World Academy of Science, Engineering and Technology

International Journal of Agricultural and Biosystems Engineering

Vol:11, No:7, 2017

Extraction of Forest Plantation Resources in Selected

Forest of San Manuel, Pangasinan, Philippines Using
LiDAR Data for Forest Status Assessment
Mark Joseph Quinto, Roan Beronilla, Guiller Damian, Eliza Camaso, Ronaldo Alberto

 Foundation, Inc, about 755,000 x 104 square meters are still

Abstract—Forest inventories are essential to assess the unclassified and the use of these forest areas has not yet been
composition, structure and distribution of forest vegetation that can determined. In fact, two thirds of the country’s total land area
be used as baseline information for management decisions. Classical
Digital Open Science Index, Agricultural and Biosystems Engineering Vol:11, No:7, 2017 waset.org/Publication/10007712

was designated as forest areas [1]. From the 2001-2003

forest inventory is labor intensive and time-consuming and
satellite imageries of the National Mapping and Resource
sometimes even dangerous. The use of Light Detection and Ranging
(LiDAR) in forest inventory would improve and overcome these Information Authority (NAMRIA), the estimation of actual
restrictions. This study was conducted to determine the possibility of forest coverage is at 7.168 million x 104 square meters or
using LiDAR derived data in extracting high accuracy forest 24.27% of the country’s total land area [2]. From then on, the
biophysical parameters and as a non-destructive method for forest need to conduct more studies towards the classification of the
status analysis of San Manual, Pangasinan. Forest resources Philippine’s forest cover as well as the forest inventory was
extraction was carried out using LAS tools, GIS, Envi and .bat scripts
recognized to provide more data in the Philippine forests and
with the available LiDAR data. The process includes the generation
of derivatives such as Digital Terrain Model (DTM), Canopy Height detailed forest resources database. Forest inventories are
Model (CHM) and Canopy Cover Model (CCM) in .bat scripts designed to measure the extent, quantity, composition, and
followed by the generation of 17 composite bands to be used in the condition of forest resources [3].
extraction of forest classification covers using ENVI 4.8 and GIS Countries all over the world are requested to provide data
software. The Diameter in Breast Height (DBH), Above Ground and estimates for forest areas [4]. The emerging method for
Biomass (AGB) and Carbon Stock (CS) were estimated for each
estimating forest areas and resources is through LiDAR
classified forest cover and Tree Count Extraction was carried out
using GIS. Subsequently, field validation was conducted for accuracy surveys due to its data accuracy and non-destructive method of
assessment. Results showed that the forest of San Manuel has 73% estimation. Forest characterization with LiDAR data has
Forest Cover, which is relatively much higher as compared to the recently garnered much scientific and operational attention.
10% canopy cover requirement. On the extracted canopy height, 80% However, the number of forest inventory attributes that may
of the tree’s height ranges from 12 m to 17 m. CS of the three forest be directly measured with LiDAR is limited; but, when
covers based on the AGB were: 20819.59 kg/20x20 m for closed
considered within the context of all the measured and derived
broadleaf, 8609.82 kg/20x20 m for broadleaf plantation and 15545.57
kg/20x20m for open broadleaf. Average tree counts for the tree forest attributes required to complete a forest inventory, LiDAR can
plantation was 413 trees/ha. As such, the forest of San Manuel has be a valuable tool in the inventory process [5].
high percent forest cover and high CS. In this paper, researchers presented the forest inventory,
using different parameters for Forest Resources Extraction
Keywords—Carbon stock, forest inventory, LiDAR, tree count. from LiDAR Surveys. LiDAR derivatives used in the study
determine the level of terrain in the forest area, intensity of
I. INTRODUCTION vegetation in the area and the canopy height of trees. On the

