MangroveGuide Web

Remote sensing techniques
for mapping and monitoring

mangroves at fine scales
Remote sensing techniques
for mapping and monitoring
mangroves at fine scales
by
Steve Schill, Valerie McNulty and Denise Perez

The Nature Conservancy
Kenichi Shono and Kim Friedman

Food and Agriculture Organization of the United Nations

Rome, 2024
Required citation:
Schill, S.R., McNulty, V.P., Perez, D., Shono, K. & Friedman, K. 2024. Remote sensing techniques for mapping
and monitoring mangroves at fine scales. Rome, FAO. https://doi.org/10.4060/cd0823en
The designations employed and the presentation of material in this information product do not imply the
expression of any opinion whatsoever on the part of the Food and Agriculture Organization of the United
Nations (FAO) concerning the legal or development status of any country, territory, city or area or of its
authorities, or concerning the delimitation of its frontiers or boundaries. The mention of specific companies
or products of manufacturers, whether or not these have been patented, does not imply that these have been
endorsed or recommended by FAO in preference to others of a similar nature that are not mentioned.
The views expressed in this information product are those of the author(s) and do not necessarily reflect the
views or policies of FAO.
ISBN 978-92-5-138796-2
© FAO, 2024
Some rights reserved. This work is made available under the Creative Commons Attribution-NonCommercial-
ShareAlike 3.0 IGO licence (CC BY-NC-SA 3.0 IGO; https://creativecommons.org/licenses/by-nc-sa/3.0/igo/
legalcode).
Under the terms of this licence, this work may be copied, redistributed and adapted for non-commercial
purposes, provided that the work is appropriately cited. In any use of this work, there should be no suggestion
that FAO endorses any specific organization, products or services. The use of the FAO logo is not permitted.
If the work is adapted, then it must be licensed under the same or equivalent Creative Commons licence. If a
translation of this work is created, it must include the following disclaimer along with the required citation:
“This translation was not created by the Food and Agriculture Organization of the United Nations (FAO). FAO
is not responsible for the content or accuracy of this translation. The original [Language] edition shall be the
authoritative edition.”
Disputes arising under the licence that cannot be settled amicably will be resolved by mediation and
arbitration as described in Article 8 of the licence except as otherwise provided herein. The applicable
mediation rules will be the mediation rules of the World Intellectual Property Organization http://www.wipo.
int/amc/en/mediation/rules and any arbitration will be conducted in accordance with the Arbitration Rules of
the United Nations Commission on International Trade Law (UNCITRAL).
Third-party materials. Users wishing to reuse material from this work that is attributed to a third party, such
as tables, figures or images, are responsible for determining whether permission is needed for that reuse
and for obtaining permission from the copyright holder. The risk of claims resulting from infringement of any
third-party-owned component in the work rests solely with the user.
Sales, rights and licensing. FAO information products are available on the FAO website (www.fao.org/
publications) and can be purchased through publications-sales@fao.org. Requests for commercial use should
be submitted via: www.fao.org/contact-us/licence-request. Queries regarding rights and licensing should be
submitted to: copyright@fao.org.
Cover photograph: Mangrove forest in Tyrell Bay, Grenada. ©TNC

CONTENTS
Acknowledgements.................................................................................................................................vii
Abbreviations..........................................................................................................................................viii
1. INTRODUCTION............................................................................................................................... 1
1.1 The purpose of the manual............................................................................................ 1
1.2 The importance of mangroves....................................................................................... 3
1.3 Global commitments and efforts to protect and restore mangroves.............................. 4
1.4 Types of mangroves....................................................................................................... 5
1.5 Approach to mangrove mapping.................................................................................... 7
2. REMOTE SENSING DATA CONSIDERATIONS............................................................................ 9

2.1 Spatial, spectral and temporal resolution.....................................................................10
2.2 Types of remotely sensed data..................................................................................... 14
2.3 Aerial photography for mapping mangrove change over time.......................................17
2.4 Optical high resolution satellite imagery (< 5 m).......................................................... 22
2.5 Synthetic Aperture Radar for defining mangrove structure.......................................... 26
2.6 Light Detection and Ranging for defining mangrove height and structure................... 27
2.7 Uncrewed aerial systems............................................................................................. 29
2.7.1 Platforms and sensors....................................................................................................29
2.7.2 Mission planning and data collection............................................................................35
2.7.3 Data processing..............................................................................................................40
3. MANGROVE MAPPING TECHNIQUES....................................................................................... 45

3.1 Visual interpretation method....................................................................................... 45
3.2 Automated classification method.................................................................................47
3.3 Remote sensing-derived indicators and methods for feature extraction...................... 49
3.3.1 Vegetation indices..........................................................................................................49
3.3.2 Change detection using Normalized Difference Vegetation Index..............................50
3.3.3 Change detection using other indices...........................................................................53
4. ONLINE RESOURCES FOR MANGROVE MAPPING................................................................ 55

4.1 Existing mangrove maps.............................................................................................. 58
4.1.1 Global Mangrove Watch..................................................................................................58
4.1.2 Ocean Data Viewer..........................................................................................................59
4.1.3 WorldCover 10-m mangrove class.................................................................................59
4.2 Open-source imagery data, software and tools for mangrove managers...................... 60
4.2.1 Imagery...........................................................................................................................60
4.2.2 Software..........................................................................................................................60
4.2.3 Tools................................................................................................................................61
5. CONCLUSIONS AND FUTURE DIRECTIONS............................................................................ 65
iii
References................................................................................................................................................67
APPENDIX............................................................................................................................................... 71
Case study 1 Parque Nacional Manglares Del Bajo Yuna, the Dominican Republic..........71
Case study 2 Caracol and Fort-Liberté, Three Bays National Park, Haiti...........................74
Case study 3 Ashton Lagoon, Union Island, Saint Vincent and the Grenadines................76
Case study 4 Tyrrel Bay, Carriacou, Grenada.......................................................................78
TABLES
1. General mangrove typologies based on their sedimentary, geomorphic
and habitat settings....................................................................................................... 5
2. Types and examples of optical and active satellite data sources with
advantages and limitations...........................................................................................16
3. Available high spatial resolution (< 5 m) satellite imagery options for
mangrove mapping and monitoring............................................................................. 22
4. Advantages and disadvantages of mapping mangroves using a UAS............................ 29
5. Advantages of multi-rotor vs. fixed-wing UAS.............................................................. 30
6. Widely used multirotor and fixed-wing UAS and technical specifications......................31
7. Popular multispectral sensors used on fixed-wing and heavy-lift UAS platforms......... 34
8. Flying heights and resulting spatial resolution and area mapped using a
DJI Phantom Pro v2 using 70 percent front and side overlap........................................ 37
9. Examples of available UAS mission planning apps....................................................... 39
10. Options for postprocessing UAS data into photogrammetric products.........................41
11. Common vegetation indices and their corresponding formulas................................... 49
12. Formulas for MSAVI and NDMA both useful for detecting changes in
mangrove forests......................................................................................................... 53
13. Directory of online resources for mangrove maps and tools......................................... 55
14. List of mangrove datasets............................................................................................ 56
FIGURES
1. A case study in Bajo Yuna National Park, Dominican Republic from TNC’s
Esri Story Map “Mapping Mangroves” showing how UAS can be used to detect
mangrove change.......................................................................................................... 2
2. Fine-scale mangrove classifications based on local geographic descriptions................. 6
3. Examples of different mangrove growth patterns on Andros Island, Bahamas............... 9
4. Near-infrared image composites of mangroves in Ashton Lagoon, Saint Vincent
and the Grenadines showing varying spatial resolutions captured by different
remote sensing platforms.............................................................................................10
5. Multispectral imagery of mangroves acquired with a UAS in Ashton Lagoon,
Saint Vincent and the Grenadines displayed using various band combinations............ 11
6. Satellite images acquired over Guadeloupe on different dates that depict
varying cloud cover percentages Image....................................................................... 11
7. Differences between three mangrove classifications using different satellite
data and methods in Andros Island, Bahamas.............................................................. 13
8. SkySat imagery over a mangrove area in west Andros Island, Bahamas....................... 14
iv
9. Synthetic Aperture Radar (SAR) data acquired over mangrove areas from
two different SAR satellites and combined into false colour composites...................... 14
10. Point cloud and vertical profile of a mangrove forest derived from LiDAR data............ 15
11. Three types of aerial photography (left to right: panchromatic; natural colour;
and colour infrared)......................................................................................................17
12. Vertical, low oblique and high oblique aerial photos over a mangrove forest
providing different perspectives...................................................................................17
13. Examples of historical aerial photos of the U.S. Virgin Islands taken in 1954 ............... 18
14. Historical aerial photos collected in January 1954 over Saint Thomas,
the United States Virgin Islands................................................................................... 19
15. Example of the selection of GCPs for georeferencing an aerial photo........................... 19
16. Appearance of mangrove forests in Jamaica in historical panchromatic and
colour infrared aerial photos....................................................................................... 20
17. Historical aerial photography and mangrove change detection in
Saint Thomas East End Reserve, the United States Virgin Islands................................. 20
18. Comparison of multiple remote sensing data showing mangrove die-off and
growth due to alternation of hydrology by a causeway near Salt River, Jamaica.......... 21
19. Examples of different high resolution satellite data displayed using different
band combinations...................................................................................................... 25
20. Different radar wavelengths penetrate and record canopy characteristics at
different depths of the forest canopy........................................................................... 26
21. Comparison of global digital elevation models (AW3D30 and CoastalDEM®)
for mangroves in Bajo Yuna National Park, Dominican Republic.................................. 27
22. Estimates of mangrove above-ground biomass and maximum canopy height
in Manglares de Estero Balsa, Dominican Republic...................................................... 28
23. Examples of a multi-rotor (left) and a fixed-wing (right) UAS........................................ 30
24. A GNSS base station collecting position correction data during a UAS mission............ 32
25. Two cost-effective UAS platforms with supplementary infrared sensors
attached for assessing mangrove extent and health.................................................... 33
26. A DJI Mavic 3 Multispectral with a 20 MP RGB camera and integrated 4-band
multispectral sensors (green, red, red-edge and NIR) for calculating vegetation
indices......................................................................................................................... 34
27. Two UAS platforms for mapping mangroves................................................................ 35
28. Examples of UAS mission plans and the area mapped at varying flight altitudes.......... 36
29. UAS imagery acquired over mangroves at different times of the day (solar noon
and early morning) in Ashton Lagoon, Saint Vincent and the Grenadines..................... 37
30. Comparison of UAS images acquired using a fast vs slow shutter speed. ..................... 38
31. User interface examples of mission planning apps include Pix4Dcapture
(top left), DroneDeploy Flight (top right), DJI GS Pro (bottom left) and
Map Pilot Pro (bottom right)........................................................................................ 39
32. UAS products of a mangrove area created from 85 overlapping stereo images
taken at different perspectives and processed using photogrammetry software......... 40
33. The rayCloud camera positions showing the location where each stereo image
was taken.................................................................................................................... 42
34. Comparison of the original image of a mangrove and a 3D point cloud generated
from stereo images using Agisoft Metashape photogrammetry software..................... 43
v
35. Example of mangrove digitization in QGIS with multiple basemaps............................. 45
36. Tracking the impacts of marina development on a mangrove forest in
Playa de Maimon, Dominican Republic using the timeline tool in Google Earth............ 46
37. Tracking the change in mangrove forests at the mouth of the Barracote River
in Bajo Yuna National Park, Dominican Republic using the timeline tool in
Google Earth................................................................................................................ 46
38. Segmenting a satellite image means creating objects around pixels that have
similar spectral properties and reflectance values.......................................................47
39. Using a DJI M300RTK multi-rotor UAS with a Micasense RedEdge MX to capture
and compute NDVI to track mangrove recovery following a hurricane......................... 50
40. TNC’s Blue Carbon Explorer showing an example of NDVI change analysis in
eastern Grand Bahama between 2015 and 2021 using Planet NICFI Dove imagery........51
41. Examples of using natural colour (RGB) UAS imagery to calculate VARI to inform
mangrove change detection in Bajo Yuna National Park, Dominican Republic............. 52
42. The GMW online platform provides remote sensing data and tools for
monitoring mangroves................................................................................................ 58
43. The ESA WorldCover online portal that provides access to global 10-m data for
land cover and forest types.......................................................................................... 59
44. The Blue Carbon Explorer’s example of mangrove restoration prioritization in
eastern Grand Bahama, 2015–2021.............................................................................. 62
45. TNC’s Mapping Ocean Wealth platform includes maps of mangrove restoration
potential and blue carbon storage............................................................................... 63
vi
ACKNOWLEDGEMENTS
This manual was developed through collaboration between the Food and Agriculture Organization of the
United Nations (FAO) and The Nature Conservancy (TNC).
The authors are grateful to our partners and collaborators for their valuable contributions to this
publication. We especially thank the following reviewers from The Nature Conservancy for their insights
and feedback: Mark Spalding; Lindsey Smart; Stefanie Simpson; Kate Longley-Wood; and Fain McGough. We
also appreciate the direction and support of Rob Brumbaugh, Shenique Albury-Smith, and Marcia Musgrave
from The Nature Conservancy. External reviewers who provided helpful editing include George Raber from
the University of Southern Mississippi, Tom Worthington from Cambridge University, and Ryan Moyer
from TerraCarbon LCC.
We also appreciate the guidance and feedback provided by FAO colleagues, including Sara Casallas
Ramirez; Frederic Castell; Simon Funge-Smith; Thomas Hofer; Pablo Martin; Maria Nuutinen; Aurelie Shapiro;
and Tiina Vähänen.
Robin Leslie provided copyediting and Kate Ferrucci designed the publication.
vii
ABBREVIATIONS
3BNP Three Bays National Park, Haiti

3D three-dimensional
AGB above-ground biomass
ALOS Advanced Land Observing Satellite
AW3D30 ALOS World 3D-30m
BGB below-ground biomass
CBD Convention on Biological Diversity
COP Conference of the Parties
CMRI Combined Mangrove Recognition Index
DBH diameter at breast height
DEM digital elevation model
dpi dots per inch
DSM digital surface model
ESA European Space Agency
ExGI Excess Green Index
EVI Enhanced Vegetation Index
FAO Food and Agriculture Organization of the United Nations
GBF Global Biodiversity Framework
GCPs ground control points
GEDI Global Ecosystem Dynamics Investigation
GEE Google Earth Engine
GEEMMM Google Earth Engine Mangrove Mapping Methodology
GIS geographic information system
GLAS Geoscience Laser Altimeter System
GMA Global Mangrove Alliance
GMW Global Mangrove Watch
GNSS Global Navigation Satellite System
GPS global positioning system
GSD ground sampling distance
IDL Interactive Data Language
ISO International Standards Organization
ICESat Ice, Cloud, and land Elevation Satellite
IUCN International Union for Conservation of Nature
JPEG Joint Photographic Experts Group
KML keyhole markup language
KNN K-Nearest Neighbours
KOMPSAT Korea Multi-Purpose Satellite
L8MI Landsat 8 Mangrove Index
LAANC Low Altitude Authorization and Notification Capability
viii
LiDAR Light Detection and Ranging
PALSAR Phased Array type L-band Synthetic Aperture Radar
MPA marine protected area
MQI Mangrove Quality Index
MVI Mangrove Vegetation Index
NASA National Aeronautics and Space Administration
NDC nationally determined contributions
NDMI Normalized Difference Mangrove Index
NDVI Normalized Difference Vegetation Index
NDWI Normalized Difference Water Index
NGO non-governmental organization
NICFI Norway's International Climate and Forests Initiative
NIR near-infrared
OTB Orfeo ToolBox
PPK Post Processed Kinematic (GPS correction technology technique)
QGIS Quantum-GIS
RADAR Radio Detection and Ranging
REDD+ Reducing Emissions from Deforestation and forest Degradation, plus the sustainable
management of forests, and the conservation and enhancement of forest carbon stocks
RGB red, green, blue natural colour composite
RMSE root mean square error
RTK Real Time Kinematic (GPS correction technology technique)
RVI Ratio Vegetation Index
SAR Synthetic Aperture Radar
SAVI Soil-Adjusted Vegetation Index
SDGs Sustainable development goals
SIDS Small Island Developing States
SPOT Satellite pour l'Observation de la Terre
SRTM Shuttle Radar Topography Mission
SWIR shortwave infrared
TGI Triangular Greenness Index
TIFF Tag Image File Format
TNC The Nature Conservancy
UAS uncrewed aerial system
UAV unmanned aerial vehicle
UNFCCC United Nations Framework Convention on Climate Change
VARI Visible Atmospherically Resistant Index
VCI Vegetation Condition Index
VIs vegetation indices
VTOL vertical take-off and landing technology
ix
©TNC
1. INTRODUCTION
1.1 THE PURPOSE OF THE MANUAL which also serve as recognized nursery environ-
Achieving food security and improved nutrition ments. Lastly, mangroves play a significant role
for the world’s population depends on preserving in climate change adaptation, providing coastal
biodiversity and ecosystem services for sustain- resilience and disaster risk reduction through
able production across the agriculture, forestry, shoreline protection, particularly in reducing the
aquaculture and fisheries sectors. However, despite impacts of storm surge.
efforts spanning several decades, biodiversity con- Given the critical importance of mangroves
tinues to be degraded. Recent global assessments to resilience and livelihoods, the need for robust
confirm that biodiversity and ecosystem services information on the status and changes in mangroves
are declining at an unprecedented rate and the to inform decision-making processes related to
pressures driving the decline are intensifying (Leal sustainable coastal development is increasingly
and Spalding, eds., 2022; Waldron et al., 2020; recognized. In response, several global tools have
Waltham et al., 2020; Friess et al., 2019). Several of been developed for mapping and monitoring
these pressures are directly or indirectly associated mangroves. However, most of these datasets
with agrifood systems, including land-use change, have been derived using 30-m spatial resolution
overexploitation, pollution and unsustainable imageries (e.g. Landsat) that make it difficult to
consumption patterns. These pressures are exac- identify mangrove patches smaller than 60 m in
erbated by climate change impacts including sea their smallest dimensions. Despite recent improve-
level rise, increased and more intense storms, and ments, for example, the free WorldCover 10 m global
altered precipitation and temperature regimes that land cover datasets produced by the European
influence mangrove vulnerability (Ward et al., 2016). Space Agency (ESA) developed using Sentinel-1
Mangrove ecosystems play a critical role in and Sentinel-2 satellite data (Zanaga et al., 2022),
harbouring biodiversity and providing a variety these global datasets typically fail to detect many
of ecosystem services. The need for developing of the narrow and fringing mangroves common
better techniques for classifying and monitoring on small islands. In addition, many local resource
mangroves is increasing, especially with the growing managers lack the resources or technical capacity
demand in blue carbon markets as mangrove forests to map and monitor at a local scale using modern
are massive carbon sinks capable of storing up to technologies, including higher resolution satellite
four to ten times as much carbon per hectare as imagery or uncrewed aerial systems (UAS), includ-
terrestrial forests (Alongi, 2020; Sanderman et al., ing drones. Many countries are challenged by the
2018). Mangrove systems provide critical habitats need to improve their mangrove cover estimates
for aquatic species of economic importance to to inform decision-making on coastal land use
coastal communities (e.g. crabs, shrimps, molluscs and to guide local conservation efforts including
and many fish) and enhance fisheries in adjacent restoration planning.
systems, such as coral reefs and seagrass beds,
1
This manual was developed to help overcome vegetation indices provides guidance for assessing
this problem by providing guidance, specifically mangrove condition and change. Options for online
targeted to build capacity for mangrove resource data resources and relevant open-source software
managers on how to use and take advantage of the offer solutions for low-budget projects.
latest technologies in mapping and monitoring This manual enables collection of data that
mangroves. Each remote sensing data source has informs conservation decision-making, guiding the
unique spatial, spectral and temporal characteristics planning of mangrove restoration and protection
that need to be appropriately matched to a research activities that will ultimately lead to enhanced
question. The plethora of remote sensing data can management and conservation of mangroves.
be overwhelming. This manual helps the reader Case studies are presented on estimating mangrove
make the best decision based on project needs extent, structure, condition and change, using a
and available budget. Before embarking on any range of remote sensing-derived sources that are
mapping project, it is important to evaluate practical focused on < 5-m spatial resolution. This includes
accuracy requirements and available skillsets. Where historical aerial photography, current high resolution
possible, low-cost options for obtaining data, as well optical and active satellite imagery, locally acquired
as open-source software for analysis can be used. UAS imagery and an inventory of existing mangrove
There are many options for assessing mangrove sources and software. We also share TNC’s new
change using historical aerial photos and archives Google Earth Engine application, the Blue Carbon
of optical and active satellite datasets. As UAS are Explorer1 that permits users to detect spatiotemporal
becoming a popular tool for local scale monitoring, changes in mangroves using derived vegetation
this manual also provides a current inventory of indices that are computed using archived satellite
available platforms and sensors as well as the data and an instructional web-based Esri StoryMap2
latest choices for mission planning applications that summarizes the results of this manual and
and postprocessing solutions. A review of pertinent includes instructional field-based videos (Figure 1).
Figure 1. A case study in Bajo Yuna National Park, Dominican Republic from TNC's Esri Story Map
“Mapping Mangroves” showing how UAS can be used to detect mangrove change
©TNC
Source: TNC & FAO. 2022. Mapping mangroves. Building data for coastal resilience and ecosystem-based climate
adaptation planning. https://storymaps.arcgis.com/stories/75615bb17fcd4704940b97fea318ad59
1
https://bluecarbon.tnc.org/
2
https://arcg.is/1WHqC5
2 REMOTE SENSING TECHNIQUES FOR MAPPING AND MONITORING MANGROVES AT FINE SCALES
©TNC
1.2 THE IMPORTANCE OF MANGROVES figure being soil carbon (Leal and Spalding, eds.,
Mangrove forests provide multiple benefits 2022). Degradation of mangroves contributes to
for people and the Earth. Mangroves serve as a climate change through loss of stored blue carbon
biodiversity hotspot and these forests are critical as CO2 is released (Spalding and Leal, 2021).
for sustaining local fisheries, with 4.1 million of Conversion of mangroves to produce agricul-
the world’s marine small-scale fisheries primarily tural commodities (aquaculture, oil palm and rice)
located in mangrove areas (Zu Ermgassen et al., accounted for 43 percent of the global mangrove
2020). losses between 2000 and 2020 (FAO, 2023). Other
Mangroves serve as natural buffers along coast- anthropogenic threats to mangroves include
lines reducing damage from storms, with global unsustainable extraction of mangrove resources,
estimates of USD 65 billion in avoided property conversion to various forms of agriculture, coastal
damage and the protection of 15 million people development and pollution. Natural retraction of
from flooding (Menéndez et al., 2020). Mangrove mangroves, including from sea level rise and coastal
forests can be up to four times more efficient erosion exacerbated by the impacts of climate
than some other forest systems at converting change, accounted for 26 percent of mangrove
carbon dioxide (CO2), a greenhouse gas, to stored losses during the same period. The climate-change-
organic carbon or blue carbon. This is due to the driven loss of mangroves further exposes vulnerable
soils being waterlogged, which slows down the communities, including in Small Island Developing
rate of decomposition (Donato et al., 2011). Global States, to disasters such as storm surges, floods and
estimates suggest mangrove forests store some tsunamis, against which healthy mangroves provide
22.86 gigatonnes of CO2, with 87 percent of this a certain level of protection (FAO, 2023).
INTRODUCTION 3
1.3 GLOBAL COMMITMENTS AND EFFORTS governments, scientists, industry, local communi-
TO PROTECT AND RESTORE MANGROVES ties and donors working towards a common goal of
Over the past 40 years, we have lost approximately conserving and restoring mangrove ecosystems, has
30 percent of the world’s mangroves (FAO, 2007; FAO, announced its own 2030 goals to help with climate
2020). Eleven of the world’s 70 mangrove species adaptation and mitigation in support of the Paris
(16 percent) are at an elevated threat of extinction Agreement (Leal and Spalding, eds., 2022). At the 27th
based on an assessment of IUCN Red List data, Conference of the Parties of the UNFCCC (COP27),
and mangrove habitats often represent primary the GMA, in collaboration with the United Nations
target areas for conversion to aquaculture and other Climate Change High-level Champions, called for
forms of agriculture, as well as coastal development signatories to a “Mangrove Breakthrough”, a unified
projects (Polidoro et al., 2010). Although the rate global approach towards mangrove conservation
of mangrove loss has declined in recent decades, (Climate Champions, 2022). This includes halting
continuing loss and degradation of mangroves loss of mangroves (an estimated 168 km2 of avoided
remain a global concern (FAO, 2023). loss), restoring half of what had been lost since
This situation has motivated governments to 1996 (some 4 092 km2 of mangroves restored) and
increase protection and restoration activities to doubling protection of an additional 61 000 km2
preserve and enhance critical ecosystem services. managed for conservation benefits (Leal and
Currently, 42 percent of mangrove areas is in Spalding, eds., 2022).
protected areas (Spalding and Leal, 2021). Many There are many other global and regional
countries have also recognized the vital role of initiatives and commitments to restoration that
mangroves in mitigating and enhancing resilience offer opportunities to further bolster mangrove
to climate change. As of October 2021, 71 countries restoration efforts. These include: the United
have included marine and coastal nature-based Nations Decade on Ecosystem Restoration
solutions in their nationally determined contri- 2021-2028;4 Target 2 under the recently adopted
butions (NDCs) under the Paris Agreement of the Kunming-Montreal Global Biodiversity Framework5
United Nations Framework Convention on Climate (GBF) to place 30 percent of degraded areas under
Change (UNFCCC) (Lecerf et al., 2021). effective restoration; the Bonn Challenge6 to bring
Under the 2030 Agenda for Sustainable 350 million hectares of degraded and deforested
Development, Sustainable Development Goal (SDG) landscapes into restoration by 2030; AFR100 (the
14 “Life below water” aims to conserve and sustain- African Forest Landscape Restoration Initiative);7
ably use the oceans, seas and marine resources for and the Initiative 20x208 in Latin America and the
sustainable development. Mangroves are closely Caribbean. Recent studies (FAO, 2023; Leal and
linked with and make strong contributions to SDG 14 Spalding, eds., 2022) highlight the dynamic nature
as they support fisheries and associated coastal com- of mangroves as coastal ecosystems transition from
munities. In addition, mangroves are also covered one to another with changes in the environmental
under Target 14.2 to sustainably manage, protect conditions. Mangroves have high potential for
and restore marine and coastal ecosystems. natural recovery where suitable habitat conditions
The Global Mangrove Alliance (GMA), a collabo-
3
can be restored. These dynamics should be consid-
ration of non-governmental organizations (NGOs), ered when planning for restoration interventions.
3
www.mangrovealliance.org
4
www.decadeonrestoration.org
5
www.cbd.int/gbf/targets/2
6
www.bonnchallenge.org
7
https://afr100.org/
8
https://initiative20x20.org/restoring-latin-americas-landscapes
©TNC
Remote sensing techniques, which are validated sediment for roots to take hold (Leal and Spalding,
with field data, play a key role in identifying the eds., 2022). Globally, there are approximately
spatial extent and condition of mangrove forests, 70 mangrove species that occur in tropical to warm
which are key information when considering bio- temperate regions, growing within deltas, estuaries,
diversity value and ecosystem service benefits, lagoons or fringing areas along shorelines. Nutrient
such as carbon sequestration. This can support availability, salinity and elevation can greatly
the identification of suitable mangrove restoration influence the species composition and structure
sites as well as the development of an appropriate of mangroves. Worthington et al. (2020) provide
technical approach to restoration. a broad-scale biophysical typology for classifying
mangroves based on their sedimentary substrate
1.4 TYPES OF MANGROVES (carbonate or terrigenous) and geomorphic setting
Mangroves are salt-tolerant evergreen forests that (deltaic, estuarine, lagoonal or open coast) (Table 1).
grow within intertidal regions along coastlines, Mangroves growing in carbonate sediments occur
where waters are calm enough and there is enough in areas where limestone deposits predominate,
Table 1. General mangrove typologies based on their sedimentary, geomorphic and habitat settings
SEDIMENTARY GEOMORPHIC DEFINITION OF GEOMORPHIC SETTINGS HABITAT
SETTINGS SETTINGS SETTINGS
Carbonate Lagoonal Shallow coastal waterbodies, intermittently Sand/shingle
separated from ocean inputs, usually formed barriers
parallel to the shore
Open coasts Sheltered environments on oceanic islands Low-energy
behind coral reefs and carbonate banks coasts
Terrigenous Deltaic Shoreline protuberances typified by a wide fan- River-dominated,
shaped alluvial plain derived from large volumes allochthonous
of river transported sediment
Estuarine Funnel-shaped main channels with bidirectional Tide-dominated,
tidal flows, characterized by large catchment allochthonous
areas and high precipitation input
Lagoonal See above Wave-dominated
barrier lagoons
Open coasts Sheltered embayments such as drowned Drowned
bedrock valleys bedrock valleys
Source: Worthington T.A., zu Ermgassen P.S.E., Friess D.A., Krauss K.W., Lovelock C.E., Thorley J., Tingey R., Woodroffe C.D., Bunting P.,
Cormier N., Lagomasino D., Lucas R., Murray N.J., Sutherland W.J. & Spalding, M. 2020. A global biophysical typology of mangroves
and its relevance for ecosystem structure and deforestation. Scientific Reports, 10:14652.
INTRODUCTION 5
Figure 2. Fine-scale mangrove classifications based on local geographic descriptions
Overwash Found on small and narrow elevated islands that may

