National Aeronautics and Space Administration

Monitoring Water Quality Using Satellite Image

Processing
Amita Mehta & Africa Flores (SERVIR Global)
5, 12, and 19 September 2018
Training Objectives

Learn to:
• Monitor water temperature and chlorophyll-a concentrations as harmful algal
bloom indicators
• Access MODIS and Landsat data for water quality monitoring
• Perform image processing of MODIS and Landsat data using SeaDAS Software

Prerequisite: Introduction to Remote Sensing of Harmful Algal Blooms

https://arset.gsfc.nasa.gov/water/webinars/HABs17

NASA’s Applied Remote Sensing Training Program 2

Course Outline

September 5 September 12 September 19

Overview and Analysis of Introduction to SeaDAS for Image Analysis Exercise
NASA Remote Sensing Image Processing and Using SeaDAS
Data for HAB Monitoring Data Analysis

NASA’s Applied Remote Sensing Training Program 3

Outline for Week 1

• About ARSET
• Remote Sensing of Water Quality (WQ)
• Monitoring WQ in Coastal and In-land Waters
• Demonstration of NASA Web-tools for WQ Data Access
– Focus: Chlorophyll a Concentration (Chlor_a) and Water Surface Temperature
(ST) Case Study: Chesapeake Bay, 1-10 May 2018
– Giovanni: http://giovanni.gsfc.nasa.gov/giovanni/
– OceanColor: https://oceancolor.gsfc.nasa.gov/
• Exercise: Analyze and Download Chlor_a and ST in Lake Victoria, Africa

NASA’s Applied Remote Sensing Training Program 4

About ARSET
NASA’s Applied Remote Sensing Training Program (ARSET)
http://arset.gsfc.nasa.gov/
• Empowering the global community Topics for Trainings Include:
through remote sensing training
• Part of NASA’s Applied Sciences
Water Resources
Program
• Seeks to increase the use of Earth
science in decision-making through
training for:
– policy makers
– environmental managers
– other professionals in the public and Air Quality
private sector Disasters

Eco

NASA’s Applied Remote Sensing Training Program 6

ARSET Team Members

Program Support Disasters & Water Resources

• Ana Prados, Program Manager (GSFC) • Amita Mehta, Instructor (GSFC)
• Brock Blevins, Training Coordinator (GSFC) • Erika Podest, Instructor (JPL)
• David Barbado, Spanish Translator (GSFC) Land & Wildfires
• Annelise Carleton-Hug, Program Evaluator • Cynthia Schmidt, Lead (ARC)
(Consultant) • Amber Jean McCullum, Instructor (ARC)
• Elizabeth Hook, Technical Writer/Editor Health & Air Quality
(GSFC)
• Pawan Gupta, Lead (GSFC)
• Selwyn Hudson-Odoi, Training Coordinator
(GSFC) • Melanie Cook, Instructor (GSFC)

• Marines Martins, Project Support (GSFC)

Acknowledgement:
• Stephanie Uz, Program Support (GSFC)
• We wish to thank Nancy Searby for her
continued support

NASA’s Applied Remote Sensing Training Program 7

ARSET Trainings
Trainings: Introductory and Advanced, On-line and In-person

100 trainings 13,000+ participants 160+ countries 3,700+ organizations

* size of bubble corresponds to number of attendees

NASA’s Applied Remote Sensing Training Program 8

ARSET Water Quality Trainings
https://arset.gsfc.nasa.gov/water/
• Introduction to Remote Sensing of Harmful Algal Blooms:
– https://arset.gsfc.nasa.gov/water/webinars/HABs17

• Introduction to Remote Sensing for Coastal & Ocean Applications:

– https://arset.gsfc.nasa.gov/land/webinars/coastal-oceans-2016

• Water Quality Monitoring Using Remote Sensing Measurements:

– https://arset.gsfc.nasa.gov/water/water-quality-2014

NASA’s Applied Remote Sensing Training Program 9

Learn More About ARSET
http://arset.gsfc.nasa.gov/

NASA’s Applied Remote Sensing Training Program 10

Remote Sensing of Water Quality (WQ)
Remote Sensing of Water Quality

• Satellites carry
instruments and sensors
to measure:
– reflected solar
radiation
– emitted infrared and
microwave radiation