F ORESTS are one of the most important natural resources

of the Philippines. It plays vital ecosystem functions to a
healthy environment such as water and air purification,
other hand, classification of forest cover determines the
different forest covers present in the area as well as the
corresponding area of each cover type. DBH of trees, as well
prevents soil erosion and plays a major role in climate change as AGB are important in CS determination. CS estimation
mitigation as healthy forest is a major carbon sink. But in the enables us to assess the amount of carbon that a forest can
latest study of the Philippine Tropical Forest Conservation store when a forest is generated, as well as the carbon loss if
the forest will be deforested. Lastly, Tree Count Extraction
allows us to assess the number of trees present in the forest
M. J. C. Quinto, G. B. Damian, and E. E. Camaso are with the Phil-LiDAR plantation area.
2 Project, Institute for Climate Change and Environmental Management,
Central Luzon State University, Muñoz, Nueva Ecija, 3120 (phone: 6344-456-
5843, fax: 6344-456-5843, e-mail: josephcalaunan@gmail.com, II. OBJECTIVE
virysses10@gmail.com, elie.camaso7@gmail.com).
R. F. Beronilla is with the Water Resources Management Center, Central The main objective of this study is to extract forest
Luzon State University, Muñoz, Nueva Ecija, 3120 (phone: 63997-334-8867, inventory parameters such as CCM, CHM, DTM, Forest
e-mail: roan17eonberonilla@gmail.com). Classifications, DBH, AGB, CS and Tree Count in the
R. T. Alberto is with the College of Agriculture, Central Luzon State
University, Muñoz, Nueva Ecija, 3120 (phone: 6344-456-5843, fax: 6344- selected forest areas for forest status assessment.
456-5843, e-mail: travert2157@gmail.com).

International Scholarly and Scientific Research & Innovation 11(7) 2017 530
 World Academy of Science, Engineering and Technology
International Journal of Agricultural and Biosystems Engineering
Vol:11, No:7, 2017

III. METHODOLOGY nc90, nc99 and the composite band of the municipality to be
classified. These bands in tif format are prerequisite in
A. Study Area
classifying LiDAR point clouds to forest cover classes.
The selected study area is a part of the protected forest
plantation in San Manuel, Pangasinan (120° 39’ 31’’E, 16° F. Forest Cover Classification
04’ 23’’N) with an area of 0.445 square kilometers or 44.5 x This particular method is applicable for forested areas and
104 square meters. Trees planted in the area composed mainly the basis of its forest extent was the NAMRIA land cover
of broadleaf trees (Fig. 1). classification. Composite band of the municipality was used in
ENVI to classify LiDAR point clouds to different forest
classes, wherein a spectral angle mapper was used as the
classification method. Moreover, majority analysis was
applied in ENVI to produce a smoother classification with a
kernel size of 15 m x 15 m.
Generated type of classification (lower class, upper class or
mangroves class) was the input to finally classify and separate
Digital Open Science Index, Agricultural and Biosystems Engineering Vol:11, No:7, 2017 waset.org/Publication/10007712

the type of forest classification in Arc GIS using added

classification toolbox.
G.DBH, AGB and CS Estimation
After classifying the different types of forest cover, each
forest cover were manually loaded and processed in the forest
type derivatives batch file. From then, the extracted
parameters using script were Diameter at DBH, AGB, CS and
CCM calculation for Broadleaf Plantation and Coniferous
Fig. 1 Selected forest plantation area of the study Plantation forest cover types.
B. LiDAR Dataset H.Tree Count Extraction
To have a more appropriate result or nearest true value of Tree count extraction allowed us to assess if the forest has
DTM, the laz files of Airborne LiDAR and partially processed sufficient number of trees or how many trees should be
data were shifted to WGS to EGM followed by converting the planted to make it sufficient/healthy. However, this developed
shifted laz files in LAS tools using LAS boundary. Output method is only for forest plantation. If not, the delineated
shape files processed in LAS boundary were superimposed in crown polygons of other forest types are expected to be poorly
ArcGIS MAP to select overlaid tiles included in the segmented. The CHM of the municipality of San Manuel with
municipality to be processed. LiDAR data was clipped using the NAMRIA coverage to
C. Tools/Software Used delineate and to produce tree count only to the selected forest
Extraction of forest resources such as canopy height, forest area regardless of the cover type present in the area.
cover, CS and tree count were carried out using LAS tools, I. Field Validation
.bat scripts, Envi tools and in ArcGIS MAP. Aside from the Field validation is important in determining the ground truth
foremost tools, Anaconda2 and Phyton2.7 software were also of LiDAR data and to assess the accuracy of the produced
used to run the programs, particularly the .bat scripts. map. Validation was conducted within the vicinity of the
D.Processing of Derivatives selected forest area. Two validation plots with an area of 400
Processing of Derivatives were carried out in batch using m2 per plot were established. The parameter that has been
LiDAR derivatives script to generate DTM, Hill Shade of validated was the DBH, Forest Cover Type, Tree Height, Tree
DTM (DTM_HS), CHM, Hill Shade of CHM (CHM_HS), Species, Fractional Gap and the Estimated Percent Cover.
CCM and Veg.laz. The programmed .bat script for generating J. Workflow
derivatives comprises LASboundary, LAStile, Las2las, Fig. 2 shows the workflow of the study, starting from the
Blast2Dem, LASheight, LASnoise, LASmerge, LASthin and LiDAR data, down to the generation of the map.
LASgrid. The above parameters were generated as
prerequisites in generating raster bands and a composite band IV. RESULTS AND DISCUSSION
to classify different forest types in the forest area.
A. LiDAR Derivatives
E. Generation of Bands
The generated derivatives were: DTM, Canopy Height and
There were 17 raster bands and one composite band which the intensity of CCM.
was generated in open_check .bat script of classification. The
generated bands were average, class_dtm, cover, density, B. Digital Terrain Model
maximum, quadratic average, standard deviation, bincentile The minimum value of the DTM is a bit higher because the
bands such as nc10, nc20, nc30, nc40, nc50, nc60, nc70, nc80, location of the forest area was away from the coastal area. Fig.

