Above Ground Forest Biomass Estimation Using Sentinel-2 Data in the Upper Blue Nile Basin, North-Western Ethiopia

doi:10.21203/rs.3.rs-4808456/v1

Download PDF

Research Article

Above Ground Forest Biomass Estimation Using Sentinel-2 Data in the Upper Blue Nile Basin, North-Western Ethiopia

https://doi.org/10.21203/rs.3.rs-4808456/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Forest biomass plays a decisive role in the global climatic condition. Therefore, rapid and accurate estimation of forest biomass is crucial for greatly reducing the uncertainty in carbon stock assessments, and for designing strategic forest management plans. Because, forest above-ground biomass (AGB) estimation is important in determining the management, environmental, and economic roles of forests in the Blue Nile basin. This study aimed to estimate forest above-ground biomass in the Upper Blue Nile Basin by integrating field-measured data with observations from Sentinel-2. Therefore, the first correlation between measured forest AGB and derived vegetation indices and biophysical parameters was investigated. A stepwise regression analysis was carried out to estimate forest AGB. The maximum forest AGB value of 268.32 t/ha was estimated in the Alemsaga natural forest. Green normalized difference vegetation index, leaf area index, fraction of absorbed photosynthetic active radiation, and fractional vegetation cover achieved moderate performance in predicting AGB with R² value > 0.5. Forest AGB was estimated with a coefficient of determination (R²) of 0.59, an adjusted R² of 0.618, and a root mean square error (RMSE) of 38.36 t/ha in comparison to field observations.

Field measured data

forest above ground biomass

vegetation indices and biophysical parameters

Forest ecosystems are a vital component in the exchange of carbon between the earth’s surface and the atmosphere (Bonan, 2008; Tao et al., 2016), serving as both carbon sinks, storing 80% of terrestrial biosphere carbon, and sources through forest growth, deforestation and degradation (Dixon et al., 1994). As a result, forests are instrumental in combating global climate change (Yue et al., 2017). Above-ground biomass (AGB) accounts for between 70% and 90% of total forest biomass and is one of the important carbon pools in forest ecosystems, besides carbon stored in soil and litter.

The Blue Nile Basin has many unique terrestrial (forest) ecosystems with thousands of plant and animal species, most of them endemic to the basin. The basin provides a wide range of environmental goods and services. Such as fuel, clean water, control of floods and erosion, sustainability of biodiversity and genetic resources, and providing opportunities for recreation and education (El-Fadel et al., 2003). Information on Forest AGB is vital for the management and monitoring of forest ecosystems. Rapid and accurate estimation and monitoring of AGB over various scales of space and time are crucial for greatly reducing the uncertainty in carbon stock assessments, and for designing strategic forest management plans (Deo et al., 2017). Therefore, forest biomass estimation in the Upper Blue Nile Basin is important for studying subsequent disturbances in the forest ecosystem (Baccini et al., 2012); a critical tool for measuring, reporting and verifying carbon stocks (Baker et al., 2010).

Various above-ground biomass estimation methods were developed and are being in the development process (Addo-Fordjour & Rahmad, 2013; Segura et al., 2018; Tetemke et al., 2019; Zhao et al., 2019). These include allometric equations using field-measured tree parameters such as tree height, diameter at breast height, crown cover, density and others. Remote sensing based measurement using mostly multi-spectral remote sensing, Radio Detection and Ranging (RADAR) and Light Detection and Ranging (LiDAR) (Lu, 2006; Lu et al., 2016). Various remote sensing products have been widely employed for forest AGB estimation across spatial and temporal scales by reducing the level of uncertainty (Pan, et al., 2011; Dou and Yang, 2018). These includes, spectral bands, vegetation indices and biophysical parameters of optical remote sensing images are Remote sensing data products have been widely employed for forest AGB estimation, to facilitate rapid and accurate forest biomass estimation across spatial and temporal scales by reducing the level of uncertainty (Pan, et al., 2011; Dou and Yang, 2018). Spectral bands, vegetation indices and biophysical parameters of optical remote sensing images are often used for AGB modeling (Du et al., 2012; Xu et al., 2013). Since different remote sensing technologies with various spectral and spatial resolutions are accessible, several potential variables may be used for AGB estimation (Lu et al., 2016). However, proper selection of features is critical for accurately estimating AGB, depending on the characteristics of forest type under investigation and remotely sensed data itself (Zhao et al., 2016).