NASA’s Applied Remote Sensing Training Program 12

Water Quality Affects Water Optical Properties

You can use remote sensing to monitor water color, which

can be an indicator of water quality:
• dissolved and suspended matter in water change the
water’s optical property, which changes its color
– dissolved organic matter includes tannin, which is caused
by organic matter coming from leaves, roots, and plant
remains
– suspended matter includes particles of clay, undissolved
minerals, planktons, and algal blooms
• some harmful algal bloom (HAB) species have unique
properties that affect water color (e.g., red tides)

Image Credit: USGS

NASA’s Applied Remote Sensing Training Program 13

Which Factors Cause Change in Water Quality?

• Nutrient loading “eutrophication”

• Pollution
• Water Temperature (warmer water affects Algal Bloom)
• Food web changes
• Introduced species
• Changes in water flow
– e.g., after major events like hurricanes, drought, or floods

NASA’s Applied Remote Sensing Training Program 14

Current Satellite Missions for Water Quality Monitoring

• Landsat 7 (4/15/1999 – present)

• Landsat 8 (2/1/2013 – present)
• Terra (12/18/1999 – present)
• Aqua (5/4/2002 – present)
• Suomi National Polar Partnership
(SNPP) (11/21/2011 – present)
• Sentinel-2A (6/23/2015 - present)
• Sentinel-2B (3/7/2017 – present)
• Sentinel-3A (2/16/2016 – present)

NASA’s Applied Remote Sensing Training Program 15

Satellites and Sensors for Monitoring Water Quality

Satellites Sensors Resolution

185 km swath; 15 m, 30 m, 60 m;
Landsat 7 Enhanced Thematic Mapper (ETM+)
16 day revisit
185 km swath; 15 m, 30 m, 60 m;
Landsat 8 Operational Land Imager (OLI)
16 day revisit
MODerate Resolution Imaging 2330 km swath; 250 m, 500 m, 1 km;
Terra & Aqua
Spectroradiometer (MODIS) 1-2 day revisit
Visible Infrared Imaging Radiometer 3040 km swath; 375 m – 750 m;
Suomi NPP
Suite (VIIRS) 1-2 day revisit
Sentinel 2A 290 km swath; 10 m, 20 m, 60 m;
Multi Spectral Imager (MSI)
and 2B 5 day revisit
Ocean and Land Color Instrument
Sentinel 3A 1270 km swath; 300 m; 27 day revisit
(OLCI)
* See Appendix A for information about the spectral bands

NASA’s Applied Remote Sensing Training Program 16

Advantages of Satellite Observations

• Available for large regions Limited Water Sampling Locations

– only source of global information for some
parameters
• Long time series and data continuity
– tracks progress
– establishes baselines and trends
• Consistency and comparability
– among multiple countries
• Diversity of measurements
– many different physical parameters
• Complements traditional statistical methods
– cross-check with in situ data
• Mostly free and open access
Image Credit (top) http://data.gcoos.org; (bottom) 2013 MODIS Aqua
image showing elevated chlorophyll-a levels

NASA’s Applied Remote Sensing Training Program 17

Water Quality Indicators Observable from Satellites

• Turbidity and Sediments Phytoplankton Bloom in the Gulf of Alaska

SNPP-VIIRS June 9, 2016
• Colored Dissolved Organic Matter
(CDOM)
• Sea Surface Temperature (SST)
• Chlorophyll-a (phytoplankton)
• Salinity
• TSS (Total Suspended Solids)
• Fluorescence Line Height
• Euphotic Depth

NASA’s Applied Remote Sensing Training Program 18

Water Quality Monitoring from Remote Sensing

• Typically optical and infrared spectral bands are used for water quality monitoring
• ETM+, OLI, MODIS, VIIRS, MSI, and OLCI measurements cover optical to infrared
spectral ranges in different spectral bands (Appendix A)
• These measurements have been used to monitor water quality over open oceans,
coastal waters and estuaries, and inland lakes

Image Credit: Chlorophyll-a from SeaWIFs

NASA’s Applied Remote Sensing Training Program 19

How Light Interacts with Water

bb(λ)
Rrs(λ,0+) ≅ C
a(λ) + bb(λ)