(tide- be an extension of a larger landmass and can have
dominated) little organic matter due to the frequent movement
of the tides.
©TNC
Fringe Grow as a relatively thin fringe along the coastlines
(river- due to the topography or nutrient availability.
and tide-
dominated)
©TNC
Riverine The tallest mangroves found in sheltered floodplains
(river- along flowing waters such as tidal rivers and creeks
dominated) that are bolstered by fluvial nutrient influx and the
daily flushing tides.
©TNC
Basin Grow in partially impounded depressions that are

(interior) flooded occasionally during the dry season and
regularly during the wet season. As a result of the
stagnant or slow flowing regime, these forests are
often stunted in growth.
©TNC
Scrub Scrub (dwarf) mangroves are more common in broad

(dwarf) flat tidal areas found in arid or cooler climates with
(interior low nutrient availability.
and tide-
dominated)
©TNC
Source: TNC & FAO. 2022. Mapping mangroves. Building data for coastal resilience and ecosystem-based climate adaptation
planning. https://storymaps.arcgis.com/stories/75615bb17fcd4704940b97fea318ad59
Graphic credit: Tracey Saxby, Integration and Application Network. https://ian.umces.edu/media-library
including islands and carbonate banks constructed appropriate approach as it provides high resolution
by coral reefs, while mangroves growing on insights and the ability to monitor changes at regular
terrigenous sediments (soil derived from land) intervals over time. For monitoring larger areas, it
can be dominated by rivers, tides, waves or a would be more appropriate to collect images and
combination of river and wave influence. species information via UAS imagery and photos at
Mangroves can be further categorized into a few representative sites, then use satellite imagery
finer classifications based on local geographic to scale up that information to a larger scale (e.g.
descriptions (Figure 2). Overwash mangroves are national) map that is re-evaluated every few years.
stands that are not contiguous with any dry land – Ready-to-use map products and decision-sup-
on the highest tides they are entirely flooded. Fringe port tools may also be available from local NGOs
mangroves grow narrowly alongside shorelines and or forestry departments depending on the region
islands due to topography or nutrient availability of interest. It is important to consider the scale of
and can be sensitive to erosion from waves and datasets and the training data that were used to
tides. Riverine mangroves grow in tidal brackish generate them; global datasets are generally not
waters along the banks of creeks and rivers. They useful for planning at the national scale but can give
typically receive large amounts of freshwater and an indication of global mangrove trends and drivers
nutrients and can have some of the tallest canopies. of degradation. Regional datasets can be useful at
Basin mangroves are found in low-lying areas the national scale, depending on where the training
located in interior settings, away from the water’s data were collected. To map carbon content, more
edge where there is minimal wave activity and, in field data are required beyond drone imagery and
some cases, only rare tidal flooding. Scrub or dwarf species information. Tree counts, height, diameter
mangroves can be found in many different mangrove at breast height (DBH), and canopy density need to
environments and have very stunted growth and be collected to estimate above-ground biomass
sparse biomass due to low nutrient availability, (AGB), while soil cores need to be collected and
high salinity or colder climates. processed to estimate below-ground biomass (BGB).
With enough field data, allometric equations can
1.5 APPROACH TO MANGROVE MAPPING be developed for estimating carbon content more
To protect mangroves and the extensive ecosystem broadly using remote sensing.
services they provide, it is critical to understand
their extent, scale and condition. Global mangrove
monitoring tools, such as Global Mangrove Watch
(GMW), can be used to determine the general
distribution or trends of mangroves at a broad scale,
however finer-scale mapping and monitoring is often
needed to inform national-scale and local-scale
conservation and management decision-making.
Three critical aspects to consider in designing a
mangrove mapping approach are 1) the research
question, 2) the scale of the area of interest and
3) available capacity for field data collection and
image processing.
For monitoring mangrove health at small sites,
such as a restoration area or a local marine pro-
tected area, UAS monitoring is likely to be the most
INTRODUCTION 7
©TNC
2. REMOTE SENSING
DATA CONSIDERATIONS
Given the various growth patterns and correspond- the analyst because both manual digitization and
ing appearance, mangrove forests can be difficult to automated classifications involve some level of
map at fine scales using remote sensing techniques. visual interpretation. For these reasons, mapping
Although all mangrove forests consist of the primary efforts can be difficult to transfer across teams
components of leaves, stems and branches, the or applications (Kuenzer et al., 2011). In addition,
reflectance patterns can vary based on age, health, species and growth patterns vary greatly across
species, season, soil type, water quality, leaf density different parts of the world, between neighbouring
and leaf geometry. The mixing of these components islands, and even on a single island, so a one-size-
and corresponding reflectance is greater in remote fits-all interpretation process and analysis technique
sensing datasets with coarser spatial resolution, can be problematic (Figure 3). Therefore, field data
which adds another layer of complexity. Mangrove are essential for interpreting these patterns and
mapping is also inherently biased being based on improving classification accuracy.
Figure 3. Examples of different mangrove growth patterns on Andros Island, Bahamas

©TNC
©TNC
©TNC
©TNC
Notes: These mangroves grow on different substrates and in different spatial patterns, which makes them difficult to map
at a fine scale. Top left: Scrub/dwarf mangroves growing on elevated black algal mats with low nutrient availability. Top
right: Expansive scrub/dwarf mangroves across a tidal flat with taller and denser fringing mangroves growing along tidal
creeks where nutrient availability is higher. Bottom left: A coppice ridge elevated area where a broadleaf and palm forest
grow separated from the low-lying mangrove forest. Bottom right: Very sparse scrub/dwarf mangroves growing near a blue
hole on a sandy substrate.
9
2.1 SPATIAL, SPECTRAL AND TEMPORAL image data source to use for mapping mangroves, it
RESOLUTION is important to evaluate three fundamental remote
Although some countries have developed their own sensing resolutions:
national-scale datasets, many developing countries 1. Spatial resolution is the actual area covered
do not have the resources and expertise to conduct on the Earth’s surface within a pixel dimension
their own mapping and have relied on global remote (Figure 4). This influences the ability to recognize
sensing-derived datasets as their principal source and map features within an image. This is also
for documenting mangrove extent. Such global called ground sampling distance (GSD). The
datasets are often based on medium resolution highest spatial resolution that can be obtained
satellite data (i.e. 30-m pixel) and miss much of the from a satellite image is a 0.3-m pixel while UAS
narrow, fringing and scrub/dwarf mangroves that sensors can record detail down to centimetre
often exist across small islands. When deciding on an (cm) per pixel.
Figure 4. Near-infrared image composites of mangroves in Ashton Lagoon, Saint Vincent and the
Grenadines showing varying spatial resolutions captured by different remote sensing platforms
0.05 m 0.5 m
©TNC
©TNC
1.85 m 4.0 m
©TNC
©TNC
10.0 m 30.0 m
©TNC
©TNC
Notes: The spatial resolution of a remote sensing dataset indicates the resolving ability per pixel. The drawback of having
better clarity in an image is that the footprint of the scene covers a smaller geographic area. These false colour composites
show vegetation as red and are useful for detecting stress levels in plants as healthier vegetation reflects higher levels of
near-infrared energy. Examples shown here are: Altum multispectral data acquired from a UAS (0.05 m) and SkySat (0.5 m)
(top row left to right); WorldView-2 (1.8 m) and PlanetScope SuperDove (4 m) (middle row); and Sentinel-2 (10 m) and
Landsat 9 (30 m) (bottom row).
2. Spectral resolution refers to the way light is light provides critical information related to
recorded on the sensor, both in the number the health of a plant. The more spectral bands
of bands but also the narrowness of how light that are recorded, the greater the ability
is being recorded (Figure 5). It can be a simple increases to separate features based on their
three-band (RGB = red, green, blue) natural reflectance patterns.
colour image from a UAS, an eight-band satellite 3) Temporal resolution refers to how often the
image from the WorldView-3 satellite or a hyper- sensor records imagery of a particular area,
spectral image acquired from an airplane with or the data collection revisiting frequency
hundreds of bands. Multispectral sensors record (Figure 6). Some satellites can take days to
wavelengths beyond the visible spectrum and revisit an area while a UAS can be deployed on
into the infrared wavelengths that do a much demand provided the environmental conditions
better job of detecting vegetation health. The are appropriate.
absorption of red light and reflection of infrared
Figure 5. Multispectral imagery of mangroves acquired with a UAS in Ashton Lagoon, Saint Vincent
and the Grenadines displayed using various band combinations
©TNC
©TNC
©TNC
Notes: Variations in spectral resolution can be seen in this multispectral imagery collected with a Micasense Altum
multispectral sensor mounted on a WingtraOne fixed-wing transitional UAS over a mangrove forest located at Ashton
Lagoon in Union Island, Saint Vincent and the Grenadines. This payload captures five bands of data (red, green, blue, red
edge and NIR spectrums) at spatial resolutions down to 3.4 cm/pixel and thermal down to 54 cm/pixel. The left image is
in natural colour (RGB), the middle and right images are false NIR colour composites (RGB = red-edge, red and green and
RGB = NIR, red-edge and red). The displayed colours will change based on what bands are assigned to the different colour
channels (e.g. red, green, blue). These infrared bands permit a much broader evaluation of plant health and vegetation
stress detection.
Figure 6. Satellite images acquired over Guadeloupe on different dates that depict varying cloud
cover percentages
©2022 PLANET LABS PBC
Notes: Temporal resolution addresses the ability to revisit a site to collect data when needed. One of the advantages of the
PlanetScope satellite constellation is that it has a very high temporal resolution that makes available a selection of daily
scenes to choose from. This can be particularly useful in the tropics where a cloud-free day is rare. These PlanetScope
SuperDove scenes were acquired over the same mangrove area in Guadeloupe but collected over three consecutive days
with different cloud cover results.
REMOTE SENSING DATA CONSIDER ATIONS 11

Each remote sensing data source has unique resolution is not always the best option because it
spatial, spectral and temporal characteristics that results in significantly larger data volume to process
need to be appropriately matched to the research and there may be an overload of information to deal
question, while considering the available budget with (e.g. shadows, more features that need to be
and size of the project area. Matching datasets classified). In addition, higher spatial resolution
appropriately can better support decision-making datasets from satellites will have narrower imaging
around improved management, conservation and swaths, so this may not be practical for mapping
restoration of mangroves. large areas.
Regarding spatial resolution, a general rule of In terms of spectral resolution, more spectral
thumb is that the spatial resolution of the imagery bands will generally provide a greater ability to
should be less than one-half the size of the feature discriminate between features. For example, it
of interest measured in its smallest dimension. For would be difficult to distinguish different mangrove
example, identifying a fringing mangrove of 2-m species using only three visible bands (RGB),
width will require an imagery with a spatial resolu- however, a WorldView-3 satellite scene with eight
tion of 1-m pixel or less. National forest definitions bands that include a red edge and two near-infrared
also need to be considered when determining the (NIR) bands, would improve the likelihood of species
appropriate spatial resolution. The highest spatial separation when combined with field data.
©TNC
Figure 7. Differences between three mangrove classifications using different satellite data and meth-
ods in Andros Island, Bahamas
©TNC
©TNC
©TNC
©TNC
Notes: Three classifications of mangroves using different satellites and methods that produced different results. Top left:
A high resolution (4-m pixel) infrared composite of a PlanetScope Dove scene over a mangrove area in western Andros
Island, the Bahamas. Through field verification, the identified coppice ridges indicated on the image are elevated land
features that consist of broadleaf forests and palms, not mangroves. Top right: The mangrove extent as mapped by the
Global Mangrove Watch 2021 dataset (25-m pixel). Bottom left: The mangrove extent as mapped by the ESA WorldCover
2021 dataset (10-m pixel). Bottom right: The classification by TNC using PlanetScope Dove imagery (4-m pixel) which
excludes the coppice ridges and includes much of the scrub/dwarf mangrove areas that are absent in the other two
datasets. When coupled with field data and higher resolution satellite imagery, more accurate boundaries can be mapped
for mangrove extent.
The temporal resolution of a sensor is important is often not practical because of the limited area that
when mapping mangroves in areas with persistent can be mapped per battery, difficulty in achieving
cloud cover. A significant advantage of using the consistency in environmental conditions, and the
Planet Dove satellite constellation is the near daily volume of data produced. In these instances, air-
cadence of data collection that provides more plane or satellite imaging is often more appropriate.
options for selecting the best possible (cloud-free) Figure 7 shows examples of three different
scene. UAS can have a very high temporal resolution mangrove classifications based on different satellite
because they can be deployed whenever suitable datasets that produced varying levels of accuracy.
environmental conditions prevail. When combined with field data, higher spatial
Finally, the image swath or footprint (i.e. image resolution satellite imagery, such as Planetscope
area of a scene) varies between data sources so the Dove (4-m pixel), can help separate out elevated
size of the project area will dictate the most appro- areas, such as coppice broadleaf ridges that are not
priate data source to use. Mapping using multiple mangroves, and identify sparse scrub mangroves.
image swaths that were acquired at various times, Compared to global products, such as the Global
seasons and environmental conditions can be prob- Mangrove Watch 2021 dataset (25-m pixel) and
lematic for achieving consistency in mapping across the ESA WorldCover 2021 dataset (10-m pixel),
seamlines. National-scale mangrove inventories are finer scale imagery can achieve higher accuracy
typically conducted using satellite imagery where in mapping mangrove extent. Field validation is
each scene covers a larger geographic extent. For important as higher spatial resolution datasets
mapping larger areas (e.g. > 10 km ), the use of UAS
2
often overestimate mangrove extent.