Landsat data have been mostly used for forest AGB estimation (Powell et al., 2010; Nguyen et al., 2019), but, faced a challenge due to its increasing data saturation in fully vegetated areas, which leads to under-estimation of biomass (Kasischke et al., 1997) and less sensitive to vegetation structure, which retards acquiring vital information about the vegetation. Hence, high-resolution data in narrow bandwidths are very useful to overcome data saturation, improve estimation reliably and accurately (Steininger, 2000). The possessed multi-spectral instrument (MSI) sensor onboard Sentinel-2 provides images with better spectral coverage, better spatial resolution (e.g., 10m, 20m) (Shoko and Mutanga, 2017), and increased temporal frequency than the Landsat series (Gómez, 2017; Pandit et al., 2018; Sun et al., 2019; Isbaex and Coelho, 2020). The use of Sentinel-2 data contributed to an improved performance in estimating AGB in various regional studies (Torabzadeh et al., 2019a; Navarro et al., 2019; Pandit et al., 2018; Castillo et al., 2017; Chen et al., 2019).

This study can benefit from the availability of high spatial and temporal resolution as well as the spectrally rich satellite images (e.g., Sentinel 2). The study aimed to estimate forest AGB by using field inventory data and Sentinel- 2 data in the Upper Blue Nile basin, by combining allometric models and Sentinel-2 derived predictor variables such as vegetation and biophysical indices. Specifically, the study aimed to: (i) analyze the relationship between field inventory biomass data with vegetation indices and biophysical parameters; and (ii) select the best predictor variable and develop AGB prediction model through correlation and regression analysis.

2.1. Description of the Study Area

This study was conducted on the Upper Blue Nile Basin, which is located in the north-western part of Ethiopia. The basin is defined by the Blue Nile River catchment, with the main tributary rivers such as the Rib, Guamara and Gilgele Abay which drains towards Lake Tana, and finally, the Blue Nile River flows downwards from Lake Tana (Fig. 1). The basin is one of the largest basins in Ethiopia, with an area of 11,314km². The Upper Blue Nile basin has a wide diversity of agro-ecological zones and vegetation types, which encompasses from lowland Acacia woodland to highland rainforest and afro-alpine forests (Asefa et al., 2020). To attain sound information for the management, protection, and restoration of these varieties of forest resources in the basin, there is an on-going need to understand and quantify the state and dynamics of forests AGB.

2.2. Field data collection

The basin was classified as forest and non-forest using Sentinel-2 satellite images to support the field measurements Eight forest patches (both natural and artificial) have been systematically selected as a sample namely, Zegie, Wanzaye, Alemsaga, Tara Gedam, Adet, Keranyo, Gezamen, and Gende Woyen from the study basin as indicated in Fig. 1 above. The type, size as well as number of sample plots was determined and spatially distributed using a stratified systematic random sampling approach(McRoberts et al., 2015).

A square-shaped sample plot measuring 60m by 60m was designed. This shape allows for easy layout of the sample plot boundary and can accommodate more than two pixels in Sentinel-2 data. When dealing with fixed area plots, there may be a large number of small trees. To address this, a set of nested sub-plots of different sizes for different diameter classes was used(McRoberts et al., 2015).

In each plot, the height and diameter at breast height (DBH) of each tree with a diameter ≥ 5cm were measured. DBH was measured from a conventional height of 1.3 m (Alamgir & Al-Amin, 2008; Losi et al., 2003) from the ground level using a tree caliper and diameter tape. Individual tree height in each plot was systematically measured using a clinometer and a graduated pole for low trees. Both vernacular and scientific names of tree species were first identified with the support of local people and forest technicians in the forest. In line with the above data collection, sample training data which was used for land use land cover classification also be collected from each land use type in the basin using handheld GPS.