Inherent Optical Properties

a = absorption by…
• phytoplankton (ph)
• non-algal particles (nap)
• colored dissolved organic matter (CDOM)
aw • water (w)
aph aCDOM
anap
bb b = scattering in forward (f) and backward
(b) directions

fluorescence bf

NASA’s Applied Remote Sensing Training Program 20

How Light Interacts with Water

bb(λ) Lw(λ)
Rrs(λ,0+) ≅ C =
a(λ) + bb(λ) Ed(λ,0+)
RRS

Inherent Optical Properties

a = absorption
ED b = scattering
Lw

aw Apparent Optical Properties

LU
Lw = water leaving radiance
aph aCDOM
anap Lu = upwelling radiance
bb Ed = downwelling irradiance
Rrs = remote sensing (rs) reflectance

fluorescence bf

NASA’s Applied Remote Sensing Training Program 21

Inherent Optical Properties (IOPs) and the ‘Color’ of Water

Light absorption (a) by photoplankton

(ph), non-algal particles (nap), water
(w), and colored dissolved organic
matter (CDOM)

a = aph + anap + aCDOM + aw

Light scattering (b) by particles in

forward (bf) and backward (bb)
direction b = bf + bb

NASA’s Applied Remote Sensing Training Program 22

Inherent Optical Properties (IOPs) and the ‘Color’ of Water

Chlorophyll

Water

CDOM

Nap/
Sediments

NASA’s Applied Remote Sensing Training Program 23

Inherent Optical Properties (IOPs) and the ‘Color’ of Water

Visible Near IR
Chlorophyll

sediments

Water

RRS(sr-1)
CDOM
CDOM
water
chlorophyll

Nap/
Sediments wavelengths (nm)

NASA’s Applied Remote Sensing Training Program 24

Remote Sensing of Water Bodies
Techniques
1. Simple image interpretation to derive 2. Different algorithms combine
qualitative information about water atmospherically corrected satellite
quality images and in situ measurements to
derive quantitative information about
water quality

In situ
observations
required

NASA’s Applied Remote Sensing Training Program 25

Remote Sensing of Water Bodies

• Satellite sensors measure top-of-

atmosphere (TOA) radiances
• The TOA radiances result from a
combination of surface and
atmospheric conditions, including
effects of clouds and aerosol particles
• Water-leaving reflectance depends
on:
– backscattering and absorption of
radiation due to water, sediments,
phytoplankton, and colored
dissolved organic matter (CDOM)

Image Credit: http://www.oceanopticsbook.info/view/remote_sensing/the_atmospheric_correction_problem

NASA’s Applied Remote Sensing Training Program 26

Atmospheric Correction for Water Quality Monitoring

• Satellite observations of reflectance Examples:

have to be corrected for atmospheric • NASA Ocean Biology Processing Group
effects for getting water surface Algorithm: https://oceancolor.gsfc.nasa.
reflectance gov/docs/technical/NASA-TM-2016-
• Various techniques exist for the 217551.pdf
atmospheric corrections • *6S: Second Simulation of the Satellite
Signal in the Solar Spectrum:
• Requires raditaive transfer modeling http://6s.ltdri.org/#
along with atmospheric conditions,
• ACOLITE: https://odnature.naturalsciences.
clouds, and aerosol information
be/remsem/software-and-data/acolite
• HydroLight: http://www.oceanopticsbook.
info/view/radiative_transfer_theory/level_2
/hydrolight

*Vermote, E.E., D. Tanré, J.L. Deuzé, M. Herman and J.-J. Morcrette, Second Simulation of the Satellite Signal in the Solar Spectrum, 6S: An Overview, IEEE Transactions on Geoscience and Remote Sensing,
Vol. 35, No. 3, p. 675-686., 1997. r12_Stumpf_Tomlinson.pdf

NASA’s Applied Remote Sensing Training Program 27

Water Quality Parameters from Remote Sensing Observations
Quantitative Technique
Algorithm Development Monitoring

Satellite TOA Atmospheric Atmospherically

Reflectance Correction Corrected Real
Over a Water Time or Current
Body Satellite
Water Leaving Overpass
Reflectance Reflectance
In Situ
Observations of
WQ Parameters
During a Satellite Statistical or
Overpass Empirical Model Derived WQ
Algorithm Coefficients Parameter
Development
Past Time Series
of Observations