2.2 TYPES OF REMOTELY SENSED DATA electromagnetic energy. Optical satellite imagery
Remote sensing data can be classified into optical that has a spatial resolution of less than 5 m provides
and active systems. adequate level of detail for mapping at the local
Most Earth observation satellites are optical scale and some systems can be tasked to acquire
remote sensing systems that record reflected data over a user-defined area (Figure 8).
Figure 8. SkySat imagery over a mangrove area in west Andros Island, Bahamas
Notes: SkySat optical four-band satellite imagery (1-m MS/0.5-m PAN) shows a natural colour composite (RGB) on the
left and a false colour infrared composite on the right. Because healthy plants reflect much more infrared energy, the
detection of vegetation is much clearer in the infrared composite, and denser plant growth (i.e. more biomass) appears in
darker shades of red.
Active systems, such as Radio Detection and dependent on the Sun’s energy. Active sensing
Ranging (RADAR) and Light Detection and Ranging systems can be used to detect mangrove forest
(LiDAR) emit and record the backscatter of their structure, map individual trees and stand diameters,
own energy (e.g. microwave, infrared) and are not and predict volume and biomass (Figure 9).
Figure 9. Synthetic Aperture Radar (SAR) data acquired over mangrove areas from two different SAR
satellites and combined into false colour composites
Left: ALOS PALSAR-1 (L-band, 23-cm) data. Right: Sentinel-1 (C-band, 5.5-cm) data. Dark colours represent smooth surfaces
where radar signals are reflected away from the sensor, and brighter surfaces indicate roughness where signals bounce
back and are recorded by the sensor. The different wavelengths and polarizations provide insight into mangrove height
and canopy structure with stronger backscatter in taller mature canopies and weaker backscattering recorded in younger,
low-density forests.
Adapted from: Simard, M. 2019. Radar remote sensing of mangrove forests. In: A. Flores, K. Herndon, R. Thapa &
E. Cherrington, eds. NASA SAR Handbook: Comprehensive methodologies for forest monitoring and biomass estimation.
NASA. DOI: 10.25966/33zm-x271
One of the main advantages of RADAR is that LiDAR data can be used to generate detailed
it can penetrate cloud cover and be acquired at three-dimensional point clouds that can be used
different wavelengths, with each band capable of to visualize, quantify and classify vegetation char-
characterizing different forest stand parameters. acteristics (Figure 10).
Figure 10. Point cloud and vertical profile of a mangrove forest derived from LiDAR data

Notes: The different colours represent the height of the laser return and are classified into different canopy height levels,
which provide information on the canopy structure.
Source: Li, Q., Wong, F.K.K. & Fung, T. 2019. Classification of mangrove species using combined WordView-3 and LiDAR data
in Mai Po nature reserve, Hong Kong. Remote Sensing, 11(18): 2114.
©TNC

Table 2 provides a summary of the widely different spatial, spectral and temporal charac-
adopted optical and active remote sensing data teristics that should be considered, depending on
sources for mapping and monitoring mangroves the mapping objectives. Higher resolution datasets
along with their associated advantages and limi- can be more expensive (not in the public domain),
tations. There are various optical satellite image cover a smaller area and scene availability may be
providers to choose from and each satellite has lower due to the smaller scene footprint.
Table 2. Types and examples of optical and active satellite data sources with advantages and limitations
TYPES SENSOR / EXAMPLES APPLICATIONS ADVANTAGES LIMITATIONS
RESOLUTION
Optical Aerial: Aerial Local-scale Ultra-detail, can be Impractical to use
high spatial photography, mapping, acquired below cloud for larger areas,
resolution UAS imagery species cover, often historical data volume
(< 0.5 m) identification, imagery available
3D modelling allowing for long-term
monitoring
Satellite: WorldView-3, Broader-scale Can be tasked to Smaller footprint
high spatial PlanetScope, mapping image a particular than medium
resolution Pléiades, applications, area (higher resolution, cost,
(0.5 m to 5 m) SkySat, canopy temporal resolution), scene availability,
KOMPSAT-5, modelling, multispectral, broader data collection
SPOT-7 species area than UAS after year 2000
level with
multispectral
data, DSM
generation
Satellite: Sentinel-2 National or Public domain Landsat Coarser spatial
medium (10 m), regional scale, is available for longer resolution that
spatial Landsat (30 monitoring time periods than misses narrow
resolution m) of large-scale newer, high-resolution and fringing
(5 m to 30 m) changes and satellite data, which mangroves
condition is important for
assessments longer term change
detection
Hyper- Airborne/ Species Powerful classification Limited
spectral UAS discrimination potential, user- availability,
and plant defined spatial and expensive, large
health spectral resolution data volume,
assessments based on the research requires extensive
question analytical skills to
process
Active Synthetic Sentinel-1, Invaluable Sentinel-1 C-band Requires
Aperture ICEYE SAR, asset for can image down to advanced
Radar (SAR) TerraSAR-X/ canopy 5 m. High frequency processing
PAZ, Capella modelling X-band provides capabilities,
and moisture better user-defined higher costs,
detection, detail, weather and requires optical
useful when daylight independent, data integration
combined flexible area coverage,
with optical complements optical
data, DSM data
generation
LiDAR Often 3D canopy Very high accuracy, Requires
airborne, modelling, speed, versatile, advanced
GEDI, ICESat point cloud automated processing
(space-based) and DSM functionality, cost- capabilities,
generation effective over large higher costs, large
areas data volume
Adapted from: Kathiresan, K. & Bingham, B.L. 2001. Biology of mangroves and mangrove ecosystems. Advances in Marine Biology,
40: 81–251; Kuenzer, C., Bluemel, A., Gebhardt, S., Quoc, T.V. & Dech, S. 2011. Remote sensing of mangrove ecosystems: A review.
Remote Sensing, 3(5): 878–928.
2.3 AERIAL PHOTOGRAPHY FOR MAPPING valuable for determining the spatial extent of
MANGROVE CHANGE OVER TIME historical mangrove cover. Early aerial photos were
While medium resolution (e.g. 30-m pixel) satellite typically acquired using two types of black and
image products have been available since the 1980s, white film: panchromatic and infrared. Starting in
high resolution satellite technology (< 5-m pixel) the 1950s, colour film became more widely used
has only existed since the early 2000s. In contrast, in both natural colour (RGB) and colour infrared
aerial photography provides an invaluable, high- formats (Figure 11). As vegetation reflects much
resolution record of land cover conditions that dates more infrared energy, infrared film offers better
back as early as the 1930s. Aerial photos are used contrast between forest types and greater ability
for a variety of applications including the creation to distinguish between healthy and diseased
of topographic maps, generation of contour lines, vegetation, delineate bodies of water and penetrate
land-use change analysis and photogrammetric atmospheric haze (Avery, 1969).
measurements. Aerial photos can be extremely
Figure 11. Three types of aerial photography (left to right: panchromatic; natural colour; and
colour infrared)
©TNC
©TNC
©TNC
Aerial photos acquired for mapping purposes can be processed into orthophoto mosaics which
are typically captured vertically, with the camera composite all photos into a single file, removing all
pointed straight down (Figure 12). The rate of distortion so they are planimetrically correct and
photo capture is sequenced so that each photo measurements can be taken. This is done using
will have at least 60 percent overlap between photogrammetric methods, capturing key points
each sequential photo and 30 percent photo among overlapping photos which generates a relief
overlap between flightlines. Because each photo model. This in turn is then used to transform the
is captured at a slightly different perspective, aerial photos into a perspective, making them
paired stereo photos can be used to visualize the appear to have been taken vertically above at an
landscape in three dimensions. Stereo photos infinite distance.
Figure 12. Vertical, low oblique and high oblique aerial photos over a mangrove forest providing
different perspectives
©TNC
©TNC
©TNC

Another important aspect of vertical aerial Historical aerial photos that have been
photos is photo scale which is the ratio of the digitized, and even georeferenced, can often be
distance between two points on a photo to the found in public web-based mapping libraries and
actual distance between the same two points on the archives. Another option is to query government
ground. One way to calculate photo scale is to divide agencies as historical photos are often archived
the focal length of the camera by the flying height and stored by planning, survey or forestry
above the ground. If a standard lens focal length departments. Originally delivered as hard copy
of 15 cm is assumed, photo scales will range from 9 x 9-inch (23 x 23-cm) photos or transparencies,
about 1:2 000 (very large scale) to approximately these can be scanned and georeferenced for use
1:135 000 (very small scale). Most aerial photos are in geographic information system (GIS) software.
acquired in scales ranging from 1:12 000 to 1:20 000. Photos or transparencies can be scanned using
The higher the flying altitude, the smaller the scale a large-format scanner and a user-defined dot-
(less detail). Other helpful information often found per-inch (dpi) analogue to the digital conversion
on the frame of an aerial photo includes the date rate (Jensen, 1996). For example, a 1:20 000 scale
and time, roll and frame number, and altimeter. photograph scanned at 500 dpi would result in a
Aerial photos can also be collected at oblique digital file with each pixel representing 1 x 1 m on
angles with the camera tilted at various angles the ground. Scanning rates greater than 2 000 dpi
between the horizon and the ground. A low oblique are not recommended because this exceeds the
photo is where the horizon is not visible and a high resolution of the emulsion properties of the film;
oblique photo is taken at a higher angle, where the moreover, the file size becomes exponentially larger
horizon can be seen (Figure 12). Oblique photos with higher scan rates (Ruzgienė, Bagdžiūnaitė and
cover a larger area and are often easier to interpret. Ruginytė, 2012). Transparencies can be scanned
However, it is not possible to take accurate meas- using a scanner that has a special hood that allows
urements from them as they have significant scale light to pass through the transparency and record
distortions due to the varying distances. the value on the top part of the scanner.
Figure 13. Examples of historical aerial photos of the United States Virgin Islands taken in 1954
©U.S. GEOLOGICAL SURVEY
Left: A panchromatic photo taken on 29 January 1954 of the eastern end of Saint Thomas, the United States Virgin Islands
at a scale of 1:28 400. Right: A colour infrared photo taken in April 1971 of Portland Bight, Jamaica. Mangroves are much
more discernible in the colour image. These photos can be scanned and georeferenced, then used in a GIS to digitize
historical mangrove extent for change detection purposes.
Once the aerial photos are scanned and in a Once sufficient GCPs (a minimum of four) have
digital format (TIFF or JPEG), they can be georefer- been identified across the image as uniformly as
enced to a coordinate system by selecting ground possible, a resampling (i.e. warping) algorithm is
control points (GCPs) that match an existing geo- applied to reproject the image into the defined
referenced base map, such as a high-resolution coordinate system. The reported root mean square
satellite image, using open-source GIS software error (RMSE) will indicate the resulting geometric
(e.g. QGIS). It is important to select GCPs that accuracy of the georeferenced image. Once the
represent permanent positions such as roads and image is georeferenced, GIS software and image
other artificial features (not soft coastlines which interpretation can be used to digitize polygon
are dynamic and change over time). There are many boundaries around the interpreted mangrove extent
online tutorials that describe how to do this.
9
(Figures 13 to 18).
Figure 14. Historical aerial photos collected in January 1954 over Saint Thomas, the United States
Virgin Islands

Notes: These photos were acquired in an overlapping (stereo-pair) format, with at least 60 percent overlap between each
photo capture and 30 percent sidelap between flightlines. The collection of stereo photos permits stereo viewing and
the creation of an orthophoto mosaic in which all geometric distortions are removed (GIS ready). This is identical to the
way a UAS mission is planned, however the overlap is generally higher (e.g. 70 percent) because the UAS is flying in closer
proximity to the ground.
Figure 15. Example of the selection of GCPs for georeferencing an aerial photo ©TNC/AERIAL PHOTOGRAPHS COURTESY
OF U.S. GEOLOGICAL SURVEY
Notes: These photos were acquired in an overlapping (stereo-pair) format, with at least 60 percent overlap between each
photo capture and 30 percent sidelap between flightlines. The collection of stereo photos permits stereo viewing and
the creation of an orthophoto mosaic in which all geometric distortions are removed (GIS ready). This is identical to the
way a UAS mission is planned, however the overlap is generally higher (e.g. 70 percent) because the UAS is flying in closer
proximity to the ground.
9
https://docs.qgis.org/3.22/en/docs/user_manual/working_with_raster/georeferencer.html

Figure 16. Appearance of mangrove forests in Jamaica in historical panchromatic and colour infrared
aerial photos

Note: It is much easier to identify mangroves using colour clues, but canopy texture is also helpful when interpreting
imagery.
Figure 17. Historical aerial photography and mangrove change detection in Saint Thomas East End
Reserve, the United States Virgin Islands
©TNC/U.S. GEOLOGICAL SURVEY

0 0.25 0.5 1 Kilometers
No change in mangrove
Mangrove extent in 1954
Mangrove extent in 2022
Note: Using historical aerial photography in Figure 17, mangrove change was detected over a 68-year period. The
construction of a marina resulted in the removal of several hectares of mangrove, however, mangroves have expanded into
other areas.
Figure 18. Comparison of multiple remote sensing data showing mangrove die-off and growth due to
alternation of hydrology by a causeway near Salt River, Jamaica
© 2022 PLANET LABS PBC


Notes: The top images are panchromatic and colour infrared aerial photos acquired in 1971. The bottom images are
from SkySat acquired on 26 May 2022 and are shown in both natural colour (RGB) and colour infrared composites. Aerial
photographs courtesy of U.S. Geological Survey/Image
©TNC

2.4 OPTICAL HIGH RESOLUTION SATELLITE at less than 5-m spatial resolution (Table 3). The
IMAGERY (< 5 M) options and price structures are constantly chang-
There are many choices of high-resolution satellite ing as satellite technology continually improves
imagery for mapping and monitoring mangroves and new satellite constellations are launched.
Table 3. Available high spatial resolution (< 5 m) satellite imagery options for mangrove mapping
and monitoring
SATELLITE SPECTRAL SPATIAL SCENE MIN. ORDER PRICE/KM2 ACQUISITION
BANDS RESOLUTION (m) SWATH (km) AREA (km2) * (USD) DATES
AIRBUS
Pléiades Pan 0.3 14 25 20–25 May 2021 to
Neo present
6-band MS 1.2
Pléiades Pan 0.5 20 25 10–15 Dec 2011 to
1A/1B present
4-band MS 2
SPOT-6/7 Pan 1.5 60 100 <5 2012/2014 to
present
4-band MS 6
MAXAR
World- Pan 0.31 13 25 20–25 Nov 2016 to
View-4 Jan2019
4-band MS 1.24
World- Pan 0.31 13 25 20–25 Aug 2014 to
View-3 present
8-band MS 1.24
8-band SWIR 3.71
World- Pan 0.46 16.4 25 15–20 Oct 2009 to
View-2 present
8-band MS 1.85
World- Pan 0.5 17.7 25 10–15 Sep 2007 to
View-1 present
GeoEye-1 Pan 0.5 15 25 15–20 Sep 2008 to
present
4-band MS 1.84
Quickbird Pan 0.6 17 25 15–20 Oct 2001 to
Dec 2014
4-band MS 2.4
IKONOS Pan 1 11.3 25 10–15 Sep 1999 to
Dec 2014
4-band MS 4
PLANET
PlanetScope 4-band MS ~4 24 25 On request Jul 2014 to
Dove/ 8-band MS 32.5 Apr 2022
SuperDove Mar 2020 to
present
SkySat Pan 0.5 11 On request 5–10 Nov 2013 to
present
4-band MS 1
ADDITIONAL OPTIONS
BlackSky Pan 1 26–59 On request On request Nov 2018 to
present
3-band MS
EROS B Pan 0.7 7 25 5–10 Apr 2006 to
present
GaoFen-2 Pan 0.8 45 25 <5 Aug 2014 to
(GF-2) present
4-band MS 3.2
SATELLITE SPECTRAL SPATIAL SCENE MIN. ORDER PRICE/KM2 ACQUISITION
BANDS RESOLUTION (m) SWATH (km) AREA (km2) * (USD) DATES
ADDITIONAL OPTIONS (continued)
GEOSAT-2 Pansharpened 0.75 12 25 5–10 Jun 2014 to
(Formerly present
Pan 1
Deimos-2)
4-band MS 4
Jilin-1 Pan 0.5 11.6 25 5–10 Oct 2015 to
present
4/5-band MS 2
Colour 1
night-time
Colour video 1
KazEOSat-1 Pan 1 20 100 <5 Apr 2014 to
present
4-band MS 4
KOMP- Pan 0.5 12 25 10–15 Aug 2014 to
SAT-3A present
4-band MS 2
KOMPSAT-3 Pan 0.5 16 25 5–10 Aug 2014 to
present
4-band MS 2
OVS-3 Natural 0.9 22.5 131 5–10 Aug 2014 to
colour present
(OVS-1, 1.9;
OVS-2, 0.9)
SuperView-1 Pan 0.5 12 25 10–15 Aug 2014 to
present
4-band MS 2
TripleSat Pan 0.8 23.8 25 5–10 Aug 2014 to
present
4-band MS 3.2
TeLEOS-1 Pan 1 12 100 <5 Sep 2008 to
present
* Estimated prices as of April 2023. Discounts may be available for NGOs and non-commercial applications.
Source: Authors’ elaborations.
This section provides a general (non-exhaustive) (0.3-m pan/1.2-m colour), a constellation of four
overview of available remote sensing data options identical satellites and along with WorldView-3, it
and technical specifications that may fit certain has the highest spatial resolution available from
budgets and project requirements. The five main space. Costing about the same as WorldView-3,
considerations in selecting satellite imagery are the Neos have a slightly larger swath, but fewer
1) spatial resolution in both panchromatic and spectral bands (6 vs 8 bands). However, the 8-band
multispectral images; 2) the number of available WorldView-3 data archive goes back to 2014. In addi-
bands; 3) the swath dimension for each scene (foot- tion to Pléiades, Airbus operates and distributes the
print); 4) the dates of available imagery; and 5) cost. TanSAR-X/TanDEM-X high resolution radar products,
Airbus operates the Pléiades satellite con-
10
the KazEOSat-1 (1-m pan/4 m-colour) and SPOT-6/7
stellation that has been collecting imagery since (1.5-m pan/6-m colour) which have been operating
December 2011 and offers more affordable 0.5-m since 2014 and have the largest imaging swath of
panchromatic (pan) and 1-m multispectral (colour) 60 km. Airbus imagery can be searched, browsed
data options with one of the largest swaths (20 km). and purchased through the GeoStore11 and custom
The company recently launched the Pléiades Neo tasking of a satellite can be scheduled.
10
www.intelligence-airbusds.com/imagery/constellation
11
www.intelligence-airbusds.com/en/4871-ordering

MAXAR12 (formerly DigitalGlobe) is a pioneer educators can also apply to access free imagery.
in high resolution satellite imaging, going back Planet also operates the SkySat satellites (0.5-m
to the first 1-m satellite, IKONOS (1-m pan/4-m pan/1-m colour) which collect 4 bands at an
colour), that was launched in 1999 and continued 11-km swath and can be tasked. In 2023, Planet
operating until March 2015. This was followed by the will upgrade their 21 SkySat satellites and start
launching of Quickbird (0.6 m-pan/2.4 m-colour) in launching their 32 Pelican satellites that will offer
2001, the first submeter satellite that operated until 0.3-m imagery with up to 30 revisits per day. Planet15
December 2014. GeoEye-1 (0.5-m pan/1.8-m colour) is a searchable database of archived scenes and
was launched in 2008, followed by WorldView-2 global basemaps mosaics that are available monthly
(0.5 m pan/1.8 m colour) in 2009 with a much larger or quarterly. Other plans include the development
swath (25 km) and the first 8-band multispectral of a hyperspectral fleet of satellites called Tanager
imaging system (coastal blue, blue, green, yellow, which will be of great benefit for mangrove mapping
red, red-edge, near-IR, near-IR2). WorldView-3 was and change detection.
launched in 2014 and was the first satellite to image BLACKSKY16 operates a constellation of satellites
below 0.5-m spatial resolution (0.3-m) with eight that provide 1-m pan scenes that are up to 60-km
1.2-m multispectral bands. WorldView-4 was similar wide as well as radar images down to 0.25-m resolu-
to WorldView-3, but only collected data between tion with a structured pricing system. Other options
November 2016 and January 2019 due to mechanical include the Chinese-operated 4-band satellites, the
failure. Maxar also distributes C-band Radarsat-2 Jilin-1 (0.5 m-pan/2-m colour) launched in 2015 and
imagery that has been collected since 2008. The the SuperView-1/2 (0.5-m pan/2-m colour) that was
new WorldView Legion six satellite constellation launched in 2016. Two Jilin-1 hyperspectral satel-
is scheduled to launch in 2023 and will provide lites were launched in January 2019 and provide
0.3-m resolution imagery with up to 15 revisits 28 bands at 5-m spatial resolution. Three Jilin-1
per day. Imagery for Maxar’s constellation can be Smart Video satellites were launched in November
searched, browsed and purchased through the 2017 and provide 4K video for real-time monitoring
Discover portal and custom tasking of a satellite
13
at a 1 m spatial resolution. A new SuperView Neo 28
can be scheduled. satellite constellation was launched in 2022 and will
Planet14 is a company that started in 2010 and provide 0.3 m-pan and 1.2-m 4-band multispectral
now manages the largest constellation of satellites imagery. Another Chinese operated satellite is
with over 200 currently in orbit. They operate the the Gaofen-2 (0.8-m pan/3.2-m colour) that was
PlanetScope Dove and SuperDove nanosatellites launched in August 2014 and has a 45-km swath.
(5 kg each) that collect daily 4-band scenes and The Korea Aerospace Research Institute17
8-band scenes at 4-m spatial resolution. These are (KARI) launched the four-band Kompsat-3A (0.5-m
most likely to be the most affordable high resolu- pan/2-m colour) in March 2015 which is a sister
tion image products unless one qualifies for the spacecraft to the previously launched Kompsat-3
Norway’s International Climate and Forest Initiative (0.7-m pan/2.8-m colour). The OVS-3 satellite
(NICFI) satellite data programme that provides free offers 0.9-m RGB imagery and 120-second video.
imagery to partners working to curb deforestation The TripleSat/DMC3 (0.8 m-pan/3.2-m colour) is an
in tropical areas of the world. Researchers and Indian constellation launched on 10 July 2015, with a
12
www.maxar.com
13
https://discover.maxar.com/
14
http://planet.com/
15
www.planet.com/products/explorer
16
www.blacksky.com
17
www.kari.re.kr/eng.do
23-km swath. More affordable options are the EROS that will be used to search for suitable imagery
B (0.7-m pan) launched in 2006, the TeLEOS-1 with acquired in the past, or to make a tasking request
1 m pan imagery launched in 2016 and the GEOSAT-2 to acquire new imagery. Data collection parameters
(formerly Deimos-2) (1-m pan/4-m colour) launched such as tide windows, maximum shooting angle
in 2014 that has a 12-km swath. and acceptable cloud cover levels can be speci-
An exciting new future satellite is the Albedo fied. Typically, there is a minimum area required
constellation, which will deliver 10 cm pan and for placing an order (larger areas when making
40 cm colour and is anticipated to be launched in new tasking requests). Older archived scenes (e.g.
2024. Satellogic is an Argentine-owned company > 6 months old) are often less expensive with smaller
that has launched 17 NewSat satellites since 2013, minimum order areas. Companies typically have
with plans to develop over 300+ microsatellites different product options in terms of radiometric
by 2025. These microsatellites provide submeter and geometric accuracy. For classifying the imagery,
resolution multispectral imagery and 29 bands, it is important to order the calibrated percent
25-m hyperspectral imagery and a 1-m resolution reflectance product, or radiometric corrections
full-motion video capability with daily updates. will need to be run. Likewise, higher levels of
Before ordering satellite imagery, the area-of- geometric accuracy will cost more, so it is important
interest will need to be defined as a polygon file to understand the project accuracy requirements
Figure 19. Examples of different high resolution satellite data displayed using different band
combinations
©2022 MAXAR, ©2022 PLANET LABS PBC

Top left: A WorldView-2 infrared colour composite (2-m pixel) of the mangroves of Caracol Bay, Haiti acquired on 3
September 2011. Top right: SkySat infrared colour composite (1-m pixel) of mangroves recovering from Hurricane Dorian in
eastern Grand Bahama (16 November 2022). Bottom left: A GeoEye-1 pan-sharpened natural colour composite (1-m pixel)
of mangrove reforestation taking place in former aquaculture areas in Bajo Yuna National Park, the Dominican Republic
(10 September 2022). Bottom right: A Planet SuperDove infrared colour composite of the Black River Morass mangroves in
Jamaica (3 November 2022). The different band combinations provide insight for various ecological investigations.