International Scholarly and Scientific Research & Innovation 11(7) 2017 531
 World Academy of Science, Engineering and Technology
International Journal of Agricultural and Biosystems Engineering
Vol:11, No:7, 2017

3 shows that highest elevation of the forest area ranging from meters to 17 meters. Overall, the mean average generated was
141 to 191 meters high. Hence, the elevation of the majority of 2 meters high (Fig. 5).
the forest area ranged from 94 - 141 meters, followed by 66
meters to 94 meters high.
Digital Open Science Index, Agricultural and Biosystems Engineering Vol:11, No:7, 2017 waset.org/Publication/10007712

Fig. 3 DTM of the forest area

Fig. 4 CCM (%)

Fig. 2 Workflow of the study

C. Canopy Cover Model
The generated Canopy Cover Model shows 74% to 100%
canopy cover. Field validation shows that only a few areas of
the forest has an open canopy, and one of which is the area
where the plot was established or located i.e. on the right side
of the map model. Types of forest cover generated and present
in the area were open broadleaf, closed broadleaf and
broadleaf plantation. Among the three cover types, the
broadleaf plantation has the highest canopy cover of
approximately 80% (Fig. 4).
D. Canopy Height Model
The generated Canopy Height of trees in the forest ranged Fig. 5 CHM
from 0 meters to 17 meters or greater than 17 meters;
however, the height of the majority of the trees ranged from 8

International Scholarly and Scientific Research & Innovation 11(7) 2017 532
 World Academy of Science, Engineering and Technology
International Journal of Agricultural and Biosystems Engineering
Vol:11, No:7, 2017

E. Forest Cover Classification direct relationship with each other, as obviously shown in the
Based on the model generated, there were three classes of Open Broadleaf Plantation as well as the other two forest
Forest Cover in the study area, namely and as Fig. 6 shows: cover, whereby, a tree with a bigger DBH has a heavier AGB
Broadleaf, Open Broadleaf and Closed Broadleaf. The area, and with a concomitant high acquisition in CS. An indication
however, was dominated by Closed Broadleaf with 21 that the forests are in a very healthy status as it has the
hectares followed by Broadleaf with 20 hectares and the Open capability to acquire high ground biomass and carbon from the
Broadleaf, which only occupies an area of 3.5 hectares (Table atmosphere. Similar results were also observed in the other
I). types of forest cover. Moreover, DBH, AGB and CS were also
found to be highest in the Closed Broadleaf as compared to the
Open Broadleaf and Plantation Broadleaf forest covers.
E. Tree Count
Shown in Fig. 7 is the delineated tree crown and tree count
of the selected forest area. The red dots (local maxima) inside
the Thiessen polygon are the representation of tree counts. The
Digital Open Science Index, Agricultural and Biosystems Engineering Vol:11, No:7, 2017 waset.org/Publication/10007712

estimated total number of trees in the forest area is equivalent

to of 18,824 trees for the three forests cover types. The
average tree count for the forest plantation is 414 tree/ha. The
red points or representation of tree counts were selected based
on the height of each delineated tree. Trees having height
greater than 1.37 meters from the ground were included in the
final estimated count.

