Professional Documents
Culture Documents
1 s2.0 S2215016121001928 Main
1 s2.0 S2215016121001928 Main
net/publication/351892744
The use of UAV-based remote sensing to estimate biomass and carbon stock
for native desert shrubs
CITATIONS READS
2 174
3 authors:
Shruthi Srinivasan
Texas Forest Service
17 PUBLICATIONS 301 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
The use of UAVs and advanced remote sensing technologies to develop ecological indicators for the restoration of arid ecosystems View project
All content following this page was uploaded by Zahraa M. Al-Ali on 29 May 2021.
MethodsX
j o u r n a l h o m e p a g e: w w w . e l s e v i e r . c o m / l o c a t e / m e x
Method Article
abstract
Unmanned Aerial Vehicles (UAVs) have started to receive more attention in the ecological field in the past 15
years, as they provide very high-resolution imagery that ranges from meters to millimeters. Very high-resolution
multispectral imagery obtained from UAVs can help in assessing and monitoring native desert vegetation.
Thus, this study use UAVs to develop a method to estimate the biomass and carbon stock of native desert
shrubs. The method integrates different techniques and software to monitor native plants’ coverage, biomass,
and carbon stock. The techniques used in this work are also applicable for other native desert shrubs in the
region and could support ecosystem managers in assessing and monitoring arid ecosystems and restoration and
revegetation programs. A three-stage image and data management are discussed, including: (1) fieldwork and
image acquisition using UAVs, (2) image pre-processing, and (3) image processing using different techniques and
software.
• Determining shrub biomass is not restricted to multispectral data only but could be applicable for RGB data
since it mainly depends on the DSM and DTM.
• Allometric parameters could help in estimating desert shrub biomass which could be measured easily and
rapidly using UAV imagery.
• SVM Supervised classification could help in distinguishing between native shrubs and grasses.
https://doi.org/10.1016/j.mex.2021.101399
2215-0161/© 2021 Published by Elsevier B.V. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/)
2 M.M. Abdullah, Z.M. Al-Ali and S. Srinivasan / MethodsX 8 (2021) 101399
article info
Method name: UAV image processing to estimate desert shrubs biomass and carbon stock
Keywords: UAVs, Annual plants, Arid ecosystems, Restoration ecology, Remote sensing, Plant-soil interaction, Biomass, carbon
stock
Article history: Received 31 March 2021; Accepted 10 May 2021; Available online 26 May 2021
Specifications table
Method details
The presented methodology in this paper integrates field base experiment and spatial analysis
using UAV-based remote sensing and GIS to estimate biomass and carbon stock of native desert shrub.
Fig. 1 presents a flow chart summarizing the main steps of estimating biomass and carbon stock of
native desert shrubs.
UAV Platform
This study used Parrot Disco-Pro AG (ready to fly kit) fixed-wing UAV, accompanied by Parrot
Sequoia multispectral sensor. The sensor contains multispectral bands including Green (550 BP 40),
Red (660 BP 40), Red Edge (735 BP 10), and Near-infrared (790 BP 40), as well as the sunshine sensors
(Fig. 2) [14] to collect the imagery data. The UAV size is 1150 × 580 × 120 mm (45 × 22 × 5 in) with
a wingspan of 1150 mm (45 in). The UAV covers 80 ha (200 ac) in a single flight at a flight altitude
of 120 m (400 ft) [13]. The multispectral bands were used to generate the NDVI layer, which helps
analyze the vegetation by capturing the amount of light they absorb and reflect. The sensor’s ground
resolution is 11.3 cm/px at 120 m flight altitude [14]. Decreasing the flight altitude will provide better
resolution, but the flight time will increase, which might require extra-flights.
M.M. Abdullah, Z.M. Al-Ali and S. Srinivasan / MethodsX 8 (2021) 101399 3
Fig. 1. Schematic flowchart for main stages of estimating biomass and carbon stock of native desert shrubs.
In this work, we examined ground resolution of 7.2 cm/px at 80 m flight altitude for our
experiment on 20th of January 2019 at 9:30 AM using automatic flight plan using the Pix4Dcapture
mobile application (www.pix4d.com). RGB and multispectral bands, including Green, Red, Red Edge,
and Near-infrared, were captured from the study area. One flight was implemented to cover the study
areas (a total area of 1 km2 ). Around 2400 images were captured during the flight to cover the
entire area with 80% imagery overlap. A high overlap percentage is recommended to prevent any
corrupted images affected by the wind speed, especially for fixed-wing UAVs. They are more sensitive
to wind speed than the multirotor UAVs, which have better wind resistance. All captured images were
georeferenced and radiometrically corrected, and mosaicked using the Pix4Dmapper software (version
4.3).