and the available budget. One option that is often such as canopy height, estimate AGB, and identify
offered by satellite companies is to pan-sharpen canopy crown (Pham et al., 2019a).
the image. This is a radiometric transformation SAR can be acquired at different wavelengths,
that fuses the high spatial resolution pan band with with each band characterizing different forest stand
the lower spatial resolution multispectral bands to parameters. Shorter wavelengths (e.g. X-band, 3-cm
produce a high spatial resolution colour image. For wavelength) strongly interact with the surface of the
example, a 0.3-m pan image is fused with the 1.85-m canopy (low penetration), while longer wavelengths
multispectral (colour) bands to produce 0.3-m (e.g. L-band, 23.5-cm wavelength) penetrate further
colour. Although pan-sharpening delivers higher into the canopy and reflect information on branches
colour resolution, the process degrades valuable and stem structure (Figure 20). While shorter
spectral information contained in multispectral wavelengths can provide estimations for canopy
imagery that can be important for classifying height, the longer wavelengths are more strongly
features when mapping or deriving information correlated with forest biomass, structure and other
that relies on calculated percent reflectance values biophysical parameters. The polarization (e.g. HH or
(Figure 19). horizontal transmit and horizontal receive and HV
or horizontal transmit and vertical receive) of the
2.5 SYNTHETIC APERTURE RADAR FOR radar (the way the radar is transmitted and received)
DEFINING MANGROVE STRUCTURE can reveal moisture content and the geometry of
SAR sends measured microwave energy and records the mangrove canopy.
the backscatter signal, which is proportional to the High resolution X-band SAR datasets that can
size, shape and water content of the leaf (Simard, be purchased include ICEYE SAR18, TerraSAR-X/
2019). One of the main advantages of SAR is that it TanDEM-X19 and Capella Space20 systems. Each SAR
can penetrate cloud cover, making it particularly system typically has different observation modes,
useful in tropical areas of the world, and operates each with a different swath and spatial resolution
both during the day and at night. High spatial reso- to address a specific mapping purpose. Radarsat-2
lution SAR data are not widely available in the public is operated by the Canadian Space Agency and
domain, however these data can be purchased and has been operating since 2007, collecting data
used as an effective tool to measure forest structure, at a spatial resolution of 3 m in spotlight mode.
Figure 20. Different radar wavelengths penetrate and record canopy characteristics at different
depths of the forest canopy
Source: Simard, M. 2019. Radar remote sensing of mangrove forests. In: A. Flores, K. Herndon, R. Thapa & E.. Cherrington,
eds. NASA SAR Handbook: Comprehensive methodologies for forest monitoring and biomass estimation. NASA. DOI:
10.25966/33zm-x271
18
www.iceye.com/sar-data
19
www.intelligence-airbusds.com/en/8694-terrasar-x-tandem-x
20
www.capellaspace.com
Lower spatial resolution SAR datasets (30 m) used for coastal flood risk modelling but can also be
available in the public domain include the ESA’s used as a screening tool to model where mangroves
Sentinel-1, a medium wavelength C-band (5.6-cm)
21
are most likely to grow based on defined elevation
sensor operated by the ESA and the Japanese ranges (i.e. a habitat suitability model). AGB can be
Advanced Land Observing Satellite (ALOS PALSAR ), 22
estimated by combining the modelled bare earth
a longer L-band sensor that produced 25-m data product with DSMs.
from 2006 to 2011 and the ALOS-2 PALSAR-2 which 23
was launched in 2014 with upgrades and can operate 2.6 LIGHT DETECTION AND RANGING
in a 3 m spotlight mode with a 25-km swath. FOR DEFINING MANGROVE HEIGHT
SAR global products (Figure 21) that have been AND STRUCTURE
used for estimating tree canopy height and structure LiDAR is another active remote sensing system
include NASADEMHGT, the latest version of the
24
that can collect data from space, airplanes, UAS
Shuttle Radar Topography Mission (SRTM) data or terrestrial platforms. Unlike SAR that transmits
that is a global 30-m C-band-derived DSM product microwave energy, a LiDAR sensor sends thousands
and AW3D30, a DSM derived from the stereo-map-
25
of infrared laser pulses and records each individual
ping PRISM sensor on board the Advanced Land return in very precise x, y, z coordinates that can be
Observing Satellite (ALOS). Climate Central’s much more accurate than SAR. These LiDAR point
CoastalDEM® is a commercial 30-m global product
26
clouds can be used to visualize, quantify and classify
based on SRTM data which uses machine learning three-dimensional (3D) vegetation characteristics.
and ICESat-2 LiDAR data to remove vegetation/ However, space-based LiDAR is not well suited for
buildings to create a more accurate “bare earth” local-scale mapping projects due to the coarse
digital elevation model. These data are most widely spatial resolution and sensitivity to clouds.
Figure 21. Comparison of global digital elevation models (AW3D30 and CoastalDEM®) for mangroves
in Bajo Yuna National Park, Dominican Republic
©TNC
©TNC
©TNC
Adapted from: Acosta-Morel, M., McNulty, V.P., Lummen, N., Schill, S.R. & Beck, M.W. 2021. Shoreline solutions: Guiding
efficient data selection for coastal risk modelling and the design of adaptation interventions. Water, 13(6): 875.
Notes: For assessing mangrove height and canopy structure using SAR, there are several freely available global datasets.
Left: The mangrove forest in Bajo Yuna National Park, the Dominican Republic. Centre: The surface elevation as measured
by AW3D30, a global 30-m DSM derived from the stereo-mapping PRISM sensor on board ALOS. Right: The CoastalDEM®
model representing “bare earth” elevations.
21
https://sentinels.copernicus.eu/web/sentinel/missions/sentinel-1
22
https://asf.alaska.edu/data-sets/sar-data-sets/alos-palsar/
23
www.eorc.jaxa.jp/ALOS-2/en/about/palsar2.htm
24
https://lpdaac.usgs.gov/products/nasadem_hgtv001/
25
www.eorc.jaxa.jp/ALOS/en/dataset/aw3d30/aw3d30_e.htm
26
https://go.climatecentral.org/coastaldem/

Figure 22. Estimates of mangrove AGB and maximum canopy height in Manglares de Estero Balsa,
Dominican Republic
©TNC
©TNC
©TNC
Source: Simard, M., Fatoyinbo, T., Smetanka, C., Rivera-monroy, V.H., Castaneda, E., Thomas, N. & Van der stocken, T.
2019. Global mangrove distribution, aboveground biomass, and canopy height. Oak Ridge, Tennessee, USA, ORNL DAAC.
doi.org/10.3334/ORNLDAAC/1665
Notes: These data were modelled by Simard et al. (2019) and were derived using a 30-m resolution global mangrove
ecotype extent map and canopy height measurements from SRTM radar data with LiDAR data from the GLAS on board the
ICESat-1 satellite.
LiDAR data are used to capture mangrove forest Analysis and Discovery (GLAD) laboratory, University
structure, map individual trees, and predict volume of Maryland28 and in the Google Earth Engine
and biomass that can be used to estimate forest (Potapov et al., 2020).
carbon stocks. When combined with SAR data, Other global mangrove models (30 m) include:
even more accurate mangrove canopy height and the use of SRTM data to measure canopy height and
structural information can be calculated (Figure 22). new mangrove distribution (Aslan and Aljahdali,
The University of Maryland has developed a global 2022); and AGB and canopy height models created
forest canopy height layer, interpolated from by integrating SRTM data with LiDAR data from
Global Ecosystem Dynamics Investigation (GEDI) 27
the Geoscience Laser Altimeter System (GLAS),
data, a full-waveform LiDAR instrument that is the primary instrument aboard NASA’s Ice, Cloud
operated from the International Space Station and and land Elevation Satellite (ICESat-1) satellite (in
measures the vertical distribution of vegetation orbit from 2003 to 2009) (Simard et al., 2019). In
by recording the amount of laser energy reflected 2018, NASA launched ICESat-229 which measures
by plant material (stems, branches and leaves) at the height of the Earth’s surface using multiple
different heights above the ground. From GEDI laser altimeter beams and satellite LiDAR. Future
waveforms, four types of structure information mangrove products can be improved using LiDAR
can be extracted: surface topography; canopy data from GEDI and ICESat-2 with radar data from
height metrics; canopy cover metrics; and vertical the TerraSAR-X that will be calibrated based on field
structure metrics. A global forest canopy height measurement data.
model is freely available from the Global Land
27
https://gedi.umd.edu/
28
https://glad.umd.edu/dataset/gedi
29
https://icesat-2.gsfc.nasa.gov/
2.7 UNCREWED AERIAL SYSTEMS areas due to battery life and flying altitude/distance
The advancement of UAS technology and the constraints, disturbance by rain and wind, and
growing availability of low-cost consumer models national and local regulations that prevent UAS
provides an emerging monitoring tool for collecting data collection. Table 4 summarizes the advantages
remotely sensed data at a local scale that were and disadvantages associated with the use of UAS
previously difficult to acquire. This adaptable for mangrove mapping.
“personal remote sensing” platform permits highly Before purchasing a UAS, it is important to
customizable and cost-effective data collection on consider the desired end products (e.g. mangrove
demand (Klemas, 2015), bridging the gap between extent, biomass, health, change) and match the
microscale field measurements and macroscale most appropriate platform and payload sensor
satellite imagery. Typically flown at altitudes based on the project area and budget, as well as
below 120 m, UAS can be rapidly deployed below the software that will be used to plan the mission
cloud cover, capturing detailed data using special- to capture the data, then process the data into the
ized sensors and payloads to map and analyse end products.
ecological spatial patterns at custom temporal
scales (Joyce, Fickas and Kalamandeen, 2023). 2.7.1 Platforms and sensors
Being able to map predetermined areas at precise Key considerations in choosing the most appro-
times offers a distinct advantage for environmental priate platform include: the spatial extent of the
monitoring, such as tracking mangrove restoration area to be surveyed; conditions and location of
success at regular intervals. UAS technology has operation; and the type of sensor(s) the platform
been used to spatially document and quantita- needs to carry to collect the required data. When
tively track mangrove patterns at multiple levels purchasing an UAS, it is also important to consider
for investigating vegetation health, estimating the maximum flight time per fully charged battery
biomass, assessing storm damage and detecting and maximum radio transmission range (km). Some
deforestation (Navarro et al., 2020; Ayub et al., 2021; companies offer protection packages for a limited
Jiang et al., 2021; Zimudzi et al., 2021). Some of the time that cover any damage to the UAS hull, gimbal
limitations of UAS include inability to map large or camera sustained during normal use.
Table 4. Advantages and disadvantages of mapping mangroves using a UAS

ADVANTAGES DISADVANTAGES
Cost-effective and easier to deploy than crewed Flight limited by battery life and flying altitude
aircraft which results in mapping of smaller areas only
Quick generation of high-resolution data over a Launch area can be limiting
small area – orthophotos and point clouds
Can be launched on demand as weather permits – Limited by radio connection distance and affected
below clouds – with minimal staff by magnetic deviations
Interchangeable payloads and sensors Regulations that limit use
Increasing availability of UAS models and software/ Easily affected by weather (e.g. rain, wind)
apps for data collection and processing
Continuing advancements in functionality, flight Photogrammetry-derived point cloud limited by
time and payload capabilities canopy
Use of RTK/PPK for precise geolocation Can cause damage to property and injury to
people

Figure 23. Examples of a multirotor (left) and a fixed-wing (right) UAS
©TNC
©TNC
Notes: Multirotor UAS are more widely adopted, can be launched from tight spaces, have increased manoeuvrability and
easier portability. Fixed-wing UAS are more energy efficient, covering longer distances and mapping larger areas on a
single battery.
Table 5. Advantages of multirotor vs. fixed-wing UAS

MULTIROTOR FIXED-WING
Increased manoeuvrability – ability to hover Longer flight times and ability to cover a much
larger area in a single flight
Size and portability for smaller platforms Increased flight stability in windy conditions
Increased payload capacity for heavy lift multi- Easier to safely land and recover from motor power
rotors (e.g. LiDAR) loss
Generally less expensive Larger platforms can carry increased payload
More widely adopted with wider selection of More suitable for mapping at higher altitudes
models
Launching and landing in confined places VTOL UAS can land and take off in more confined
spaces
Better camera control – rotation around the x-axis Ideal for vertical (i.e. nadir) mapping operations
There are two broad categories of UAS: Fixed-wing UAS fly like an airplane and are more
multirotor and fixed-wing, each having its own energy efficient; they can cover longer distances
advantages and limitations (Figure 23, Table 5). and map larger areas on a single battery. Due to
Multirotor drones are more widely adopted and their larger size, they are more stable in wind and
often less expensive as there are more commercial can carry more payload. Some limitations include
products. The advantages of this type of UAS include larger size, low manoeuvrability, higher cost, lack
the ability to launch from tight spaces, increased of hovering capacity, less camera control, and the
manoeuvrability such as hovering, rotation on its need for larger areas for take-off and landing. New
own axis, good camera control (cameras in fixed- developments in Vertical Take Off and Landing
wings are not typically mechanically gimballed) and (VTOL) technology have overcome the problem of
more portability. Some of the limitations include requiring large open spaces for take-off and landing
shorter flight times, more moving parts to maintain of fixed-wing UAS.
(or serve as potential fail points), increased noise and Table 6 contains a listing of popular UAS plat-
smaller payload capacity (unless more expensive forms for mapping.
and larger heavy-lift multirotor systems are used).
Table 6. Widely used multirotor and fixed-wing UAS and technical specifications
MAKE MODEL PRICE MAX WEIGHT MAX SHUTTER SENSOR MAX MAX
(USD)* FLIGHT (Kg) RANGE SIZE RESO- VIDEO
TIME (km) (inch) LUTION RESO-
(min) (MP) LUTION
(fps)
MULTIROTOR (fixed payload)
Autel EVO Max 4T 9 000 42 1.6 20 Electronic 0.8 50 8K/30
Autel EVO II Pro 3 590 42 1.11 13 Electronic 1 20 6K/30
Autel EVO Lite+ 1 200 40 0.835 12 Electronic 1 20 4K/30
Autel EVO Nano 825 28 0.249 10 Electronic 1/2 48 4K/30
Autel EVO Nano+ 800 28 0.249 10 Electronic 0.8 50 4K/30
DJI Mavic 3E 4 000 45 0.915 15 Mechanical 4/3 20 4K/30
DJI Mavic 3M 5 000 45 0.951 15 Mechanical 4/3 20 G,R, 4K/30
(Multispectral) Multispectral RE,NIR
DJI Air2S 999 31 0.595 12 Electronic 1 20 5.4K/30
DJI Mini 3** 469 38 0.249 18 Electronic 1/1.3 12 4K/30
DJI Mini 3 Pro** 759 34 0.249 18 Electronic 1/1.3 48 4K/60
Parrot Anafi Ai 4 500 32 0.898 4 Electronic 1/2 48 4K/30
Parrot Anafi USA 7 000 32 0.496 5 Electronic 1/2 21 4K/30
Yuneec H520E/E90X 25 1.86 7 Electronic 1 20 4K/30
HEAVY-LIFT MULTIROTOR (interchangeable payloads)
DJI Inspire 3/ 16 500 28 4 (can 15 Mechanical Full 42 8K/75
X9-8K Air lift 1.3) frame
DJI Matrice 300 17 000 55 6.3 (can 8 Electronic 1/1.7 20 4K/30
RTK/ lift 2.7)
Zenmuse H20
DJI Matrice 350 19 000 55 6.5 (can 20 Mechanical Full 45 4K/60
RTK /Zen- lift 2.7) frame
muse P1
FIXED-WING
AgEagle SenseFly eBee 10 995 90 1.6 8 Mechanical 1 20 N/A
(non- VTOL) X/S.O.D.A.
Censys Sentaero 5/ 95 100 70 10.8 80 Mechanical Full 61 4K/30
Technologies Sony A7R IV frame
Delair UX11 16 000 59 1.5 53 Mechanical 1 21.4 N/A
Deltaquad Deltaquad 24 200 110 6.2 50 Mechanical Full 61 4K/30
Pro #Map/ frame
Sony A7R IV
Event 38 E400/Sony 18 900 90 9 6 Mechanical Full 42 N/A
RX1R II frame
FIXAR 007/ 20 400 59 5 60 Mechanical Full 61 4K/30
Sony A7R IV frame
Quantum Trinity Pro/ 29 000 90 5.75 7.5 Mechanical Full 42 N/A
Systems Sony RX1R II frame
Sentera PHX/Double 8 000 59 1.9 3.2 Electronic 1/2.3 12 B,G, N/A
(non- VTOL) 4K R,RE,NIR
Wingtra WingtraOne 29 900 59 3.7 10 Mechanical Full 61 N/A
(Gen II)/RGB61 frame
*Prices as of December 2023. Heavy-lift multirotors and fixed-wing platforms have interchangeable payloads and may not reflect
the listed price.
**For those with limited budgets, these less expensive drones can be used for mapping but may not be compatible with most mission
planning apps. These models can be programmed to fly using subscriptions to Dronelink, Map Pilot Pro, or Drone Harmony. Due
to their smaller size, the sensors will not provide the same level of detail that more expensive drones can deliver at the same flying
altitude. Low-cost post-processing options for creating orthomosaics include OpenDroneMap (local processing) and cloud-based
subscriptions to WebODM Lightning or Map Pilot Pro.

As UAS technology continues to improve, the the first of which is maintained by the United States
sensor payload is becoming more compact and of America and is called the Global Positioning
sophisticated regarding the data that can be col- System (GPS). Other systems include GLONASS
lected. Most consumer UAS come with a built-in (the Russian Federation), Galileo (the European
optical RGB (visible range) camera. In multirotor Union) and BeiDou (the People’s Republic of
versions this primary camera is mounted on a China). The position reported by a GNSS receiver
three-axis (tilt, roll, pan) gimbal that provides can be improved to centimetre level when it is
stability during flight and precise control during corrected using GNSS data collected at a known
data capture. Cameras vary in terms of the size location called a “base station” (Figure 24). This
of the sensor, the maximum image resolution (i.e. also requires a special receiver installed in the
megapixels), dynamic range and shutter type (i.e. UAS that acts as a “rover” and makes it capable of
global mechanical vs rolling electronic). The size recording the raw GNSS satellite signals, not just
of the sensor, quality of the lens and the image the position of the UAS. When this is done in real
resolution dictate the spatial resolution achieved at time with a data link between the base and the
a set flying height. Holding other camera parameters GNSS system in the UAS, it is said to be supplying
constant (particularly focal length), a 42-megapixel Real-Time Kinematic (RTK) “live” corrections.
camera flying at a height of 120 m will be able to
achieve a similar spatial resolution (1.5 cm/pixel) Figure 24. A GNSS base station collecting
as a 20-megapixel camera flying at 55 m. When position correction data during a UAS mission
travelling at similar speeds, mapping at the higher

altitude would cover almost double the area. A
larger detector (e.g. 2.5 cm) will collect imagery of
higher quality or fidelity. Image classification can
be done using JPEG image format, however, RAW
and TIFF formats are uncompressed and produce
higher quality images.
The type of shutter will affect the amount and
type of distortion in each image. A global mechanical
shutter opens and closes at the predefined shutter
speed, ensuring that each image on the image
sensor is exposed to light at the same time. This
helps to minimize distortion compared to a rolling
electronic shutter. A rolling electronic shutter, on
the other hand, exposes pixels on the image sensor
in a line-by-line order so it will introduce distortion
effects as it moves forward depending on the rate
at which each line is exposed to the scene.
In most modern systems, images are geotagged
as they are collected with the location coordinates
©TNC
embedded in the file header. The accuracy of the

Notes: An Emlid Reach RS2+ multiband GNSS receiver
coordinates is dependent on the quality of the with centimetre precision can be used to collect
GCPs or act as a base station for PPK corrections to
Global Navigation Satellite System (GNSS) receiver geotagged UAS images. This is especially critical for
change detection of UAS products when pixels need to
in the UAS. Several governments or multinational line up with centimetre accuracy.
groups have launched satellite positioning systems,
When the raw satellite data on the base and rover high levels of NIR and as the amount of chlorophyll
are stored and then processed after the UAS flight, produced in a plant decreases, less NIR is reflected.
it is said to provide Post Processed Kinematic (PPK) The Normalized Difference Vegetation Index (NDVI)
corrections. The primary advantage of RTK is that equation can be used to detect plant stress by
the positions are known when they are recorded, comparing the relationship between reflected
and no postprocessing is required. However, it intensities of NIR and red light. More information
requires a data link between the base and the can be found in the vegetation indices section of this
UAS (e.g. cellular, radio connection). PPK does not manual. There are lower cost NIR camera options
require this connection, can be more accurate and that can be mounted on smaller consumer drones,
will also tolerate a greater distance between the such as those made by MAPIR30 and Sentera31 that
base station and the rover (< 30 km). offer single and multi-sensor cameras and come in
If mapping objectives require assessment of a variety of filter transmission options (Figure 25).
mangrove health, a camera capable of recording The multirotor DJI Mavic 3 Multispectral has an
NIR reflectance is essential. A healthy plant reflects integrated RGB camera and 4-band multispectral
Figure 25. Two cost-effective UAS platforms with supplementary infrared sensors attached for
assessing mangrove extent and health
©TNC
©TNC
©TNC
©TNC
Top row: A Parrot Sequoia multispectral 4-band sensor attached to a DJI Phantom 4 Pr. Bottom row: a MAPIR Survey 3
NIR camera attached to a DJI Mavic 2 Pro. On each UAS there is a sunshine sensor attached to the top that is used for
automatic adjustment of readings to ambient light. Both sensors can be used to compute NDVI used to detect plant stress
by comparing the relationship between reflected intensities of NIR and red light.
30
www.mapir.camera
31
https://sentera.com

system (Figure 26). More precise and advanced produce ~ 1.6 cm pixel multispectral imagery flying
cameras can be operated from fixed-wing and/ at 100 m. The DUAL lens (the NDVI and Red-edge)
or heavy-lift multirotors (Figure 27). AgEagle32 can generate 4 bands at 10 MP and the QUAD lens
and Agrowing33 offer a variety of compact options can generate 12 bands at 12 MP. Many companies
including five multispectral bands (blue, green, red, offer cloud-based services that use artificial
red-edge and infrared) as well as panchromatic intelligence to automatically detect and identify
and thermal sensors. Adapted lenses compatible plant health and diseases based on a variety of
with the Sony A6100 and A7R/4 camera bodies can vegetation indices.
Figure 26. A DJI Mavic 3 Multispectral with a 20 MP RGB camera and integrated 4-band multispectral
sensors (green, red, red-edge and NIR) for calculating vegetation indices
©TNC
©TNC
Table 7. Popular multispectral sensors used on fixed-wing and heavy-lift UAS platforms
SENSOR WEIGHT SPECTRAL RGB GSD AT 120-M PRICE
(g) BANDS OUTPUT FLYING HEIGHT (USD)*
(cm)
Sony A7R-IV 835 10 narrow bands 12 MP per 1.65 14 500
with QUAD (405;430;450; band
multispectral 550; 560;570;650; Mechanical
lens 685;710;850) shutter
Micasense 350 Blue, green, red, 5.1 MP per 7.7 (MS) 10 000
RedEdge P red-edge, NIR band 3.98 (Pan)
Mechanical
shutter
Micasense 406.5 Blue, green, red, 3.2 MP per 5.28 (MS) 16 000
Altum PT + red-edge, NIR band 33.5 thermal
DLS 2 12.4 MP (Pan)
Mechanical
shutter
320 × 256
thermal
*Prices as of December 2023.
32
https://ageagle.com
33
https://agrowing.com
Figure 27. Two UAS platforms for mapping mangroves
©TNC
©TNC
©TNC
©TNC
Top row: A Parrot Sequoia multispectral 4-band sensor attached to a DJI Phantom 4 Pr. Bottom row: a MAPIR Survey 3
NIR camera attached to a DJI Mavic 2 Pro. On each UAS there is a sunshine sensor attached to the top that is used for
automatic adjustment of readings to ambient light. Both sensors can be used to compute NDVI used to detect plant stress
by comparing the relationship between reflected intensities of NIR and red light.
2.7.2 Mission planning and data collection the use of a mission planning app where key data
It is important to note that the mission planning collection parameters (such as area boundary with
takes much longer than the actual flying time. an adequate buffer, camera type and gimbal angle,
A variety of factors needs to be considered and flying altitude, front and side overlap, and shutter
planned prior to the launch, such as finding a safe speed) are specified. Most mission planning apps
launch area, planning for weather and suitable stream high resolution satellite image base maps to
environmental conditions, and having adequate enable accurate specification of the mapping area.
mission approval flying. Data from a UAS can be col- Some apps even permit the importing of Keyhole
lected manually or in autonomous mode where the Markup Language (KML) files that are designed in
GPS coordinates (i.e. waypoints) of flight lines are Google Earth or GIS vector layers. Flight mission
uploaded wirelessly and the UAS follows the flight planning will also reveal the amount of time needed
lines, capturing each image at a predetermined to execute the mapping task over the area of inter-
overlap setting. When flying in autonomous mode, est. Most mission planning apps allow for planning
good planning is the key to collecting accurate data of multiple battery missions within a large area,
that meet the research objective. This involves where the UAS will automatically return to home