NASA’s Applied Remote Sensing Training Program 28

Example: Chlorophyll-a Estimates from Ratios of Reflectance
Surface Remote Sensing Reflectance
10.0
Example: Ratio
of Rrs value at

VIIRS Chl a
486 nm and
1.0
550 nm

0.1

0.1 1.0 10.0

In situ Chl a (mg m-3)

Algorithm description: http://oceancolor.gsfc.nasa.gov/cms/atbd/chlor_a

NASA’s Applied Remote Sensing Training Program 29

Monitoring WQ in Coastal and Inland Waters
Remote Sensing Data Source Considerations

• What geographical and atmospheric (especially persistent cloud cover) conditions

exist?
• What is the spatial resolution of the data and how appropriate is it, relative to the
size of the water body to be monitored?
• What is the temporal resolution in terms of potential frequency of acquisition of
non-cloudy observations compared to the desired frequency of monitoring?
• What are the spectral regions, and bands within them, and how do these relate to
the potential for distinguishing water quality?
• What is the longevity of the image archive length – does this meet the historical
mapping needs?
• What are the cost implications of these data in terms of purchase and analysis?
• What are the future satellite development and launch commitments?

NASA’s Applied Remote Sensing Training Program 31

NASA Ocean Color Web from Ocean Biology Processing Group
https://oceancolor.gsfc.nasa.gov/
• Primarily designed for coastal and
open oceans
• Useful for monitoring ‘large’ in-land
lakes and estuaries
• Provides historical and current data
from various satellites and sensors
• Focuses on processing remote sensing
imagery to derive Chlorophyll
Concentration (Ch) and Sea Surface
Temperatures (SST)

NASA’s Applied Remote Sensing Training Program 32

NASA Ocean Color Data Products
https://oceancolor.gsfc.nasa.gov/atbd/
• Algorithms are derived based on
spectral band ratios and SeaBASS in
situ measurements
– https://seabass.gsfc.nasa.gov

NASA’s Applied Remote Sensing Training Program 33

NASA Ocean Color Data Products Algorithms
https://oceancolor.gsfc.nasa.gov/atbd/
• Algorithms to derive Ocean Color (OC) • The algorithm for Particulate Organic
products from MODIS and VIIRS Carbon (POC) is a power-law
• The OC algorithm is a fourth-order relationship between a ratio of Rrs from
polynomial relationship between a MODIS and POC
ratio of Rrs and chlor_a poc=a×(Rrs(443)/ Rrs(557))b
• Particulate Inorganic Carbon (PIC)
– 2-band approach: normalized water-
leaving radiances in two bands near
443 and 555 nm
• SSTs are derived from infrared window – 3-band approach: spectral top-of-
radiances (11 and 4 micron) atmosphere reflectance at three
wavelengths near 670, 750, and 870
nm

NASA’s Applied Remote Sensing Training Program 34

NASA Ocean Color Data Products Algorithms
https://oceancolor.gsfc.nasa.gov/atbd/
• Kd490, a diffuse attenuation coefficient:
– Indicates how strongly light intensity at a given wavelength is attenuated within
the water column
– Useful for
• characterizing water optical properties
• classifying water types (e.g., Case 1 or 2)
– Is a critical parameter for accurate estimation of the light intensity at depth
– Useful for measuring Water Turbidity and Transparency
– Derived by using an empirical relationship between Kd(490) and the blue‐green
normalized water‐leaving radiance ratio (NASA’s Ocean Color Processing
Group)

NASA’s Applied Remote Sensing Training Program 35

NASA Ocean Color Data Products Algorithms
https://oceancolor.gsfc.nasa.gov/atbd/
• Normalized Fluorescence Line Height (normalized to solar irradiance)
– A measurement of solar-induced phytoplankton chlorophyll fluorescence
emission at ~678 nm
– Useful to improve ocean photosynthesis estimates
– Helps resolve climate-phytoplankton interactions
– Characterize iron stress in the global ocean
• Derived from difference between the observed nLw(678) and a linearly
interpolated nLw(678) from two surrounding bands. (NASA’s Ocean Color
Processing Group)