4 M.M. Abdullah, Z.M. Al-Ali and S. Srinivasan / MethodsX 8 (2021) 101399
Image classification
The Normalized Difference Vegetation Index (NDVI) was utilized to distinguish between healthy
vegetation and bare ground. NDVI is widely used in monitoring and assessing vegetation coverage
and distribution. The index can separate green vegetation from its background soil brightness and is
identified as the difference between the Near-Infrared (NIR) and the Red (RED) bands normalized by
the sum of those bands [10], as shown in the following equation:
(NIR − RED )
NDVI =
(NIR + RED)
Afterword, the ENVI software was used to stack the generated multispectral bands and the NDVI layer,
and the training regions of interest (ROIs) were defined from the stacked layer for three major classes
of interest, including (1) shrubs, (2) grasses, and (3) bare ground. The NDVI was utilized to select
the ROIs, as the NDVI helps identify healthy green vegetation. The Support Vector Machine (SVM)
classifier in ENVI was then used to classify the image due to its high accuracy in detecting desert
vegetation in previous works [1,3,18]. The SVM classifier is a binary machine-learning algorithm that
classifies pixels by locating the optimal statistical boundaries between classes [7].
The major step in estimating native desert shrub maximum height using 2D imagery is by
subtracting the Digital Surface Model (DSM) from the Digital Terrain Model (DTM). Therefore, the
Pix4Dmapper was used to generate the DSM and DTM and the spectral bands for the UAV image.
The processing steps included aligning images, building a dense point cloud, building a 3D mesh, and
building a field geometry. The DSM layer represents the height of all objectives in the image, including
shrubs, fences, buildings, and ground. However, the DTM represents the ground elevation used to
separate the vegetation heights from the ground elevation. The shrub classified class generated in the
classification section (Section 2.4.3) was then extracted from the DSM to determine the surface height
of the shrub class using GIS-ArcMap. The DSM for the shrub layer was then subtracted by the DTM to
determine the shrub height.
M.M. Abdullah, Z.M. Al-Ali and S. Srinivasan / MethodsX 8 (2021) 101399 5
Fig. 3. Created grids map using fishnet method considering the size of field quadrate (3m x 1m) [2].
A field experiment was designed to quantitatively measure the heights of the Rhanterium shrubs in
the field to validate the accuracy and reliability of the estimated plant maximum heights using UAV
imagery. The maximum shrub height refers to the main stem’s length above the ground surface to
the maximum plant height. In this work, 70 Rhanterium shrub sample points were selected randomly
using the very high-resolution UAV image in ArcMap. The sample points were then transferred to
the ArcPad GPS to ground truth each point using a measuring tape to record each selected sample’s
maximum height. The results of the estimated shrub heights were then compared with the ground-
truthing points measured in the field. The regression analysis test, which includes the coefficient
of determination (R²), and Root Mean Square Error (RMSE), was implemented using JMP statistical
software (version 13.0) to validate the results. The results illustrated a reasonable correlation between
the estimated shrub heights and the ground-truth measurements (R² = 0.66, p-value <0.01).
In this study, AGB for Rhanterium epapposum shrub was determined based on an actual field
experiment and laboratory analysis accomplished by [4], as shown in Eq. (1). He used a 1 x 3 m
quadrats for the same type of plant community in Kuwait during the growing seasons (winter/spring)
in three years (1997, 1998, 20 0 0). In our work, we implemented the same experiment using the spatial
analysis tools in ArcMap software (10.5) by developing a grid map for the entire study area using the
fishnet method with the same size of the plot dimension (1 x 3 m) used by [4]. These plots were
then used to estimate shrub coverage (%) was calculated for each plot (quadrate), as shown in Fig. 3.