Figure 28. Examples of UAS mission plans and the area mapped at varying flight altitudes
©TNC
©TNC
©TNC
©TNC
Notes: When planning a UAS mission, the flying height and sensor will dictate the imagery spatial resolution and area that
can be mapped. These maps show the spatial extent of single-battery flight missions planned at different flying heights
(120 m, 75 m, 50 m and 25 m) using a DJI Phantom 4 Pro. Each mission was planned with 70 percent front overlap and 70
percent sidelap settings and a maximum flying time of 25 minutes.
when battery levels are low. The pilot can then put In setting the flying height, a balance between
in a fresh battery and the UAS will return to the last the area coverage and the desired spatial resolution
data collection point and resume the mission. needs to be struck. Flying higher will cover more
It is best practice to plan flight missions in the area, but the data will have poorer spatial resolution
office where the area and flight parameters can be (Figure 28). To determine the appropriate flying
carefully assessed and reviewed. This makes field height, it is recommended to first acquire test data
execution much more efficient and timelier. During at varying heights (e.g. 50 m, 75 m, 100 m) under the
mission planning, the camera or the sensor to be same environmental and solar conditions to assess
used need to be specified as this influences the the resulting spatial resolution and the ability to
predicted overlap requirements and the calculated discriminate and identify features of interest. Sun
spatial resolution (i.e. the GSD). A vertical gimbal angle and weather can have significant influence on
angle (i.e. the camera points straight down, 90°) the data quality. Ideally, the flight height should be
is commonly used, however, more oblique angle set as high as possible to map a larger area, while
settings can be used for capturing the sides of still maintaining sufficient spatial resolution.
features when a more accurate 3D model needs to Specifying adequate forward overlap (along
be created. More information on UAS best practices a flight line) and side overlap (between flight
is available from the World Wildlife Fund.34 lines) is extremely important for the generation of
34
https://space-science.wwf.de/drones/WWF_CT_Drones_2020_web.pdf
Table 8. Flying heights and resulting spatial resolution and area mapped using a DJI Phantom Pro v2
using 70 percent front and side overlap
FLYING HEIGHT SPATIAL RESOLUTION AREA MAPPED ON ONE FULLY
(ABOVE-GROUND LEVEL) (cm) CHARGED BATTERY
(m) (ha)
25 0.7 5
50 1.4 21
75 2.1 45
100 2.7 78
120 3.3 94
Source: Authors’ calculated values.
Figure 29. UAS imagery acquired over mangroves at different times of the day (solar noon and early
morning) in Ashton Lagoon, Saint Vincent and the Grenadines
©TNC
©TNC
Notes: Time of day can produce significantly different illumination patterns when acquiring UAS data over mangroves.
The imagery on the left was acquired at high solar noon when sun glint on water and sunspots over the canopy was at the
highest intensity. The imagery on the right was collected early morning when the sun angle was low to minimize the glint
on the water.
high-quality point cloud and orthophoto mosaic. solar noon can minimize shadowing, which can sig-
Insufficient overlap will result in data gaps and nificantly affect high resolution multispectral data
outputs are likely to have artefacts or errors in and cause problems with classification. However, if
image alignment. There should be at least 70 there is a water surface in the study area, glint may
percent forward overlap and side overlap (Table 8). result from the higher sun angles and cause issues
To produce accurate terrain models, a minimum with postprocessing. Therefore, flying at high solar
of 80 percent forward overlap and 75 percent side noon under sunny conditions should be avoided as
overlap is recommended. Lower flying altitude may this can result in bright sunspots appearing in the
require greater percentage overlap (as well as a data (Figure 29). Data collection during overcast
faster shutter speed), but this depends on the type conditions can result in soft, diffuse lighting with
of sensor used. minimal shadowing.
Time of day (i.e. sun angle) and tidal cycles Calibrated reflectance panels can be used to
are important factors to consider when mapping create radiometrically calibrated data and should
mangroves. Collecting data within two hours of local be measured immediately before and after each

Figure 30. Comparison of UAS images acquired using a fast vs slow shutter speed
©TNC
©TNC
©TNC
©TNC
Notes: A slow shutter speed (right) during data capture will result in image motion blur which will affect the image clarity
and the ability of the photogrammetry software to recognize features and identify key points. Minimum shutter speeds
of at least 1/2 000 (left) are recommended (i.e. shutter priority) with the ISO and aperture set to automatic to adjust to
ambient light conditions. The likelihood for image blur is greater with lower flight altitudes (e.g. < 80 m).
mission flight. It is best to avoid capturing data when will indicate when such authorizations are required
there are rolling clouds which will intermittently and recommend steps to obtain permission.
change illumination conditions throughout the Relevant local and national laws should always be
flight. Under these conditions, it is best to not set consulted and obeyed, and priority placed on safety.
the camera’s white balance setting to automatic, The mission should be planned based on what
as this will cause non-uniform exposures in the detail and spectral information are needed in the
data. Instead, set it to cloudy or sunny depending data to answer the research question. There are
on the prevailing conditions. For optimal data various mission planning apps to choose from
collection, it is recommended to use the fastest along with their supported operating systems and
shutter speed possible based on light conditions estimated prices (Table 9). Examples of a few user
(e.g. 1/2 000 sec) and let the aperture and the ISO interfaces are shown in Figure 31. It is important
float to minimize image blur and facilitate feature to review the supported platforms and features
detection (Figure 30). of each app. Apps, such as Dronelink and Drone
When planning missions, it is important to check Harmony, employ a virtual stick approach and can
the airspace restrictions and regulations in the area be used with more affordable drones (e.g. Mavic
or the country where the mission will take place. Mini and Mavic Air series) that previously have
Some countries prohibit drone flights altogether, not been available for autonomous mapping. All
while others require remote pilot licences and/or mission planning apps allow for flight missions to
permits from the local aviation authority prior to be saved, exported and recalled for repeat use,
collecting field data. To operate UAS in restricted ensuring consistency in data collection between
airspaces, special authorizations (e.g. Low Altitude time periods. Some apps have a companion online
Authorization and Notification Capability or LAANC) account that enables synchronization of different
may be required. Many of the mission planning apps devices so that missions can be easily retrieved
Table 9. Examples of available UAS mission planning apps
SOFTWARE OS PRICE*
Pix4Dcapture Pro iOS and Android Free
DJI Pilot 2 iOS and Android Free
DJI GS Pro iOS tablets only Free
DroneDeploy Flight iOS and Android, desktop Free, integrated with paid
version of postprocessing app
Map Pilot Pro iOS and Android Free
PrecisionFlight iOS and Android Free
Autel Explorer iOS and Android for Autel drones Free
Spexi iOS and Android Free
Mission Planner Windows desktop only for Free
drones using ArduPilot-Pixhawk
Copterus iOS, supports most DJI drones USD 5 one time
Litchi iOS and Android USD 25 one time
DJIFlightPlanner iOS and Android USD 99 one time
dronelink Windows USD 50 one time & tiered
subscription
drone harmony Windows Tiered subscription
UgCS photogrammetry tool Windows, iOS and Linux Starting at USD 830
Site Scan for ArcGIS Windows Contact Sales
Figure 31. User interface examples of mission planning apps include Pix4Dcapture (top left),
DroneDeploy Flight (top right), DJI GS Pro (bottom left) and Map Pilot Pro (bottom right)
©TNC
©TNC
©TNC
©TNC

in the field. A few apps (DJI Pilot 2, Map Pilot Pro, 2.7.3 Data processing
DroneDeploy, DroneHarmony) have a feature called Once data have been collected in the field and
“Terrain Awareness” that integrates global eleva- downloaded to a computer, photogrammetry
tion products to adjust flightline flying altitude so software can be used to process individual images
the UAS maintains sufficient altitude above the into three primary output products: point clouds,
terrain and reduces the risk of collision in high Digital Surface Models (DSMs) and orthophoto
topographic areas. mosaics (Figure 32). A point cloud is a set of mil-
There are several companion apps that can lions of georeferenced points in 3D space that is
assist in preparing for field data collection. These generated from surface matching and triangulation
include: AirMap and AirHub Launch (for reviewing within overlapping stereo images. The location
airspace and restrictions); Aloft (LAANC author- accuracy of each point is based on the precision
ization and flight tools); UAV Forecast (weather of the GPS that was used in the geotagging of the
forecasts specific to remote pilots); Sun Surveyor images. A DSM is a georeferenced surface that is
(planning for sun angles during the missions); created from the point cloud and represents a
Measure Ground Control (drone deployment and solid plane connecting all the points. The DSM will
management); and GPSdiagnostic (GPS planning represent all features from which the point cloud is
tools). There are also some insurance apps such generated – trees, buildings and the ground when
as SkyWatch and Verifly offering protection plans it is not obscured. The orthophoto mosaic is the
that can be purchased by the hour, month or year final product that is based on the DSM, in which
with limits up to USD 5 000 000. all geometric distortion has been removed from
individual images (orthographically corrected) and
a seamless GIS-ready mosaic is calculated.
Figure 32. UAS products of a mangrove area created from 85 overlapping stereo images taken at
different perspectives and processed using photogrammetry software
©TNC
©TNC
©TNC
©TNC
©TNC
©TNC
Notes: The resulting orthophoto mosaic was computed to a 1.3 cm pixel based on 4.9 million points generated in the point
cloud. The vegetation index shown is the VARI method that estimates presence of vegetation based only on the visible
range of the spectrum (i.e. RGB).
Table 10. Options for postprocessing UAS data into photogrammetric products
SOFTWARE OS PRICE*
COLMAP Windows, MacOS, Linux Free – 3D models only
Meshroom Windows, Linux Free – 3D models only
openMVG Windows, MacOS, Linux, Free – 3D models only
Android, iOS (code only)
Regard3D Windows, MacOS, Linux Free – 3D models only
VisualSFM Windows, MacOS, Linux Free – 3D models only
WebODM Windows, MacOS, Linux Free or USD 57 for installer
WebODM Lightning Cloud-based Pay as you go, from USD 35/month
MapsMadeEasy Cloud-based Pay as you go, from USD 5/month
Photomodeler Windows From USD 59/month
Mapware Cloud-based From USD 99/month
Agisoft Metashape Windows, MacOS, Linux From USD 550 (academic)
PIX4Dmapper Windows, MacOS, Cloud From USD 260/month
Simactive Correlator3D Windows From USD 295/month
3DFlow Zephyr Windows From USD 300/month
SpexiGeo Cloud-based From USD 300/month or by credit
DroneDeploy Cloud-based From USD 329/month
Autodesk ReCap Windows USD 360/year
iWitnessPro Windows USD 2 495
Bentley ContextCapture Windows From USD 3 647
RealityCapture Windows By credit or USD 3 750 unlimited
Skyline – PhotoMesh Windows or Linux Call for price
TBC photogrammetry Windows Call for price
Mapper UgCS Windows, MacOS, or Linux USD 50/month or USD 500/
perpetual
ArcGIS Drone2Map Windows Call for price
Site Scan for ArcGIS Windows Call for price
There are desktop and cloud-based options Some of the widely used photogrammetric software
available for photogrammetry software. Desktop packages are listed in Table 10 with their system
software provides the user with more control of the operating requirements and pricing. Several free
output and the ability to customize more parame- options are available, however they might not create
ters. However, it requires a robust workstation to an orthophoto mosaic. The more expensive options
process the huge amount of data that is generated. offer trial periods for users to experiment with and
This option is best for remote areas with slow inter- test the software. It is up to the users to evaluate
net access. Cloud-based options have a simpler the workflow and orthophoto mosaic generation
interface, can produce results faster (provided that process to determine which one best meets their
there is a fast internet connection) and enable team needs based on budget, accuracy requirement and
collaboration through web-sharing of products. skill level.

Of the low-cost options, WebODM (Open Drone between overlapping images and calculates each
Map) is a popular open-source drone mapping point’s 3D coordinates. The resulting dense point
app that has been improving over the years as it cloud is then used to create a 3D polygonal mesh
is maintained by a large community of software model and DSM. Finally, the software uses the DSM
developers. The software allows for creation of to project every pixel and generate the orthophoto
diverse types of maps from JPEG and TIFF images mosaic which is a planimetrically correct image
and processes multispectral images to calculate with all geometric distortions removed. The
different vegetation indexes. WebODM Lightning is geometric correction Root Mean Square Error
a cloud-based version and offers a Pay As You Go (RMSE) is reported in both horizonal and vertical
option as well as a Pro option for USD 35/month for dimensions. A rayCloud feature can be used to
unlimited maps and 100 GB of cloud space. visually assess the quality of the reconstruction
The postprocessing workflow starts with the showing the GPS-derived position of the cameras
orientation and alignment of all images using the (Figure 33).
exchangeable image file format (EXIF) header, geo- The collection of UAS stereo images from both
tagged image information and advanced bundle vertical and varying oblique angles can be used to
block adjustment. Next, an algorithm searches create a realistic 3D model of a mangrove forest
for and matches millions of points (i.e. key points) which can be useful for estimating above-ground
Figure 33. The rayCloud camera positions showing the location where each stereo image was taken
©TNC
Notes: Using photogrammetry software, data can be assessed and interpreted using the rays connecting the 3D point
cloud to the original input images.
Figure 34. Comparison of the original image of a mangrove and a 3D point cloud35 generated from
stereo images using Agisoft Metashape photogrammetry software
©TNC
©TNC
biomass and carbon storage (Figure 34). When orthophoto mosaic fails, it is typically the result
acquiring these data, it is important to avoid of low key point generation between overlapping
having the sky in the images which can confuse images. This can be due to insufficient overlap, a
the photogrammetric algorithm. Point clouds homogeneous environment or poor image quality,
may have spurious artefacts that can be manually such as image blur (due to slow shutter speed).
cleaned up using tools available in certain software Following data processing, a quality report is typ-
(e.g. Metashape, Pix4D). Without high precision ically generated by the software providing product
geometric correction (either RTK or PPK), geometric details including output GSD, total area mapped,
accuracy of the resulting orthophoto mosaic using median shutter speed, geometric accuracy (i.e.
the onboard GPS is typically between 3 cm to 6 cm RMSE), points generated in the point cloud, and
horizontal (depending on the UAS model used). the percent of images that are aligned (i.e. good
When conducting change detection between stereo coverage vs poor coverage). For sharing
images collected at different times, it is important data, point clouds and DSMs can be uploaded to
to align the pixels using RTK or PPK methods or OpenTopography.org and orthophoto mosaics to
by using GCPs based on accuracy requirements OpenAerialMap.org.
(usually centimetre level). If the processing of the
35
https://tnc.app.box.com/s/3zv56y84042wdjrnya37yusk1532mt54

©TNC
3. MANGROVE MAPPING
TECHNIQUES
3.1 VISUAL INTERPRETATION METHOD and mapping their location. This method is labour-
The simplest method of mapping mangrove forests intensive, highly subjective to the digitizer and subject
is to digitize the boundaries using head-up manual to human error. As such, field accuracy assessments
techniques based on visual interpretation of high and review through local knowledge are important.
spatial resolution imagery (< 5 m pixel) coupled Having imagery with sufficient spatial resolution
with fieldwork and local knowledge. In order to and access to infrared bands also facilitates the
successfully recognize a mangrove forest, the image recognition of mangrove presence or absence.
analyst must consider a variety of image interpretation There are free GIS resources that can digitize
elements such as the forest canopy’s tone and polygon features using a georeferenced image
colour, size, shape, texture, location and situation. base map (Figure 35). Open-source solutions,
Mangroves require specific biophysical conditions to such as Quantum GIS36 (QGIS) software, provide
establish and grow (e.g. elevation, salinity, nutrient open access to stream-archived high resolution
requirements) so recognizing where these conditions image basemaps maps, such as those available in
are likely to occur is critical for successfully identifying Google Earth and Microsoft Bing aerial imagery.
Figure 35. Example of mangrove digitization in QGIS with multiple basemaps

©TNC
©TNC
©TNC
Notes: Mangrove extent can be digitized using open-source GIS software such as QGIS that permits users to stream high
resolution satellite image basemaps for free. Figure 35 shows three satellite basemaps of different satellites and collection
periods pulled from satellite archives. It is important to check each basemap to determine the optimal imagery to digitize from.
36
www.qgis.org/en/site
37
https://livingatlas.arcgis.com/wayback/
38
www.planet.com/stories
45
In the desktop version of Google Earth, historical Other sources for older imagery include Esri’s
imagery going back to the early 2000s is available World Imagery Wayback 37 tool and Planet’s
using the Timeline tool. Using Google Earth, polygons Stories 38 with which users can build custom image
can be digitized in KML format, then imported into change animations for anywhere in the world going
GIS for further analysis (Figure 36 and Figure 37). back to 2015.
Figure 36. Tracking the impacts of marina development on a mangrove forest in Playa de Maimon,
Dominican Republic using the timeline tool in Google Earth
©TNC
©TNC
©TNC
©TNC
©TNC
©TNC
Figure 37. Tracking the change in mangrove forests at the mouth of the Barracote River in Bajo Yuna
National Park, Dominican Republic using the timeline tool in Google Earth
©TNC
©TNC
©TNC
©TNC
3.2. AUTOMATED CLASSIFICATION METHOD neighbouring (often spectrally similar) forest types.
Automated methods for mapping mangroves Spatial patterns of mangroves vary greatly between
require specialized image processing software that regions and even between neighbouring countries,
uses pixel and object-based classification algo- so field data and local knowledge are crucial for
rithms. Pixel-based means each pixel is assigned informing localized mangrove classifications.
to a class while object-based uses groups of pixels Automated pixel-based classification can be
that have similar spectral values and are classified performed using QGIS which runs on Linux, Unix,
together (Figure 38). These methods can be much Mac OSX, Windows and Android operating systems.
less labour-intensive than manual digitization, but The latest stable version and the required packages
require quality imagery, adequate field data and and plugins should be downloaded and used. 39
technical expertise. Classifications can be informed There are numerous tutorials40 on how to classify
and improved using additional GIS data layers on imagery using open-source software. Prior to
related ecosystem components (e.g. soil, elevation) conducting the classification, the user will need
and field data. When collecting field data, GPS- to determine which and how many classes there
referenced terrestrial and/or aerial UAS images will be. This decision should be determined by the
or videos of habitats can help identify different research question, available data and resources.
types of mangroves and the transition boundaries In a supervised classification, the user selects
to other habitat types. UAS images and videos along representative training data from the image for
strategically placed transects can be particularly each class on which the classification is based.
useful in spatially documenting complex mangrove These training data should represent the full var-
growth patterns and distinguishing mangroves from iation that exists within each class. They are also
Figure 38. Segmenting a satellite image means creating objects around pixels that have similar
spectral properties and reflectance values
©TNC
©TNC
©TNC
©TNC
©TNC
©TNC
Notes: The scale of the objects can be specified to make them larger or smaller, depending on the mapping objecive.
Object-based classification means it is the objects, not the individual pixels, that are being classified. These results show
the same scale of objects applied to three different satellite images with different spatial resolutions over mangroves in
Ashton Lagoon, Saint Vincent and the Grenadines (from left to right): SkySat (0.5 m), PlanetScope SuperDove (4 m) and
Sentinel-2 (10 m). Beneath are the results of a simple three class classification applied to the objects. As the pixel size
increases, the segments become coarser and the detail of the classification becomes less clear.
39
https://plugins.qgis.org/
40
https://servirglobal.net/Portals/0/Documents/Articles/ChangeDetectionTraining/Module3_LC_Classification_Accuracy_Assessment.pdf
MANGROVE MAPPING TECHNIQUES 47