NASA’s Applied Remote Sensing Training Program 36

Water Quality Data Product Algorithms

• Various spectral combinations are used in different algorithms to derive water

quality parameters (Gholizadeh et al. 2016)
• In addition to NASA Ocean Color data based on MODIS and VIIRS, several studies
have used Landsat 5, 7, and 8 to derive water quality parameters

Gholizadeh, M., Melesse, A., & Reddi, L. (2016). A Comprehensive Review on Water Quality Parameters Estimation Using Remote Sensing Techniques. Sensors, 16(8), 1298. doi:10.3390/s16081298

NASA’s Applied Remote Sensing Training Program 37

Harmful Algal Blooms (HABs)
https://arset.gsfc.nasa.gov/water/webinars/HABs17
• HABs occur when colonies of algae • Generally, chlor_a is used as an
grow out of control indicator for HABs
– algae: simple fresh and sea water • Chlor_a anomalies (departure from
plants mean value), and inherent optical
• HABs: properties can also be used an
– Produce toxins indicator of HABs
– Cause economic losses
– Contaminate drinking water
– Smother benthic organisms
– Deplete oxygen
– Attenuate light to submerged
aquatic vegetation or corals
Image: Landsat 8 (OLI) Aug 1, 2014

NASA’s Applied Remote Sensing Training Program 38

Demonstration of NASA Web Tools for Water
Quality Data Access
Web Tools

• Data search, spatial and temporal subsetting, analysis, and visualization:

– Giovanni: http://giovanni.gsfc.nasa.gov/giovanni/
– OceanColor Web: https://oceancolor.gsfc.nasa.gov/

NASA’s Applied Remote Sensing Training Program 40

Case Study: Chesapeake Bay Spring Algal Bloom 2018
http://eyesonthebay.dnr.maryland.gov/eyesonthebay/habs.cfm

Satellite detection of algal bloom in the Chesapeake Bay on May 1, 2018. Source NOAA, Sentinel 3

NASA’s Applied Remote Sensing Training Program 41

Exercise: Lake Victoria
Appendix A
Landsat Bands

Landsat 7 ETM+ Landsat 8 OLI

Spatial Spatial
Band Band Range (μm) Band Band Range (μm)
Resolution (m) Resolution (m)
1 0.45 – 0.515 1 0.433 – 0.453
2 0.525 – 0.605 2 0.450 – 0.515
3 0.63 – 0.69 30 3 0.525 – 0.60
4 0.775 – 0.90 4 0.630 – 0.680 30
5 1.55 – 1.75 5 0.845 – 0.885
6 10.4 – 12.5 60 6 2.10 – 2.30
7 2.08 – 2.35 30 7 0.500 – 0.680
8 0.52 – 0.9 15 8 2.08 – 2.35 15
9 1.36 – 1.39 30

NASA’s Applied Remote Sensing Training Program 44

MODIS Bands Relevant for HAB Monitoring

Chlorophyll Concentration from Aqua

Band Band Range µm MODIS, June 2017
8 0.405-0.420
9 0.438-0.448
10 0.483-0.493
11 0.526-0.536
12 0.546-0.556
13 0.662-0.672
14 0.673-0.683
15 0.743-0.753

Spatial resolution: 1 km

NASA’s Applied Remote Sensing Training Program 45

VIIRS Bands Relevant for HAB Monitoring

Phytoplankton Bloom in the Gulf of Alaska, from VIIRS, June 9, 2016

Band Band Range µm

M1 0.402-0.422
M2 0.436-0.454
M3 0.478-0.488
M4 0.545-0.565
M5 0.662-0.682
M6 0.739-0.745

Spatial Resolution: 750 m

NASA’s Applied Remote Sensing Training Program 46

MSI bands
https://earth.esa.int/web/sentinel/user-guides/sentinel-2-msi
Algal Bloom in the Middle of the Baltic
Sea, Sentinel-2 MSI, Aug 7, 2015

Image: Copernicus Sentinel data (2015)/ESA, CC BY-SA 3.0 IGO

NASA’s Applied Remote Sensing Training Program 47

OLCI Bands
https://sentinel.esa.int/web/sentinel/user-guides/sentinel-3-olci
Sentinel-3 OCL-Based
Chlorophyll
Concentration

Image Credit: ESA/ACRI-ST

NASA’s Applied Remote Sensing Training Program 48

Thank You