Next, coverage (%) of 76 shrubs were selected randomly and used to estimate the AGB was estimated
according to the following equation:
where AGB: Above Ground Biomass (g/m2) and VC: estimated vegetation coverage (%)
6 M.M. Abdullah, Z.M. Al-Ali and S. Srinivasan / MethodsX 8 (2021) 101399
Fig. 4. Allometric parameters were utilized to determine the AGB. (A) Crown area was measured for each shrub using ArcMap
tools. (B) Shrub Volume (SV) was determined using a roughly spherical geometric shape formula
At this stage of the work, we investigated the influence of allometric parameters on the AGB
of desert shrubs. Regression analysis was used to develop a model to determine the relationship
between the allometric parameters and AGB. The following allometric parameters were investigated
in this study, including the crown area (CA) [6], vegetation heights (VH), and Shrub Volume (SV)
[9]. The parameters were then measured for the 76 shrub samples using ArcMap software before
implementing them in the allometric equations. We used the following equations to calculate the CA
and SV (Fig. 4).
CA = R1 ∗ R2 ∗ π (2)
where CA: crown area (m2 ), R1: radius of the longest crown diameter (m), R2: radius of the crown
diameter perpendicular to R1 (m) and Л: 3.14.
4
SV = π a2 b (3)
3
where SV: shrub volume (m3 ), a: shrub radius (m), b: shrub height (m) and Л: 3.14
However, we used the shrub heights, which were generated in the previous section. Finally, the
regression analysis was implemented to determine the relationship between the AGB and the three
examined allometric parameters using the JMP statistical software (Version 11, SAS Institute Inc.). We
found that the obtained AGB were significantly correlated with the allometric parameters, including
the CA (R2 = 0.81), SV (R2 = 0.74), however, the SH showed a lower correlation (R2 = 0.24) (Fig. 5).
Root extraction is the traditional method of estimating BGB, which is considered a destructive
method as it impacts the uprooting of shrubs and disturbs soils. It is also costly, especially when
implemented in a large-scale landscape. Thus, we used the root to shoot ratio (R/S) method to
estimate the BGB [15], which depends on multiplying the AGB (which was estimated in the previous
section) by R/S [11]. However, since this ratio mainly depends on the type of species and biomass, we
used the R/S designed for desert landscapes, ranging from 0.5 to 1, with an average of 0.75 [11,16,17].
The mean of the R/S of arid lands was implemented to estimate the BGB, as shown in the following
Eq. (4).
Fig. 5. Scatterplots between Above Ground Biomass (AGB) and shrub structure parameters. (a) Crown Area (CA), (b) Shrub
Volume (SV), and (c) Shrub Height (SH).
where BGB: Below Ground Biomass (g/m2 ), AGB: Above Ground Biomass (g/m2 ), R/S conversion factor:
0.75
Finally, total biomass (TB) was calculated using the following equation:
The TB results were used to estimate the carbon stock due to the robust relationship between
biomass and carbon storage [5]. The TB was converted to carbon stocks using the IPCC conversion
factor of 0.47 [8,19], using equations (6). Then, the carbon stock was converted into CO2 equivalent
(CO2-eq ) using Eq. (7) [12].
T C = T B ∗ 0.47 (6)
where TC: Total Carbon stocks (g/m2 ), TB: Total Biomass (AGB+BGB) (g/m2 ) and IPCC conversion
factor: 0.47
Conclusion
This paper presents an integrated method to assess and evaluate native desert plants using UAVs.
The work gives clear evidence on the effectiveness of using UAVs to estimate the above-ground
biomass of desert shrubs. It was found that the use of very high-resolution imagery can provide a
rapid and accurate assessment of native desert vegetation. The methods used in this work provide
a useful guideline on integrating different techniques and software for monitoring native plant’s
coverage, biomass, and carbon stock. The methods are also applicable for other native desert shrubs in
the region and could support ecosystem managers in assessing and monitoring arid ecosystems and
restoration and revegetation programs. However, even though UAVs could provide detailed ecological
information, including plant monitoring ecological assessment, they are not always beneficial. They
mainly depend on the scale and purpose of the study, making it applicable for small-scale studies,
which is one of the major limitations of utilizing UAVs. Thus, the integration of aerial and satellite
imagery could help in developing biological indicators that could provide ecological information for
large-scale studies.
Acknowledgment
The research was supported by the NASA Globe program at Kuwait Foundation for the
Advancement of Science (KFAS). This research is also supported by the Public Authority of Agriculture
Affairs and Fish Resources (PAAF) by providing a protected area to implement the study.