©TNC
used in the classification algorithm to learn the and Deep Learning methods, which are quickly
spectral patterns for each class and establish the emerging, utilize libraries of specimen training
rules upon which each pixel will be assigned to a labels of mangrove variations and assign classes
class. Unsupervised classification does not require based on recognized learned patterns, both spectral
training data and automatically assigns pixels to a and contextual. QGIS has several classification
statistically defined spectral cluster based on their plugins such as the Semi-Automatic Classification
shared spectral signatures. It is then up to the user plugin41 and other Deep Learning plugins42 that can
to determine what each of the clusters represents be used to classify imagery. In addition, the Orfeo
in terms of land cover class. ToolBox43 (OTB) is an open-source software library
There are various image classification algo- for processing imagery and operates within the
rithms and each one has its advantages. The popular QGIS interface. A user can perform pixel-based or
Random Forest algorithm mines the input data, object-based classification using the OTB. Whatever
then builds and operates on complex decision trees classification method is used, it is important to
to assign each pixel to a class. Support Machine refine outputs with manual corrections and conduct
Vector works best with highly complex imagery and an accuracy assessment (in field or expert review)
K-Nearest Neighbors (KNN) performs well with noisy when available resources permit.
training data. Finally, Convolution Neural Network
41
https://plugins.qgis.org/plugins/SemiAutomaticClassificationPlugin/
42
https://plugins.qgis.org/plugins/tags/deep-learning/
43
www.orfeo-toolbox.org
3.3 REMOTE SENSING-DERIVED INDICATORS Soil-Adjusted Vegetation Index (SAVI). VIs based
AND METHODS FOR FEATURE EXTRACTION on RGB bands include the Triangular Greenness
Index (TGI), the Visible Atmospherically Resistant
3.3.1 Vegetation indices Index (VARI) and the Excess Green Index (ExGI)
The spectral resolution of a remote sensing system (Table 11). TGI is more effective in separating the
dictates the segments of the electromagnetic vegetation from water than VARI and ExGI slightly
spectrum that are recorded in spectral bands, such outperformed other indices in discriminating
as red, green, blue, NIR and SWIR. These individual vegetation cover.
spectral bands can be used as inputs in vegetation Specialized VIs have been developed to distin-
indices (VIs) to highlight a particular property of guish mangroves from non-mangrove vegetation
vegetation. VIs are derived using the reflectance (Ali and Nayyar, 2020). These include the Combined
properties of vegetation and each one is designed to Mangrove Recognition Index (CMRI), the Normalized
accentuate a particular vegetation property. Some Difference Mangrove Index (NDMI), the Landsat 8
of the more common VIs that are used to reliably Mangrove Index (L8MI) and the Mangrove Vegetation
extract vegetation pixels from non-vegetation Index (MVI). CMRI distinguishes mangroves using
pixels include the Normalized Difference Vegetation NDVI and the Normalized Difference Water Index
Index (NDVI), the Ratio Vegetation Index (RVI), (NDWI) that are negatively correlated and are used
the Enhanced Vegetation Index (EVI) and the to help separate mangroves from other vegetation.
Table 11. Common vegetation indices and their corresponding formulas

VEGETATION INDEX FORMULA
Normalized Difference Vegetation Index (NDVI) (NIR – Red)/(NIR + Red)
Ratio Vegetation Index (RVI) NIR/Red
Enhanced Vegetation Index (EVI) 2.5*((NIR – Red)/(NIR +6 * Red-7.5*Blue+1))
Soil-Adjusted Vegetation Index (SAVI) ((NIR – R)/(NIR + R + L)) * (1 + L)
soil brightness correction factor (L) defined as 0.
Triangular Greenness Index (TGI) GREEN-(0.39*RED)-(0.61*BLUE)
Visible Atmospherically Resistant Index (VARI) GREEN-RED
GREEN+RED-BLUE
Excess Green Index (ExGI) 2 * GREEN – (RED + BLUE)
Combined Mangrove Recognition Index (CMRI) NDVI – NDWI
where NDWI = (Green – NIR)/(Green + NIR)
Normalized Difference Mangrove Index (NDMI) (SWIR2 – Green)/(SWIR2 + Green)
Landsat 8 Mangrove Index (L8MI) L8MI1 = [ASST1 >T21] and SAVI > T1
L8MI2 = [ASST2 > T22] and SAVI > T1
Mangrove Vegetation Index (MVI) |NIR – Green|/|SWIR – Green|
Sources: Ali, A. & Nayyar, Z.A. 2020. Extraction of mangrove forest through Landsat 8 Mangrove Index (L8MI). Arabian Journal of
Geosciences, 13(21): 1–12; Baloloy, A.B., Blanco, A.C., Ana, R.R.C.S. & Nadaoka, K. 2020. Development and application of a new mangrove
vegetation index (MVI) for rapid and accurate mangrove mapping. ISPRS Journal of Photogrammetry and Remote Sensing, 166: 95–117;
Barr, J.R., Green, M.C., DeMaso, S.J. & Hardy, T.B. 2018. Detectability and visibility biases associated with using a consumer-grade
unmanned aircraft to survey nesting colonial waterbirds. Journal of Field Ornithology, 2018, 89: 242–257; Boon, M.A., Drijfhout, A.P.
& Tesfamichael, S. 2017. Comparison of a fixed wing and multirotor UAV for environmental mapping applications: A case study. The
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2017, XLII-2/W6, 47–54; Gupta, K.,
Mukhopadhyay, A., Giri, S., Chanda, A., Majumdar, S.D., Samanta, S., Mitra, D., Samal, R.N., Pattnaik, A.K. & Hazra, S. 2018. An index
for discrimination of mangroves from non-mangroves using LANDSAT 8 OLI imagery. MethodsX, 5: 1129–1139; Huete, R.A. 1988.
A soil-adjusted vegetation index (SAVI). Remote Sensing Environment, 25(3): 295–309; Jordan, C.F. 1969. Derivation of leaf-area
index from quality of light on the forest floor. Ecology, 50(4): 663–666; McFeeters, S.K. 1996. The use of the Normalized Difference
Water Index (NDWI) in the delineation of open water features. International Journal of Remote Sensing, 17(7): 1425–1432; Ramsey
Ill, E.W. & Jensen, J.R. 1996. Remote sensing of mangrove wetlands: relating canopy spectra to site-specific data; Rouse, J.W. Jr.,
Hasas, R.H., Schell, J.A. & Deering, D.W. 1973. Monitoring the vernal advancement and retrogradation (green wave effect) of natural
vegetation. Remote Sensing Center, Texas A&M University, College Station; Shi, T., Liu, J., Hu, Z., Liu, H., Wang, J. & Wu, G. 2016. New
spectral metrics for mangrove forest identification. Remote Sensing Letters, 7(9): 885–894.

NDMI extracts mangrove forest pixels with some 3.3.2 Change detection using Normalized
inclusion of barren land pixels. NDMI takes advantage Difference Vegetation Index
of the decreased reflectance of SWIR in mangrove It is important to monitor changes in mangrove
forests and the increase in reflectance of the green biomass to recognize where mangroves are under
band in mangrove forests that naturally grow denser. stress and where mangroves are growing. Such
This index has shown success in identifying forest dis- information can guide restoration planning and
turbance and recovery in terms of water stress. MVI mitigation of threats.
discriminates the distinct greenness and moisture Biomass can be monitored through change
of mangroves from other vegetation utilizing three analyses of various vegetation indices, most
Sentinel-2 bands (Green, NIR and SWIR1) with high commonly NDVI. NDVI is a normalized ratio of red
accuracy. The NIR-Green relationship captures the and NIR spectral reflectance and is one of the most
difference of greenness between mangrove forests widely used vegetation indices. NDVI is determined
and other trees, while the SWIR-Green relationship by the degree of absorption by chlorophyll in the
identifies the distinct moisture of mangroves without red wavelengths, which is proportional to leaf
the need for additional water indices. In comparison chlorophyll density, and by the reflectance of NIR
with other vegetation indices, MVI values are high radiation, which is proportional to green leaf density
for mangrove vegetation only and can be utilized as and serves as a proxy of vegetation productivity.
a single input for index-based mangrove mapping Therefore, NDVI can be used as a surrogate to help
(Ali and Nayyar, 2020). evaluate the health status of mangroves. Figure 39
shows how NDVI can be used to monitor the recovery
of mangroves following a hurricane.
Figure 39. Using a DJI M300RTK multirotor UAS with a Micasense RedEdge MX to capture and compute
NDVI to track mangrove recovery following a hurricane
©WILL GREENE/PE RRY INSTITUTE FOR MARINE SCIENCE
©TNC
©TNC
©TNC
©TNC
Left: A DJI M300RTK multirotor UAS with a dual-sensor payload consisting of a DJI P1 full frame camera (RGB) and a Micasense
RedEdge MX Dual (10-band multispectral system). This setup allows for pixel-aligned outputs with centimetre-level accuracy
and coverage up of to 81 ha on a single battery. Flown at 120 m, it provides 1.5 cm RGB and 8 cm calibrated multispectral
imagery. Centre: Hurricane-damaged red mangroves in the Bahamas depicted with this system. Right: Computed NDVI values
provide spatial investigations into mangrove recovery. The red and orange hues indicate regrowth while the greens and blues
signal no photosynthesizing activity. The top half shows the ability of the system to pick up individual surviving plants, while
the bottom shows broader spatial mortality patterns that are difficult to distinguish with visual data alone.
Figure 40. TNC’s Blue Carbon Explorer showing an example of NDVI change analysis in eastern Grand
Bahama between 2015 and 2021 using Planet NICFI Dove imagery
©TNC
Notes: Hurricane Dorian ravaged mangroves in Grand Bahama and Abaco Island in the Bahamas in 2019. Using the Blue
Carbon Explorer, these results show levels of recovery based on NDVI (areas in red indicate loss and green indicate recovery)
and temporally through an NDVI time series chart. These types of data can be used to detect where reforestation efforts are
needed.
Change in biomass can be monitored by ana- to which NDVI analysis can be clipped, showing
lysing differences in NDVI across time periods, both growth and loss of biomass within the footprint of
spatially through change maps and temporally a mangrove area. The varying levels of NDVI values
through NDVI time series (as shown in TNC’s Blue can be used to investigate changes in mangrove
Carbon Explorer) (Figure 40). This type of analysis health, biomass, extent, and seasonal change.
requires a mangrove spatial footprint (vector layer)
©TNC

Figure 41. Examples of using natural colour (RGB) UAS imagery to calculate VARI to inform mangrove
change detection in Bajo Yuna National Park, Dominican Republic
©TNC
©TNC
©TNC
©TNC
©TNC
©TNC
©TNC
©TNC
©TNC
©TNC
Analysis of NDVI change can highlight areas NDMA by Aljahdali, Munawar and Khan (2021), MSAVI
and the timing of die-off events that allow for performed best in capturing various trend patterns
further investigation of events such as hurricanes related to greenness to vegetation, and NDMI better
or deforestation with resultant mangrove loss. A identified forest disturbance and recovery in terms
study completed on Abaco, the Bahamas, showed of water stress.
little evidence of a relationship between NDVI Vegetation health can be assessed at a larger
and hurricanes or drought events, but found that scale via analyses of various environmental
leaf-eating herbivores likely facilitated the spread indicators, such as the Mangrove Quality Index
of disease, contributing to the dieback (Rossi et al., (MQI) developed by Faridah-Hanum et al. (2019)
2020). These indices can also be used to track the which considers contributing components of a
recovery of restoration actions or natural expansion mangrove forest, including the soil, surrounding
in areas such as river deltas (Figure 41). Seasonality marine ecosystem, hydrology and the socio-
in mangrove biomass is apparent in NDVI, EVI and economic variables. While field measurements
NDWI analyses, where greenness is negatively such as DBH, canopy closure and canopy density
correlated with litterfall (Pastor-Guzman, Dash are extremely valuable for monitoring mangrove
and Atkinson, 2018). As such, it is recommended to health and biomass at the site level, MQI can be
compare NDVI using an annual average to under- used to scale up monitoring efforts while reducing
stand long-term trends, or before and after specific human resources, time and cost (Faridah-Hanum
events for spot damage assessments. et al., 2019). Analysis of environmental indicators or
pressures can also help identify drivers of change.
3.3.3 Change detection using other indices Maina et al. (2021) found that catchment erosion,
Beyond NDVI, EVI has also been found to be useful human pressure, sea level and macroclimate are the
in detecting changes in mangrove cover both main drivers of the present-day ecological condition
seasonally and long term, particularly in large of mangroves, and that NDVI was more sensitive to
ecosystems (Berlanga-Robles and Ruiz-Luna, 2020). these drivers than vegetation condition index (VCI).
The Modified Soil-Adjusted Vegetation Index (MSAVI) Changes in area, or extent, of mangroves require
and Normalized Difference Moisture Index (NDMA) derivation of the mangrove footprint (vector layer)
can also be used as proxies for plant biomass at various points in time. This can be done using
(Table 12). In an analysis of NDVI, EVI, MSAVI and CMRI, L8MI, NDMI or MVI as already discussed.
Table 12. Formulas for MSAVI and NDMA both useful for detecting changes in mangrove forests
INDEX FORMULAS
MSAVI (2 * NIR + 1 − sqrt((2 * NIR + 1)² − 8*(NIR − Red))/2
NDMA (NIR − SWIR1)/(NIR + SWIR1)

©TNC
4. ONLINE RESOURCES
FOR MANGROVE MAPPING
There are numerous published datasets for historical summarizes mangrove data covering habitat maps,
mangrove coverage, with the earliest starting in 2000. biological characteristics such as mangrove height
Table 13 provides a directory of online resources and biomass, as well as ecosystem services, such as
for mangrove maps and tools while Table 14 coastal protection, tourism and blue carbon storage.
Table 13. Directory of online resources for mangrove maps and tools
APPLICATION MAPS AND RESOLUTION GEOGRAPHIES URLS
TOOLS (m)
Ready-to-use TNC Blue 1 and 4 TNC has mapped https://BlueCarbon.tnc.org
maps Carbon mangroves at regional
Explorer and national scales at
1 m and 4 m resolution
Global ~25 (0.8 arc Global www.globalmangrovewatch.
Mangrove seconds) org
Watch
UNEP-WCMC Various Global https://data.unep-wcmc.org/
Ocean Data mangrove
Viewer datasets
WorldCover 10 Global https://worldcover2021.esa.int/
Imagery Planet NICFI 4.77 The four-band www.planet.com/nicfi
sources mosaics cover most
tropical forested
regions of the world
EarthExplorer Various datasets USA & global datasets https://earthexplorer.usgs.gov/
Analysis QGIS N/A N/A www.qgis.org/en/site
Orfeo ToolBox N/A N/A www.orfeo-toolbox.org
Google Earth Various datasets Global https://earthengine.google.
Engine com/
Ecosystem Blue Carbon 4.77 (Planet Caribbean, Indonesia, https://BlueCarbon.tnc.org
service tools Explorer NICFI), and Papua New Guinea
10 (Sentinel-2), (as of April 2023)
and 30
(Landsat-7 and
Landsat-8)
Mapping Varies Global https://maps.oceanwealth.org/
Ocean Wealth
InVEST Flexible, can be Global https://naturalcapitalproject.
used at local, stanford.edu/software/invest
regional or
global scales
Adapted from: Worthington, T.A., Andradi-Brown, D.A., Bhargava, R., Buelow, C., Bunting, P., Duncan, C., Fatoyinbo, L., Friess, D.A.,
Goldberg, L., Hilarides, L. & Lagomasino, D. 2020. Harnessing big data to support the conservation and rehabilitation of mangrove
forests globally. One Earth, 2(5): 429–443.
55
56
Table 14. List of mangrove datasets
DATASETS DESCRIPTION NOMINAL RESOLUTION MANGROVE DOWNLOAD OR VIEWER REFERENCES
YEAR EXTENT
USED
Mangrove Composite extent map using remote-sensing and visual- 1999–2003 – – https://data.unep-wcmc.org/ Spalding,
extent and interpretation approaches datasets/5 Kainumu and
change Collins (2010)
First globally consistent remote-sensing-based map of mangrove 2000 30 m – https://data.unep-wcmc.org/ Giri et al. (2011)
extent datasets/4
Giri et al. (2011) dataset refined by the removal of areas above an 2000 – Giri et al. (2011) Tang et al. (2018)
elevation threshold
Global analyses of mangrove deforestation based on the Global Forest Annual 30 m – http://faculty.salisbury.edu/ Hamilton and
Cover dataset 2000–2012 sehamilton/mangroves Casey (2016)
Most current global analysis of extent captures both losses and gains 1996, 25 m – https://data.unep-wcmc.org/ Bunting et al.
over a 20-year period 2007–2010, datasets/45 (2018)
2015–2020
Mangrove Climate-driven model of potential mangrove AGB – – Spalding, Hutchison et al.
biomass Kainumu and (2014)
Collins (2010)
Mangrove Canopy height maps based on a digital elevation model and LiDAR 2000 30 m Giri et al. (2011) https://doi.org/10.3334/ Simard et al.
height and altimetry ORNLDAAC/1665 (2019)
biomass
Canopy height maps based on a digital elevation model; biomass 2000 – Giri et al. (2011) Tang et al. (2018)
derived from global allometric model
Freshwater Changes in mangrove extent are modelled against human alteration – – Bunting et al. Maynard et al.
and sediment to free-flowing rivers (2018) (2019)
impact on
mangrove
condition
REMOTE SENSING TECHNIQUES FOR MAPPING AND MONITORING MANGROVES AT FINE SCALES
Mangrove Global analyses of the change in fragmentation metrics over time Annual 0.2° x 0.2° Hamilton and Bryan-Brown
fragmentation 2000–2012 Casey (2016) et al. (2020)
Table 14. List of mangrove datasets (continued)
DATASETS DESCRIPTION NOMINAL RESOLUTION MANGROVE DOWNLOAD OR VIEWER REFERENCES
YEAR EXTENT
USED
Soil carbon Covariates of climate and location data modelled against – ~10 km Giri et al. (2011) Jardine and
measurements of soil carbon Siikamäki (2014)
Assessment of how soil carbon stocks vary across latitude, 2014 – Hamilton and Atwood et al.
hemispheres and mangrove community composition Casey (2016) (2017)
Fine-scale three-dimensional variation in soil-carbon density as 2000 30 m Giri et al. (2011) https://dataverse.harvard.edu/ Sanderman et al.
assessed by machine-learning approaches dataset.xhtml?persistentId= (2018)
doi:10.7910/DVN/OCYUIT
Variation in soil carbon examined in relation to coastal environmental – ~25 km Hamilton and Available from the Rovai et al.
settings via climate-geophysical models Casey (2016) corresponding author upon (2018)
special request
Mangrove soil-carbon stocks across different classifications of coastal – 30 m Hamilton and Twilley, Chen
environmental settings Casey (2016) and Hargis
(1992)
Above-ground Field measurements modelled against latitude for estimating total – – WRI and IIED Twilley, Chen
and below- biomass carbon (1986) and Hargis
ground carbon (1992)
Mangrove AGB modelled against latitude; BGB assessed as a relative – ~9 km Giri et al. (2011) Siikamäki et al.
fraction of AGB (2012)
Total carbon Annual assessment of total carbon stocks and losses from Annual 30 m Hamilton https://dataverse.harvard.edu/ Hamilton and
deforestation 2000–2012 and Casey dataverse/GMCSD Friess (2018)
(2016)
Mangrove Analysis of TripAdvisor website to identify mangrove attractions and Up to 2015 – – https://maps.oceanwealth.org Spalding and
tourism their usage Parrett (2019)
Coastal Global valuation model of the role of mangroves in reducing annual 2010 20 km Giri et al. (2011) https://osf.io/ecs4p/ Losada et al.
protection coastal flood damages to people and property (2018)
Conservation Conservation hotspots identified by the intersection of threatened 2016 ~20 km Hamilton https://megafauna.wetlands. Sievers et al.
hotspots megafauna distributions with areas of high mangrove loss and Casey app (2019)
(2016)
ONLINE RESOURCES FOR MANGROVE MAPPING

Adapted from: Worthington, T.A., Andradi-Brown, D.A., Bhargava, R., Buelow, C., Bunting, P., Duncan, C., Fatoyinbo, L., Friess, D.A., Goldberg, L., Hilarides, L. & Lagomasino, D., 2020. Harnessing big data to support the
conservation and rehabilitation of mangrove forests globally. One Earth, 2(5): 429–443.
57
4.1 EXISTING MANGROVE MAPS data can be viewed and downloaded for mangroves
globally or for individual countries. For each data
4.1.1 Global Mangrove Watch layer, dropdown menus are available to custom-
Global Mangrove Watch (GMW) provides a free
44
ize date range. A Global Mangrove Watch leaflet
online tool (Figure 42) to visualize maps of man- which includes a listing of all available data layers
groves worldwide (Bunting et al., 2022; Spalding in the GMW online platform is available in multiple
and Leal, 2022). The maps were developed from languages at www.wetlands.org/publications/
the baseline of mangrove extent determined from global-mangrove-watch-leaflet.
2010 Landsat imagery with an overall accuracy of As the GMW maps are developed with coarser
95 percent (Bunting et al., 2018) and has since been resolution (25-m) data that are better suited for
updated for 2022 using SAR data (Leal and Spalding, global or regional analysis, there are limitations
eds., 2022). The GMW website has a user-friendly with regards to mapping mangroves at fine scales
interface with an interactive web map to view (Leal and Spalding, eds., 2022). Initial maps orig-
different data layers visualizing mangrove extent, inally published in 2018 had increased error for
net change over time, blue carbon storage and areas with narrow and/or fragmented mangrove
level of protection. In version 3.0, an additional habitat (Bunting et al., 2018). However, the most
map of restoration potential was also developed recent 2022 update, with a reported 95 percent
based on loss of mangroves between 1996 and 2020 accuracy (accuracy is variable by region), has
(Leal and Spalding, eds., 2022). The map includes improved mapping of mangroves in smaller areas
a restorability score that represents the feasibility especially for Small Island Developing States (SIDS)
of restoration and benefits to fisheries and carbon (Leal and Spalding, eds., 2022).
storage (Leal and Spalding, eds., 2022). Reports and
Figure 42. The GMW online platform provides remote sensing data and tools for monitoring mangroves
©TNC
Source: Global Mangrove Alliance. 2023. Global Mangrove Watch. [Cited 1 April 2023].
www.globalmangrovewatch.org
44
www.globalmangrovewatch.org
4.1.2 Ocean Data Viewer including mangroves at 10-m resolution (Figure 43).
The Ocean Data Viewer hosted by the United
45
The maps are updated continuously with versions
Nations Environment Programme and the World now available for 2020 and 2021, using Sentinel-1
Conservation Monitoring Centre provides access and Sentinel-2 data. The 2020 version has an
to a variety of global mangrove datasets including: estimated accuracy of 75 percent (Zanaga et al.,
the World Atlas of Mangroves, version 3.1; Global 2021), while the 2021 version has an accuracy of 76.1
Distribution of Modelled Mangrove Biomass, version 1; percent (Zanaga et al., 2022). Different algorithms
Global Biophysical Topology of Mangroves, version were used to produce the forest cover maps, so
2.2; and the USGS Global Distribution of Mangroves any differences between years are attributable to
version 1.4. difference in land cover and methods used. The
dashboard not only provides options to view the
4.1.3 WorldCover 10-m mangrove class forest cover maps, but also allows users to view
The ESA in collaboration with a consortium of the RGB and false-colour composites from Sentinel
European research organizations, hosts WorldCover data. Data can be accessed and downloaded via the
online, a global map of land cover and forest
46
webmap viewer, Google Earth Engine or Terrascope
types, which includes 11 different land classes with Python.
Figure 43. The ESA WorldCover online portal that provides access to global 10-m data for land cover
and forest types
Source: ESA. 2021. WorldCover. [Cited 1 April 2023].