Fig. 6 Forest Cover Classification

TABLE I
AREA OF THE GENERATED DIFFERENT FOREST COVER TYPES
CANOPY COVER TYPE AREA (ha)
Broadleaf Plantation 20.10
Closed Broadleaf 17.96
Open Broadleaf 2.50

TABLE II Fig. 7 Extracted Thiessen polygon and local maxima for the tree
ESTIMATED MEAN AVERAGE DBH, AGB AND CS OF DIFFERENT FOREST
COVER
count of forest area. The red points are the representation of tree
counts
Forest Cover Type DBH (mm) AGB (kg) CS (kg)
Broadleaf Plantation
Minimum 265.16 2731.02 1283.58
V. SUMMARY AND CONCLUSION
Maximum 849.77 8609.82 4046.61 In this study, researchers were determined and used the
Mean 595.13 5912.65 2778.94 important parameters, namely: CCM, CHM, DTM Forest
Closed Broadleaf Cover Types, DBH, AGB and the generated Tree Count in
Minimum 366.29 4955.69 2329.18 forest status assessment. These parameters are of great help to
Maximum 1137.48 20819.59 9785.21
assess forest inventory and general status of the forest. Based
Mean 679.49 11276.03 5299.73
on the results, the Canopy Height of the forest ranged from 12
Open Broadleaf
m to 17 m, with a mean canopy height of 2 m, while the
Minimum 191.97 1936.09 909.96
Maximum 801.09 15545.57 7306.42
elevation of the forest location was from 94 m to 141 m high.
Mean 500.86 6316.66 2968.83 The selected forest plantation has high percentage Canopy
Cover reaching as high as 80% of the forest area and with an
D. DBH, AGB and CS estimated CS of 21138.24 kg. The total Tree Count is 18,824
As Table II shows, among the three types of forest cover, trees or with an average of 414 trees per hectare at an
the Closed Broadleaf was observed to have the highest DBH approximate distance of 5 m X 5 m in between trees.
of 1137.48 mm, AGB of 20819.59 kg and CS of 9785.21 kg. This study could provide the needed data to the Philippine
This was followed by the Open Broadleaf with 801.09 mm data base on forest status assessment. By giving the forest
DBH, 15545.57 kg AGB and 7306.42 kg CS, and the lowest inventory and forest status to the local communities, they will
value was Broadleaf Plantation that has 849.77 mm DBH, be able to plan on how to protect and manage their forests. It
8609.82 kg AGB, and 4046.61 kg CS. will indicate which areas should be conserved and areas which
It was observed further that the three parameters have a need rehabilitation.

International Scholarly and Scientific Research & Innovation 11(7) 2017 533
 World Academy of Science, Engineering and Technology
International Journal of Agricultural and Biosystems Engineering
Vol:11, No:7, 2017

It is recommended that development of a specific process

for tree count delineation for accurate segmentation be
considered as the tree count parameter is only applicable for
forest plantations. Through this, estimating tree count on
different forest types in the Philippines will be much easier.

ACKNOWLEDGMENT
Researchers would like to thank the Department of Science
and Technology (DOST) for funding the Phil- LiDAR Project
as well as the Department of Science and Technology-
Philippine Council for Industry, Energy and Emerging
Technology Research and Development (DOST- PCIEERD)
for the management and monitoring of the project.
Digital Open Science Index, Agricultural and Biosystems Engineering Vol:11, No:7, 2017 waset.org/Publication/10007712

REFERENCES
[1] Philippines, Department of Environment and Natural Resources-Forest
Management BureauPhilippines Forestry Outlook. Asia-Pacific Forestry
Sector Outlook Study II: Working Paper No. APFSOS II/ WP/2009/10.
Bangkok: Food and Agriculture Organization of the United Nations,
Regional Office for Asia and the Pacific. 2009.
[2] Philippine Tropical Forest Conservation Foundation, Inc. Status of
Philippine Forests. http://www.ptfcf.org/wp-
content/uploads/2015/07/Status_of_Philippine_Forests. pdf
[3] Kangas, A., J. H. Gove and C. T. Scott. Introduction. In A. Kangas and
M. Maltamo. Forest Inventory. 2006. Springer, The Netherlands, pp. 3–
11.
[4] Marklund L. G. and M. Henry. Food and Agriculture Organization of the
United Nations. Status of forest biomass and carbon stock assessment in
South and Central America. 2003
[5] Wulder M. A., Bater C. W., Coops N. C., Hilker T. and J. C. White.
2012. The role of LiDAR in sustainable forest management. The
Forestry Chronicle. Downloaded from pubs.cif-ifc.org by
216.183.154.252 on 01/19/12.

International Scholarly and Scientific Research & Innovation 11(7) 2017 534