The authors declare that they have no known competing financial interests or personal
relationships that could have appeared to influence the work reported in this paper.
The authors declare the following financial interests/personal relationships which may be
considered as potential competing interests.
References
[1] MM Abdullah, L Addae-Wireko, Tena-Gonzalez GA, Assessing native desert vegetation recovery in a war-affected area using
multispectral and hyperspectral imagery: a case study of the Sabah Al-Ahmad Nature Reserve, Kuwait., Restor. Ecol. (2017),
doi:10.1111/rec.12527.
[2] MM Abdullah, ZM Al-Ali, MM Abdullah, S Srinivasan, AT Assi, S Al Atiqi, Investigating the applicability of UAVs in
characterizing desert shrub biomass and developing biological indicators for the selection of suitable revegetation sites,
J. Environ. Manage. 288 (2021) 112416.
[3] Z Al-Ali, M Abdullah, N Asadalla, M Gholoum, A comparative study of remote sensing classification methods for monitoring
and assessing desert vegetation using a UAV-based multispectral sensor, Environ. Monitor. Assess. (2020) 192.
[4] G Brown, Species richness, diversity and biomass production of desert annuals in an ungrazed Rhanterium epapposum
community over three growth seasons in Kuwait, Plant Ecol. 165 (2003) 53–68.
[5] W Chen, J Zhao, C Cao, H Tian, Shrub biomass estimation in semi-arid sandland ecosystem based on remote sensing
technology, Glob. Ecol. Conserv. 16 (2018) e00479.
[6] G Conti, L Enrico, F Casanoves, S Díaz, Shrub biomass estimation in the semiarid Chaco forest: a contribution to the
quantification of an underrated carbon stock, Ann. Forest Sci. 70 (2013) 515–524.
[7] C Huang, L Davis, J Townshend, An assessment of support vector machines for land cover classification, Int. J. Remote Sens.
23 (2002) 725–749.
[8] Ipcc2006 IPCC Guidelines for National Greenhouse Gas Inventories, Institute for Global Environmental Strategies, Hayama,
Kanagawa, Japan, 2006.
[9] JA Ludwig, JF Reynolds, PD Whitson, Size-biomass relationships of several Chihuahuan desert shrubs, Am. Midland Nat.
(1975) 451–461.
[10] CM Malmstrom, HS Butterfield, C Barber, B Dieter, R Harrison, J Qi, D Riaño, A Schrotenboer, S Stone, CJ Stoner, Using
remote sensing to evaluate the influence of grassland restoration activities on ecosystem forage provisioning services,
Restor. Ecol. 17 (2009) 526–538.
[11] X Nie, Y Yang, L Yang, G Zhou, Above-and belowground biomass allocation in shrub biomes across the northeast Tibetan
Plateau, PLoS One 11 (2016) e0154251.
[12] R Othman, SZaJIJOES Kasim, Assessment of plant materials carbon sequestration rate for horizontal and vertical landscape
design, Int. J. Environ.Sci. Develop. 7 (2016) 410.
[13] Parrot (n.d.-a) Parrot Disco-Pro. Retrieved from https://www.parrot.com/business- solutions- us/parrot- professional/
parrot- disco- pro- ag.
M.M. Abdullah, Z.M. Al-Ali and S. Srinivasan / MethodsX 8 (2021) 101399 9
[14] Parrot, (n.d.-b) Sequoia Multispectural Sensor. Retrieved from https://www.parrot.com/business- solutions- us/
parrot-professional/parrot-sequoia.
[15] NH Ravindranath, M Ostwald, Carbon Inventory Methods: Handbook for Greenhouse Gas Inventory, Carbon Mitigation and
Roundwood Production Projects, Springer Science & Business Media, 2007.
[16] P Rundel, AC. Gibson, Ecological Communities and Processes in a Mojave Desert Ecosystem, Cambridge University Press,
2005.
[17] P Rundel, P Nobel, Structure and Function in Desert Root Systems, Blackwell Scientific Publications, Oxford, 1991.
[18] L Su, MJ Chopping, A Rango, JV Martonchik, DP Peters, Support vector machines for recognition of semi-arid vegetation
types using MISR multi-angle imagery, Remote Sens. Environ. 107 (2007) 299–311.
[19] UndpGrassland Carbon Stock Calculation and Preparation of Water Balance Model for Vashlovani Protected Areas, United
Nations Development Programme, 2014.