https://esa-worldcover.org/en
45
https://data.unep-wcmc.org/
46
https://esa-worldcover.org/
ONLINE RESOURCES FOR MANGROVE MAPPING 59

4.2 OPEN-SOURCE IMAGERY DATA, SOFTWARE United States Geological Survey EarthExplorer
AND TOOLS FOR MANGROVE MANAGERS The United States Geological Survey hosts
EarthExplorer,52 an online archive of remotely sensed
4.2.1 Imagery imagery that is freely available to the public. The
Norway’s International Climate & Forests archive includes a wide range of satellite platforms
Initiative – Planet access from multispectral imagery, such as Landsat, and
Through Norway’s International Climate & Forests hyperspectral imagery, to the Hyperion sensor on
Initiative (NICFI), free access can be provided to
47
the EO-1 satellite. Additional platforms include
Planet’s high resolution, analysis-ready mosaics radar, aerial and UAS imagery, as well as additional
of the world’s tropics in order to help reduce and spatial datasets such as digital elevation and
reverse the loss of tropical forests, combat climate vegetation monitoring using NDVI. The archive can
change, conserve biodiversity and facilitate sustain- be searched using a predefined area with a choice of
able development. Planet is a constellation of small platforms and datasets pooled from commercial and
satellites that can image the entire Earth on a daily government entities, such as NASA and the National
basis. Now in its third generation, satellite sensors Oceanic and Atmospheric Administration.
have been updated from 4-band multispectral
imagery to 8-band imagery with the SuperDove 4.2.2 Software
constellation launched in 2020. Planet has a search- Quantum GIS
able database of basemaps and global mosaics that Quantum GIS (QGIS)53 is a free and open-source
began in January 2016 and are available on a monthly geographic information system with a wide range
or quarterly basis. A public application programming of resources, tutorials and integrated toolsets that
interface (API) is available for use with a Python
48
can be used for creating maps as well as analysing
library and the command line interface, and an open- remotely sensed imagery. QGIS can be integrated
source code is available on GitHub.49 For disaster with an online Planet account to allow for access
areas, Planet imagery will make available the entire
50
to Planet’s SkySat, PlanetScope and RapidEye
archive to the public with an explorer account for a imagery, through installation of the Planet plugin.
minimum of up to 30 days. Multiple training courses Instructions on how to integrate QGIS with Planet’s
are available through Planet’s University online image search are available online through Planet
courses that offer guidance on using basemaps
51
University’s online courses.54
and how to integrate image workflows within
various software, such as ArcGIS Pro, Google Earth Integrated Valuation of Ecosystem Services
Engine and QGIS. The NICFI satellite data program and Tradeoffs
was specifically designed for monitoring changes Integrated Valuation of Ecosystem Services and
in tropical forests to reduce deforestation. NICFI Tradeoffs (InVEST)55 is an open-source software
makes monthly imagery available from September that includes a suite of models for mapping eco-
2020 to August 2022, and historical imagery collected system services, such as coastal blue carbon of
every six months from 2015 to 2020. mangrove and seagrass ecosystems. The model
47
https://university.planet.com/page/nicfi
48
https://developers.planet.com/open/
49
https://github.com/planetlabs
50
www.planet.com/disasterdata
51
https://university.planet.com/
52
https://earthexplorer.usgs.gov/
53
www.qgis.org/en/site
54
https://university.planet.com/
55
https://naturalcapitalproject.stanford.edu/software/invest
can be analysed from local, regional to global scales cloud computing-based platform that combines
with input of biophysical or economic information. the Google Earth Engine and open-source software
The software works independently, but results like the Orfeo Toolbox, Python, Jupyter, GDAL, R,
must be viewed in GIS software, such as QGIS or R Studio Server, R Shiny Server, SNAP Toolkit and
ArcGIS. The software targets users with beginner the Open Foris Geospatial Toolkit.
to intermediate GIS skills. The InVEST Coastal Blue
Carbon uses a bookkeeping approach that models 4.2.3 Tools
three pools of carbon: AGB and BGB, dead litter and Orfeo ToolBox
sediment. Detailed breakdown of the parameters The Orfeo ToolBox63 is an open-source software
and algorithms used in the model is hosted on the library developed by the French Space Agency,
InVEST Coastal Blue Carbon model. The user 56
National Center for Space Studies (CNES), for
must provide the following inputs for the model: imagery processing in remote sensing. The library
spatial information (mangrove footprint or extent) is based on C++, and can be accessed with QGIS,
and biophysical information (amount of carbon Python, the command line or Monteverdi. Features
storage and accumulation rate in all three carbon available with the Orfeo Toolbox include preprocess-
pools). The biophysical information should ideally ing imagery, such as radiometric and atmospheric
be sourced from field or local data, but global data corrections, feature extraction, change detection
can be sourced from peer-reviewed literature and and image classification including supervised,
used as a proxy if local data are unavailable. machine-learning algorithms, and object-based
image analysis.
Open Foris
FAO’s Open Foris57 suite of tools is free open-source The Blue Carbon Explorer
software and tools that facilitate flexible and The Nature Conservancy, in collaboration with FAO
efficient data collection, analysis and reporting. and Planet, released the Blue Carbon Explorer in
These include Collect Mobile which is an Android
58
2023 to allow users to explore blue carbon oppor-
app for field-based surveys, Collect Earth,59 a Java- tunities across the insular Caribbean, Indonesia and
based tool that enables data collection through Papua New Guinea (Figure 44). This Google Earth
Google Earth, Collect Earth Online that allows 60
Engine app aims to guide decision-making in terms
users to collect reference data using high resolution of conservation by identifying mangroves that are
satellite images and big data analysis through the degraded and in need of restoration, and mangroves
Google Earth Engine, Earth Map, which permits61
that are healthy and in need of protection, as oppor-
the monitoring of land cover change in an easy, tunities for blue carbon projects. The mangrove
integrated and multitemporal manner, and SEPAL 62
extent map for the Caribbean was developed in
that allows users to query and process satellite 2021 through hand-digitization, while the map for
data quickly and efficiently, tailor their products the Bahamas was refined in 2023 using automated
for local needs, and produce sophisticated and detection from PlanetScope imagery and extensive
relevant geospatial analyses quickly. SEPAL is a field data and input from local experts. These maps
56
http://releases.naturalcapitalproject.org/invest-userguide/latest/en/coastal_blue_carbon.html
57
https://openforis.org/
58
https://openforis.org/tools/collect-mobile/
59
https://openforis.org/tools/collect-earth/
60
https://openforis.org/tools/collect-earth-online/
61
https://earthmap.org/login
62
https://openforis.org/tools/sepal/
63
www.orfeo-toolbox.org
64
https://bluecarbon.tnc.org/

Figure 44. The Blue Carbon Explorer’s example of mangrove restoration prioritization in eastern
Grand Bahama, 2015–2021
©TNC
Note: This analysis based on Planet NICFI Dove imagery and TanDEM-X canopy height data (Simard et al., 2023) was used by
TNC and partners to support site selection for mangrove restoration on the islands of Grand Bahama and Abaco, Bahamas.
are being developed on a country-by-country basis degraded areas of short, non-propagule-producing

for the insular Caribbean and will be added to the mangroves that were remote or isolated from taller
tool over time. For Indonesia and Papua New and healthier propagule-producing mangroves.
Guinea, the GMW mangrove extent is used in the The resulting prioritization was reviewed with
tool. The tool not only shows where mangroves are, restoration partners on both islands, overlaid with
but also gives an estimation of what condition they current restoration sites and marine protected
are in. It allows users to assess changes in health areas (MPAs) to select 10 000 ha of mangroves as
or biomass of mangroves over time using NDVI priorities for restoration.
change maps and time series. It also utilizes MVI
to assess changes in the extent of mangroves over Mapping Ocean Wealth
time. Canopy height from GEDI (Potapov et al., 2020) TNC’s Mapping Ocean Wealth65 platform enables
and TanDEM-X (Simard et al., 2023) can be overlaid, the sharing of ecosystem service data generated
as well as protected areas and benthic habitats. through many collaborative projects. The platform
The Blue Carbon Explorer can also support includes map viewers and data on potential blue
prioritization of mangrove areas for restoration carbon storage, mangrove tourism, and restoration
using NDVI thresholds and canopy height. This sites based on area loss since 1996, using the extent
has been demonstrated by TNC and restoration of mangroves sourced from GMW (Worthington and
partners in the Bahamas. Many of the mangroves Spalding, 2018) (Figure 45). Areas that were per-
on Grand Bahama and Abaco were devastated manently lost due to erosion or urbanization were
during Hurricane Dorian in 2019. With guidance excluded from consideration for restoration. From
from partners on the ground, TNC used the tool to the dropdown menu, restoration potential can be
prioritize mangrove areas for restoration that were viewed globally or by country. The dashboard also
not likely to recover naturally. Such areas comprised has a layer for blue carbon which sums estimates of
65
https://maps.oceanwealth.org/
Figure 45. TNC’s Mapping Ocean Wealth platform includes maps of mangrove restoration potential
and blue carbon storage
©TNC
Source: Worthington, T. & Spalding, M. 2018. Mangrove restoration potential: A global map highlighting a critical opportunity.
Apollo - University of Cambridge Repository. https://doi.org/10.17863/CAM.39153
AGB and soil carbon for mangroves using the GMW access to the Landsat catalogue and will create a
2016 map. The AGB was estimated from a global composite image based on time range of interest
dataset at 30-m resolution for the year 2000, which and cloud cover limit.
included maximum canopy height and individual If higher resolution imagery is needed, the MVI
tree height weighted in proportion to the basal Mapper is a fully automated tool for use in Google
area (Simard, 2019). The below-ground carbon was Earth Engine (Baloloy et al., 2020). MVI is an index
estimated from a machine-learning algorithm using developed to distinguish mangroves from other
mangrove forest extent in 2000, to map soil carbon vegetation types with the use of green, NIR and
at 30 m resolution (Sanderman et al., 2018). SWIR bands.
The band combination helps to distinguish
Google Earth Engine mangroves from other vegetation due to differences
The Google Earth Engine Mangrove Mapping in the level of greenness and moisture content. The
Methodology (GEEMMM) is a free tool developed MVI Mapper in GEE gives access to atmospherically
for use within the Google Earth Engine (GEE) envi- corrected Sentinel-2 (level-2A) imagery. The final
ronment by non-specialists for coastal management outputs include RGB and false colour composites,
(Yancho et al., 2020). The tool is freely available and vegetation and mangrove raster format files.
on GitHub, and utilizes GEE’s cloud computing
66
The MVI Mapper can also be used outside of the
capability to create multidate habitat maps that Google Earth Engine and offline, with Interactive
can be used in change detection. The tool is divided Data Language (IDL®) software, available commer-
into three modules to define the region of interest, cially. However, this method requires additional
classification and accuracy assessment. The tool is steps for sourcing and preprocessing of satellite
also customizable for entering local datasets such imagery before applying the index in IDL.
as mangrove baseline extent DSMs or use of globally
available datasets within GEE. GEEMMM provides
66
https://github.com/Blue-Ventures-Conservation/GEEMMM

©TNC
5. CONCLUSIONS AND
FUTURE DIRECTIONS
As remote sensing technology continues to advance, comprehending the spatial dynamics of mangrove
it is expected that higher spatial and spectral reso- changes, monitoring illegal deforestation activities,
lution data with increased revisit times will become and evaluating the effectiveness of restoration
available. There will also be more integration with initiatives. The methodologies outlined in this
active sensors such as radar and LiDAR to better document play a crucial role in supporting national
define mangrove extent and structure. The rise of endeavors and pledges aimed at conserving,
big data and cloud computing will permit faster sustainably managing, and restoring mangrove
access to and processing of a vast amount of geo- ecosystems along with their diverse ecological and
spatial data that will lead to better understanding socio-economic benefits.
and mapping of mangrove patterns, conditions and Remote sensing will especially play a critical
change. Increased collaboration among mangrove role in serving the needs of SIDS, where information
networks will further promote awareness of the on the status of mangroves is often inadequate, as
need to continue to prioritize, protect and restore their decision-makers seek to better understand
mangroves. the distribution of mangroves and the important
Given that mangrove forests provide critical role they play in human well-being. Progress is still
benefits to people, including food security, shoreline needed in providing data products that can support
stabilization, flood reduction, carbon sequestration improved management at local scales, as well as
and biodiversity conservation, increased attention in improving access to data and ensuring that such
is needed on how we prioritize areas for restoration, access is equitable (Worthington et al., 2020). The
facilitate improved mangrove management and challenge will be to translate the growing body of
enhance ecosystem services provided by mangroves. information into policy and action that will result in
Enhanced remote sensing capabilities at finer scales on-the-ground conservation.
are essential for accurately documenting and
65
©TNC
REFERENCES
Aljahdali, M.O., Munawar, S. & Khan, W.R. 2021. Mexico, based on time series of enhanced
Monitoring mangrove forest degradation and vegetation index (EVI) data. Wetlands Ecology
regeneration: Landsat time series analysis of and Management, 28(2): 229–249.
moisture and vegetation indices at Rabigh Bryan-Brown, D.N., Connolly, R.M., Richards, D.R.,
Lagoon, Red Sea. Forests, 12(1): 52. Adame, F., Friess, D.A. & Brown, C.J. 2020.
Alongi, D.M. 2020. Global significance of mangrove Global trends in mangrove forest fragmentation.
blue carbon in climate change mitigation Scientific Reports, 10: 7117.
(Version 1). Science, 2(3): 57. Bunting, P. Rosenqvist, A., Hilarides, L., Lucas, R.M.,
Aslan, A. & Aljahdali, M.O. 2022. Characterizing Thomas, T., Tadono, T., Worthington, T.A.,
global patterns of mangrove canopy height and Spalding, M., Murray, N.J. & Rebelo, L-M. 2022.
aboveground biomass derived from SRTM data. Global Mangrove Extent Change 1996 – 2020:
Forests, 13(10): 1545. Global Mangrove Watch Version 3.0. Remote
Atwood, T.B., Connolly, R.M., Almahasheer, H., Sensing, 14(15): 3657.
Carnell, P.E., Duarte, C.M., Ewers Lewis, C.J., Bunting, P., Rosenqvist, A., Lucas, R.M., Rebelo, L.M.,
Irigoien, X. et al. 2017. Global patterns in Hilarides, L., Thomas, N., Hardy, A., Itoh, T.,
mangrove soil carbon stocks and losses. Nature Shimada, M. & Finlayson, C.M. 2018. The Global
Climate Change, 7: 523–528. Mangrove Watch - a new 2010 global baseline of
Avery, T.E. 1969. Forester's guide to aerial photo mangrove extent. Remote Sensing, 10(10).
interpretation. No. 308. Washington, DC, US CBD (Convention on Biodiversity). 2022. Official
Department of Agriculture, Forest Service. https:// CBD press release. Convention on Biological
naldc.nal.usda.gov/download/CAT87209514/PDF Diver sit y. 202 2. w w w.cbd.int /ar ticle/
Ayub, A.S., Nugroho, F., Suci, A.N.N. & Anggoro, A. cop15-cbd-press-release-final-19dec2022
2021. Utilization of unmanned aerial vehicle Climate Champions. 2022. The Mangrove
(UAV) technology for mapping mangrove Breakthrough: A call to action for a critical
ecosystem. Journal of Sylva Indonesiana, 4(02): ecosystem. Climate Champions.2022. https://
70–77. climatechampions.unfccc.int/the-mangrove-
Baloloy, A.B., Blanco, A.C., Ana, R.R.C.S. & breakthrough
Nadaoka, K. 2020. Development and application Donato, D.C., Kauffman, J.B., Murdiyarso, D.,
of a new mangrove vegetation index (MVI) for Kurnianto, S., Stidham, M. & Kanninen, M. 2011.
rapid and accurate mangrove mapping. ISPRS Mangroves among the most carbon-rich forests
Journal of Photogrammetry and Remote Sensing, in the tropics. Nature Geoscience, 4(5): 293–7.
166: 95–117 FAO. 2007. The world´s mangroves 1980-2005.
Berlanga-Robles, C.A. & Ruiz-Luna, A. 2020. FAO Forestr y Paper 153. Rome, FAO.
Assessing seasonal and long-term mangrove https://agris.fao.org/agris-search/search.
canopy variations in Sinaloa, northwest do?recordID=XF2008433305
67
FAO. 2020a. FAO Strategy on Mainstreaming Heenkenda, M.K., Joyce, K.E., Maier, S.W.
Biodiversity across Agricultural Sectors. Rome, & Bar tolo, R. 2014. Mangrove species
FAO. doi.org/10.4060/ca7722en identification: Comparing WorldView-2 with
FAO. 2020b. Global Forest Resources Assessment. aerial photographs. Remote Sensing, 6(7):
Rome. [Cited 25 January 2023]. https://fra-data. 6064–6088.
fao.org. Hutchison, J., Manica, A ., Swetnam, R.,
FAO. 2021. 2021-23 Action plan for the implementation Balmford, A. & Spalding, M. 2014. Predicting
of the FAO strategy on mainstreaming biodiversity global patterns in mangrove forest biomass.
across agricultural sectors. Rome, FAO. www.fao. Conservation Letters, 7: 233–240.
org/3/cb5515en/cb5515en.pdf Jardine, S.L. & Siikamäki, J.V. 2014. A global
FAO. 2023. The world’s mangroves 2000–2020. Rome. predictive model of carbon in mangrove soils.
https://doi.org/10.4060/cc7044en Environmental Research Letters, 9: 104013.
Faridah-Hanum, I., Yusoff, F.M., Fitrianto, A., Jensen, J.R. 1996. Introductory digital image
Ainuddin, N.A., Gandaseca, S., Zaiton, S., processing: a remote sensing perspective.
Norizah, K. et al. 2019. Development of a Second edition. Prentice-Hall Inc. www.scirp.
comprehensive mangrove quality index (MQI) org/(S(351jmbntvnsjt1aadkposzje))/reference/
in Matang Mangrove: Assessing mangrove ReferencesPapers.aspx?ReferenceID=2028387
ecosystem health. Ecological Indicators, 102: Jiang, Y., Zhang, L., Yan, M., Qi, J., Fu, T., Fan, S.
103–117. & Chen, B. 2021. High-resolution mangrove
Friess, D.A., Rogers, K., Lovelock, C.E., Krauss, K.W., forests classification with machine learning
Hamilton, S.E., Lee, S.Y., Lucas, R. et al. 2019. using worldview and UAV hyperspectral data.
The state of the world's mangrove forests: Remote Sensing, 13(8): 1529.
past, present, and future. Annual Review of Joyce, K.E., Fickas, K.C. & Kalamandeen, M. 2023.
Environment and Resources, 44: 89–115. The unique value proposition for using drones
Giri, C., Ochieng, E., Tieszen, L.L., Zhu, Z., to map coastal ecosystems. Cambridge Prisms:
Singh, A., Loveland, T., Ma-sek, J. & Duke, N. Coastal Futures, 1: p.e6.
2011. Status and distribution of mangrove Klemas, V.V. 2015. Coastal and environmental
forests of the world using earth observation remote sensing from unmanned aerial vehicles:
satellite data. Global Ecology and Biogeography, An overview. Journal of Coastal Research, 31(5):
20: 154–159. 1260–1267.
Goldberg, L., Lagomasino, D., Thomas, N. & Leal, M. & Spalding, M.D., eds. 2022. The State of
Fatoyinbo, T. 2020. Global declines in human‐ the World’s Mangroves 2022. Global Mangrove
driven mangrove loss. Global Change Biology, Alliance. www.mangrovealliance.org/wp-
26(10): 5844–5855. content/uploads/2022/09/The-State-of-the-
Hamilton, S.E. & Casey, D. 2016. Creation of a high Worlds-Mangroves-Report_2022.pdf
spatio-temporal resolution global database Losada, I. J., Menéndez, P., Espejo, A .,
of continuous mangrove forest cover for the Torres, S., Díaz-Simal, P., Abad, S., Beck, M.W.
21st century (CGMFC-21). Global Ecology and et al. 2018. The global value of mangroves for
Biogeography, 25: 729–738. risk reduction. Technical Report. The Nature
Hamilton, S.E. & Friess, D.A. 2018. Global carbon Conservancy. www.conservationgateway.
stocks and potential emissions due to mangrove org /Conser vationPrac tices/Marine/crr/
deforestation from 2000 to 2012. Nature Climate l i b r a r y/ D o c u m e n t s /G l o b a l M a n g r o v e s
Change, 8: 240–244. RiskReductionTechnicalRepor t10.7291/
V9DV1H2S.pdf
Lucas, R.M., Ellison, J.C., Mitchell, A., Donnelly, B., Potapov, P., Li, X., Hernandez-Serna, A.,
Finlayson, M. & Milne, A.K. 2002. Use of stereo Tyukavina, A., Hansen, M.C., Kommareddy, A.,
aerial photography for quantifying changes in Pickens, A. et al. 2020. Mapping and monitoring
the extent and height of mangroves in tropical global forest canopy height through integration
Australia. Wetlands Ecology and Management, of GEDI and Landsat data. Remote Sensing of
10(2): 159–173. Environment, 112165.
Maina, J.M., Bosire, J.O., Kairo, J.G., Bandeira, S.O., Rossi, R.E., Archer, S.K., Giri, C. & Layman, C.A.
Mangora, M.M., Macamo, C., Ralison, H. & 2020. The role of multiple stressors in a dwarf
Majambo, G. 2021. Identifying global and local red mangrove (Rhizophora mangle) dieback.
drivers of change in mangrove cover and the Estuarine, Coastal and Shelf Science, 237: 106660.
implications for management. Global Ecology Rovai, A.S., Twilley, R.R., Castan ̃ eda-Moya, E.,
and Biogeography, 30(10): 2057–2069. Riul, P., Cifuentes-Jara, M., Manrow-Villalobos,
Maynard, J., Tracey, D., Williams, G.J., Andradi- M., Horta, P.A., Simonassi, J.C. et al. 2018.
Brown, D.A., Grill, G., Thieme, M. & Ahmadia, G. Global controls on carbon storage in mangrove
2019. Mangrove cover change between 1996 and soils. Nature Climate Change, 8: 534–538.
2016 near river-ocean outlets: A global analysis Ruzgienė, B., Bagdžiūnaitė, R. & Ruginytė, V. 2012.
to identify priority rivers for conservation. Gland, Scanning aerial photos using a non-professional
Switzerland, World Wildlife Fund. https://doi. scanner. Geodesy and Cartography, 38(3):
org/10.6084/m9.figshare.8094245. 118–121.
Menéndez, P., Losada, I.J., Torres-Ortega, S., Sanderman, J., Hengl, T., Fiske, G., Solvik, K.,
Narayan, S. & Beck, M.W. 2020. The global flood Adame, M.F., Benson, L., Bukoski, J.J. et al.
protection benefits of mangroves. Scientific 2018. A global map of mangrove forest soil
Reports, 10: 4404. carbon at 30 m spatial resolution. Environmental
Navarro, A., Young, M., Allan, B., Carnell, P., Research Letters, 13(5).
Macreadie, P. & Ierodiaconou, D. 2020. The Sievers, M., Brown, C.J., Tulloch, V.J.D., Pearson, R.M.,
application of unmanned aerial vehicles (UAVs) Haig, J.A., Turschwell, M.P. & Connolly, R.M.
to estimate above-ground biomass of mangrove 2019. The role of vegetated coastal wetlands
ecosystems. Remote Sensing of Environment, for marine megafauna conservation. Trends in
242: 111747. Ecology and Evolution, 34: 807–817.
Pastor-Guzman, J., Dash, J. & Atkinson, P.M. 2018. Siikamäki, J., Sanchirico, J.N. & Jardine, S.L.
Remote sensing of mangrove forest phenology 2012. Global economic potential for reducing
and its environmental drivers. Remote Sensing carbon dioxide emissions from mangrove loss.
of Environment, 205: 71–84. Proceedings of the National Academy of Sciences
Pham, T.D., Xia, J., Ha, N.T., Bui, D.T., Le, N.N. of the United States of America, 109: 14369–14374.
& Tekeuchi, W. 2019a. A review of remote Simard, M., Fatoyinbo, T., Thomas, N., Stovall, A.,
sensing approaches for monitoring blue carbon Parra, A., Barenblitt, A., Bunting, P. et al. 2023.
ecosystems: Mangroves, seagrasses and salt A New Global Mangrove Height Map. Manuscript
marshes during 2010–2018. Sensors, 19(8): 1933. in prepreparation.
Polidoro, B.A., Carpenter, K.E., Collins, L., Simard, M. 2019. Radar remote sensing of mangrove
Duke, N.C., Ellison, A.M., Ellison, J.C., forests. In: A. Flores, K. Herndon, R. Thapa &
Farnsworth, E.J. et al. 2010. The loss of species: E. Cherrington, eds. NASA SAR Handbook:
mangrove extinction risk and geographic areas Comprehensive methodologies for forest
of global concern. PLoS ONE, 5. monitoring and biomass estimation. NASA 2019.
DOI: 10.25966/33zm-x271
REFERENCES 69
REFERENCES
Simard, M., Fatoyinbo, T., Smetanka, C., Ward, R.D., Friess, D.A., Day, R.H. & Mackenzie, R.A.
Rivera-monroy, V.H., Castaneda, E., Thomas, N. 2016. Impacts of climate change on mangrove
& Van der stocken, T. 2019. Global mangrove ecosystems: a region by region overview.
distribution, aboveground biomass, and canopy Ecosystem Health and Sustainability, 2(4): e01211.
height. Oak Ridge, Tennessee, USA, ORNL DAAC. Worthington, T. & Spalding, M. 2018. Mangrove
doi.org/10.3334/ORNLDAAC/1665 restoration potential: A global map highlighting
Spalding, M. & Parrett, C.L. 2019. Global patterns a critical opportunity. Apollo - University of
in mangrove recreation and tourism. Marine Cambridge Repository. https://doi.org/10.17863/
Policy, 110: 103540. CAM.39153
Spalding, M.D., Kainumu, M. & Collins, L. 2010. WRI (World Resources Institute) & IIED
World Atlas of Mangroves. UK, Earthscan. (International Institute for Environment and
https://books.google.co.th/books/about/ Development). 1986. World Resources 1986.
World_ Atlas _of_ Mangroves.html?id= Pollutants and forest decline. Washington,
wzSCkulW9SQC&redir_esc=y DC, WRI and London, UK, IIED. www.wri.org/
Tang, W., Zheng, M., Zhao, X., Shi, J., Yang, J. & research/world-resources-1986.
Trettin, C. 2018. Big geospatial data analytics Yancho, J.M.M., Jones, T.G., Gandhi, S.R.,
for global mangrove biomass and carbon Ferster, C., Lin, A. & Glass, L. 2020. The Google
estimation. Sustainability, 10: 472. Earth Engine Mangrove Mapping Methodology
Taureau, F., Robin, M., Proisy, C., Fromard, F., (GEEMMM). Remote Sensing, 12(22): 1–35.
Imbert, D. & Debaine, F. 2019. Mapping the Zanaga, D., Van De Kerchove, R., De Keersmaecker, W.,
mangrove forest canopy using spectral unmixing Souverijns, N., Brockmann, C., Quast, R.,
of very high spatial resolution satellite images. Wevers, J. et al. 2021. ESA WorldCover 10 m 2020
Remote Sensing, 11(3): 367. v100. Paris, France. European Space Agency.
Tian, Y., Huang, H., Zhou, G., Zhang, Q., Tao, J., https://doi.org/10.5281/zenodo.5571936
Zhang, Y. & Lin, J. 2021. Aboveground mangrove Zanaga, D., Van De Kerchove, R., Daems, D.,
biomass estimation in Beibu Gulf using machine De Keersmaecker, W., Brockmann, C., Kirches,
learning and UAV remote sensing. Science of the G., Wevers, J. et al. 2022. Paris, France, European
Total Environment, 781: 146816. Space Agency. ESA WorldCover 10 m 2021 v200.
Twilley, R.R., Chen, R.H. & Hargis, T. 1992. Carbon doi.org/10.5281/zenodo.7254221
sinks in mangroves and their implications to Zimudzi, E., Sanders, I., Rollings, N. & Omlin, C.W.
carbon budget of tropical coastal eco-systems. 2021. Remote sensing of mangroves using
Water, Air, & Soil Pollution, 64: 265–288. unmanned aerial vehicles: Current state and
Waldron, A., Adams, V., Allan, J., Arnell, A., future directions. Journal of Spatial Science,
Asner, G., Atkinson, S., Baccini, A. et al. 2020. 66(2): 195–212.
Protecting 30% of the planet for nature: costs, Zu Ermgassen, P.S., Mukherjee, N., Worthington, T.A.,
benefits and economic implications. University Acosta, A., da Rocha Araujo, A.R., Beitl, C.M.,
of Helsinki. doi:10.13140/RG.2.2.19950.64327 Castellanos-Galindo, G.A. et al. 2020. Fishers
Waltham, N.J., Elliott, M., Lee, S.Y., Lovelock, C., who rely on mangroves: Modelling and mapping
Duarte, C.M., Buelow, C., Simenstad, C. et al. the global intensity of mangrove-associated
2020. UN decade on ecosystem restoration 2021– fisheries. Estuarine, Coastal and Shelf Science,
2030—what chance for success in restoring coastal 247: 106975.
ecosystems? Frontiers in Marine Science: 71.
APPENDIX
Case studies in mapping
mangrove change
Four case studies that demonstrate the use of remote production. The need for monitoring mangroves is
sensing for detecting mangrove change at fine scales not only to understand where mangroves are being
are presented. High resolution satellite imagery lost, but also where mangroves are expanding. The
and UAS data were used to map and monitor man- objectives of these projects were to establish a base-
grove changes in the Dominican Republic, Grenada, line for gauging future change and understanding
Haiti and Saint Vincent and the Grenadines. These the threats so that adaptive management plans
countries face similar, as well as different, threats to can be designed and put in place to safeguard the
mangroves, ranging from aquaculture and coastal mangroves.
development, to harvesting of biomass for charcoal
CASE STUDY 1 the objects across spatial, spectral and textural

PARQUE NACIONAL MANGLARES DEL scales. These segments represent distinct patches
BAJO YUNA, THE DOMINICAN REPUBLIC of uniform mangrove habitat. Object-oriented
The Parque Nacional Manglares Del Bajo Yuna is methods yield accuracy improvements compared to
located in a karst estuarine wetland ecosystem conventional pixel-based image analysis techniques
at the western end of Samaná Bay, encompassing as the non-spectral attributes of the imagery such
an area of 12 000 ha. Declared a Ramsar site, the as texture, spatial and contextual information are
park is one of the largest semi-closed bays in the integrated into the classification workflow. The soft-
Caribbean and supports various endemic species. ware used for mapping in this study was eCognition
The total mangrove extent measured using satellite (v. 9.1, Trimble Inc.). Image segmentation requires
imagery acquired in December 2017 was 3 456 ha. identifying the appropriate scale of segments that
Field surveys were conducted in 2016 and 2017 can adequately represent the features that are being
collecting, inter alia, the following field parameters: mapped. After segmentation, statistics pertaining
1) mangrove species composition; 2) canopy height; to the spectral and textural properties of the sat-
3) DBH; and 4) species richness and percent cover ellite imagery were used to identify all mangrove
(by species). segments. Mangrove segments that were adjacent
Quickbird imagery was used to map the man- to each other were merged; the results were visually
grove extent in 2003 and WorldView-2 imagery interpreted and manually cleaned to improve the
was used for 2013 and WorldView-3 for 2017. An final accuracy for both time periods.
object-based image analysis approach was applied WorldView-3 satellite imagery of Parque
to delineate mangrove extent. This approach seg- Nacional Manglares Del Bajo Yuna collected on 6 July
ments satellite imagery into landscape objects that 2017, was used to calculate NDVI values and compare
have ecologically meaningful shapes, and classifies them with the baseline dataset of 3 December 2013
71
©TNC
both in terms of mangrove extent and biomass. detected in the north in areas that had been cleared
A pixel-by-pixel NDVI comparison between the for aquaculture but were later abandoned. The
2013 and 2017 NDVI datasets was performed and results in Figure A1.1 show the area extent for
a difference threshold that represented a varying each magnitude of change in biomass based on
amount of change between the two NDVI images the difference threshold.
was computed. The five threshold classes above Changes in mangrove cover since 2003 were also
and below the no change class represent the assessed. In the north, near the town of Sanchez,
magnitude of biomass change. The thresholds are a total of 233 ha of mangroves had been removed
based on standard deviations from the baseline within the national park boundary. Records show
map. Change (+5) represents the largest areas that that this was largely due to the expansion of shrimp
have gained mangrove biomass, while Change (-5) aquaculture and polluted water generated by the
indicates the largest loss of mangrove biomass. The municipality’s landfill. However, mangrove cover has
classes in between are scaled respectively. increased by 206 ha since 2001 around the mouth
Results of the NDVI analysis indicated that of the Barracote River. This is a result of sediment
biomass decreased in 96.32 ha of mangroves deposition that has decreased water depth and
between 2013 and 2017 while biomass increased facilitated the natural establishment of mangrove
in 160.27 ha of mangroves particularly near the propagules across a wider area. Figure A1.2 and
mouth of the Barracote River to the south where Figure A1.3 shows that this area has high mangrove
mangroves had expanded. Growth was also biomass indicating a highly productive system.
FIGURE A1.1 NDVI class magnitude of change in hectares
Change (+5) 38.80
Change (+4) 42.79
Change (+3) 9.23
Change (+2) 24.12
Change (+1) 45.32
Change (-1) 38.57
Change ( -2) 22.45
Change (-3) 23.42
Change (-4) 8.80
Change (-5) 3.01
 Positive change  Negative change

Source: TNC
FIGURE A1.2 Mangrove change detection (2003 FIGURE A1.3 Areas of change in mangrove
to 2017) in Parque Nacional Manglares Del Bajo biomass based on NDVI values in Parque
Yuna, Dominican Republic, showing areas of Nacional Manglares Del Bajo Yuna between
mangrove loss and growth since 2003 using December 2013 and July 2017
Quickbird and WorldView-2 satellite imagery
©TNC
©TNC
APPENDIX 73
CASE STUDY 2 images equalled 4 146.27 ha. Using WorldView-2
CARACOL AND FORT-LIBERTÉ, 2014 imagery, a total of 4 257.42 ha for a net increase
THREE BAYS NATIONAL PARK, HAITI of 111.15 ha was identified for the entire park. Total
Three Bays National Park (3BNP) was designated loss of mangrove areas detected from 2003 to 2014
by the Haitian Government in 2013. The MPA was estimated to be 91.48 ha, mostly due to harvest-
encompasses three bays: Limonade, Caracol and ing of mangrove wood for charcoal production. On
Fort-Liberté, as well as one of the largest inland the other hand, there was an expansion of 202.63 ha
brackishwater lagoons – the Important Bird Area of of mangroves into new areas as a result of ongoing
Lagon aux Boeufs – covering an area of 75 618 ha. reforestation initiatives that engaged communities
3BNP represents one of the largest protected areas located within the park to sow mangrove propagules
of mangrove and coastal wetlands in Haiti and has a in areas of die-off. These activities were also sup-
total mangrove area of 4 257.42 ha as measured from ported by the construction of mangrove nurseries
satellite imagery acquired in April 2014, representing and awareness-raising on the importance of man-
about 20 percent of Haiti’s remaining mangroves. groves, encouraging environmental stewardship.
Field surveys were carried out in June and Results of NDVI analysis indicated that a total of
November 2015 as well as July 2017 to document 57.47 ha of mangroves had been lost while a total
existing conditions within the mangrove and coastal of 101.1 ha of mangroves had established between
wetland habitats of 3BNP. The total mangrove area 2014 and 2017 (Figures A2.1, A2.2 and A2.3).
mapped within 3BNP using Quickbird 2002/2003
©TNC
FIGURE A2.1 Mangrove loss and expansion (ha) between 2014 and 2017 in 3BNP
200
150
100
50
0
3 BNP Caracol Fort Liberté Lagon aux Boeufs
 Growth  Loss
Source: TNC
FIGURE A2.2 NDVI biomass change detection in Caracol Bay from 2014 to 2017
©TNC
FIGURE A2.3 NDVI biomass change detection in Forte Liberté Bay and Lagon aux Bouefs from 2014 to 2017
©TNC
APPENDIX 75
CASE STUDY 3
ASHTON LAGOON, UNION ISLAND,
SAINT VINCENT AND THE GRENADINES Figure A3.1 NDVI difference between 2009
Ashton Lagoon is the largest lagoon in Saint Vincent and 2012
and the Grenadines and home to the largest man-
grove forest in the country. The lagoon harbours
abundant seagrass beds, fringing, patch and
barrier coral reefs, as well as an offshore island
(Frigate) and mangrove wetlands that have been
designated as an Important Bird Area by Birdlife
International. Because of its rich ecological
importance to the country, this area was formally
designated a Conservation Area (under Schedule 11,
Regulation 20, The Fisheries Act, 1986) in 1987.
©TNC
The total mangrove extent measured from drone
imagery (acquired in December 2015) was 19.81 ha.
Field surveys were conducted in December 2015
and March 2016 to document the extent and density
of mangroves. IKONOS imagery acquired in March
2009 and WorldView-2 imagery acquired in June Notes: Changes in mangrove extent over three years, from
2009 to 2012. The dark blue and purple shades indicate
2012 revealed a total of 18.09 ha and 19.81 ha of mangrove die-off, while green, yellow and brown indicate
mangrove growth. Note signs of die-off in the centre of the
mangroves, respectively. One major threat that forest, largely due to an alteration of hydrology when an
access road was built to the marina along the edge of the
negatively impacted the mangroves was a 300-berth mangrove. Expansion of mangroves was detected along
the causeways and to the north.
marina project that started in 1994 and became
Figure A3.2 Reduction and increase in mangrove biomass as modeled by remotely sensed NDVI values
in Ashton Lagoon between 2009-2012
©TNC
©TNC
Left: Areas where mangrove biomass was reduced in Ashton Lagoon between 2009 and 2012 (3.01 ha).
Right: Areas where mangrove biomass increased over the same period (1.1 ha).
non-operational a year after. Nonetheless, massive die-off can be seen in the central part of the wetland,
dredging and causeway construction had already mangrove growth continues along the abandoned
occurred and modified the natural hydrology in causeways and there is a slow expansion to the
the bay, reducing nutrient inflow and stunting north of the wetland.
mangrove growth. Although a steady mangrove
Figure A3.3 Aerial UAS image of Ashton Lagoon, Saint Vincent and the Grenadines taken from a DJI
Phantom 4 Pro at an altitude of 120 m on 24 June 2017
©TNC
Figure A3.4 A colour infrared orthophoto mosaic, the resulting DSM and computed NDVI values of
Ashton Lagoon, Saint Vincent and the Grenadines
©TNC
©TNC
©TNC
Left: A colour infrared orthophoto mosaic acquired by a Sony QX1 camera from a 3DR Solo drone on 24 June 2017 at a flying
height of 120 m which yielded a pixel resolution of 2 cm. Centre: The resulting DSM. Right: The computed NDVI values which
indicate mangrove biomass
APPENDIX 77
CASE STUDY 4 imagery acquired on 12 November 2014; and UAS
TYRREL BAY, CARRIACOU, GRENADA imagery acquired from 122 m (2-cm resolution)
Tyrrel Bay is part of the Sandy Island Oyster Bed on 30 November 2015. The main threat to this
Marine Protected Area established in 2009 and mangrove area is the continued construction of a
comprises an area of 787 ha on the southwest coast marina in Tyrrel Bay which began in 2003. Dredging
of Carriacou, Grenada. The mangroves in Tyrrel Bay to accommodate yachts has destroyed seagrass
provide habitat for mangrove oysters that grow on beds, reduced water quality and continues to
the roots of the red mangroves. These mangroves negatively impact the mangrove ecosystem.
also serve as nursery grounds for several species Assessment results show that 13.8 percent of the
of fish and are used by local boats to secure their mangroves were removed through clear-cutting
vessels during tropical storms. The total mangrove and back-filling. In the north-eastern part of the
extent measured from satellite imagery acquired in mangrove forest, dwarf mangroves have slowly
November 2014 was 26.48 ha. expanded into the mud flats with a total expansion
To calculate the mangrove spatial extent and of 2.57 ha. These dwarf mangroves have restricted
biomass in Tyrrel Bay, the following datasets growth and low biomass due to their locations at
were used: IKONOS satellite imagery acquired on higher elevations with minimal tidal flushing and
2 November 2000; WorldView-2 satellite imagery therefore limited nutrient inputs.
acquired on 10 January 2010; WorldView-2 satellite
Figure A4.1 The extent and change detection of mangroves in Tyrrell Bay, Carriacou, Grenada
©TNC
©TNC
©TNC
Left: WorldView-2 satellite imagery acquired on 10 January 2010. Centre: WorldView-2 image acquired on 12 November 2014.
Right: The mangrove change detection map.
Figure A4.2 Aerial view of Tyrrell Bay, Carriacou, Grenada

©TNC
FAO Forestry Division –
Natural Resources and Sustainable Production ISBN 978-92-5-138796-2
https://www.fao.org/forestry/en/

Rome, Italy 9 789251 387962
CD0823EN/1/07.24

MangroveGuide Web

Uploaded by

Copyright:

Available Formats

MangroveGuide Web

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

MangroveGuide Web

Uploaded by

Copyright:

Available Formats

Remote sensing techniques

for mapping and monitoring

Steve Schill, Valerie McNulty and Denise Perez

Kenichi Shono and Kim Friedman

Food and Agriculture Organization of the United Nations

Cover photograph: Mangrove forest in Tyrell Bay, Grenada. ©TNC

2. REMOTE SENSING DATA CONSIDERATIONS............................................................................ 9

3. MANGROVE MAPPING TECHNIQUES....................................................................................... 45

4. ONLINE RESOURCES FOR MANGROVE MAPPING................................................................ 55

5. CONCLUSIONS AND FUTURE DIRECTIONS............................................................................ 65

3BNP Three Bays National Park, Haiti

Overwash Found on small and narrow elevated islands that may

Basin Grow in partially impounded depressions that are

Scrub Scrub (dwarf) mangroves are more common in broad

Figure 3. Examples of different mangrove growth patterns on Andros Island, Bahamas

©2022 PLANET LABS PBC

©2022 PLANET LABS PBC

REMOTE SENSING DATA CONSIDER ATIONS 11

REMOTE SENSING DATA CONSIDER ATIONS 13

©2022 PLANET LABS PBC

©2022 PLANET LABS PBC

©2022 PLANET LABS PBC

REMOTE SENSING DATA CONSIDER ATIONS 15

REMOTE SENSING DATA CONSIDER ATIONS 17

©U.S. GEOLOGICAL SURVEY

©U.S. GEOLOGICAL SURVEY

REMOTE SENSING DATA CONSIDER ATIONS 19

©U.S. GEOLOGICAL SURVEY

©U.S. GEOLOGICAL SURVEY

©TNC/U.S. GEOLOGICAL SURVEY

©TNC/U.S. GEOLOGICAL SURVEY

©TNC/U.S. GEOLOGICAL SURVEY

0 0.25 0.5 1 Kilometers

©TNC/U.S. GEOLOGICAL SURVEY

© 2022 PLANET LABS PBC

© 2022 PLANET LABS PBC

© 2022 PLANET LABS PBC

REMOTE SENSING DATA CONSIDER ATIONS 21

REMOTE SENSING DATA CONSIDER ATIONS 23

©2022 MAXAR, ©2022 PLANET LABS PBC

©2022 MAXAR, ©2022 PLANET LABS PBC

REMOTE SENSING DATA CONSIDER ATIONS 25

REMOTE SENSING DATA CONSIDER ATIONS 27

Table 4. Advantages and disadvantages of mapping mangroves using a UAS

REMOTE SENSING DATA CONSIDER ATIONS 29

Table 5. Advantages of multirotor vs. fixed-wing UAS

REMOTE SENSING DATA CONSIDER ATIONS 31

travelling at similar speeds, mapping at the higher

embedded in the file header. The accuracy of the

REMOTE SENSING DATA CONSIDER ATIONS 33

REMOTE SENSING DATA CONSIDER ATIONS 35

REMOTE SENSING DATA CONSIDER ATIONS 37

REMOTE SENSING DATA CONSIDER ATIONS 39

REMOTE SENSING DATA CONSIDER ATIONS 41

REMOTE SENSING DATA CONSIDER ATIONS 43

Figure 35. Example of mangrove digitization in QGIS with multiple basemaps

MANGROVE MAPPING TECHNIQUES 47

Table 11. Common vegetation indices and their corresponding formulas

MANGROVE MAPPING TECHNIQUES 49

MANGROVE MAPPING TECHNIQUES 51