AGB Estimation using Allometric Equation

AGB can be estimated through various models of allometric equations. These can be destructive (Aboal et al., 2005), non-destructive (Peltier et al., 2007), general allometric equation (Houghton, 2003), and species-specific allometric equation (Kairo et al., 2009). In this study general non-destructive method of allometric equation was employed for the estimation of AGB in the study basin. Estimates of biomass are largely the results of a common equation applied over a large area (Houghton, 2003). The advantage of applying general allometric equations is that the equations were derived from a large number of trees with a wide range of DBH. This could improve the accuracy of biomass estimation (Brown, 2002). Allometric methods were developed from non-destructive surrogate measurements, such as diameter at breast height. The non-destructive or indirect method attempts to estimate tree biomass by measuring variables that are more accessible and less time consuming to assess (e.g., wood volume and density) (Peltier et al., 2007).

In order to represent all forest types in the analysis and to convert field measurements into above ground biomass estimates the allometric equation proposed by Chave et al., (2014) was used which was developed with a very high coefficient of determination (R² ≥ 99%) for many different forest types across various tropical countries.

$$\:\varvec{A}\varvec{G}\varvec{B}=0.0673\varvec{*}{\left(\varvec{W}\varvec{D}\varvec{*}{\varvec{D}\varvec{B}\varvec{H}}^{2}\varvec{*}\varvec{H}\right)}^{0.976}$$

………..

Whereby WD is the wood density (g/cm3), DBH is the diameter at breast height (cm), and H is the total height of the tree (m).

According to Chave et al., (2014), the inclusion of country specific wood density in the equation significantly improves biomass estimation. Therefore Ethiopia conducted a comprehensive study to determine the most suitable wood density estimate for the country, covering 421 indigenous and exotic tree species. The average wood density for these species was found to be 0.612 g/cm3, which is consistent with global and tropical Africa averages (Brown & Lugo, 1990; Chave, et al., 2009; IPCC, 2006; Reyes et al., 1992). The lowest wood density observed was 0.262 g/cm3 for Moringa species, while the highest was 1.04 g/cm3 for Dodonaea angustifolia species (EFRL-UNFCC-2016). These specific wood density values were used for each type of tree species in the study basin. The study estimated the available above-ground tree biomass by applying allometric equations to the inventoried individual trees and extrapolated the results to the entire area.

2.3. Remote sensing data Preprocessing

Sentinel-2B_MSIL1C data was acquired during January 2022 with low cloud cover season from the Copernicus Scientific Data Hub website (www.scihub.copernicus.eu) (ESA 2019). The raw image image was processed using Sentinel Application Platform (SNAP) and Quantum GIS (QGIS) software. Radiometric correction, contrast stretching, and histogram equalization were applied to improve the visual interpretation of the image. Contrast stretching is a technique used to widen the range of brightness values in the output image for better visibility (Lillesand Kiefer & Chipman, 2000). The image was also re-projected to the appropriate datum and projection for Ethiopia, namely the UTM coordinate system, map zone 37 N of Clack WGS 1880, to avoid data distortion due to datum conflict and to allow for the use of overlay techniques. Subsequently, all image bands were re-sampled to a 10m resolution using the SNAP software. Atmospheric calibration was performed using the Sen2Cor software, which is specific to Sentinel-2 data (Muller et al., 2013).

In addition, the Global Digital Elevation Model (GDEM) data, with a spatial resolution of 12.5m, was re-sampled to a 10m cell size and used for topographic correction of the Sentinel-2 data using the C-correction model (Sola et al., 2016). Finally, the pre-processed and re-sampled Sentinel-2 image, using nearest neighbor interpolation, was utilized to derive vegetation indices (VIs) and biophysical variables (BPVs) with the combination of the multispectral bands as input parameters for predicting forest AGB.

Land use land cover classification

The main goal of the image classification in this study is to differentiate the total forest area from other land use types in the Blue Nile Basin and estimate the total forest above-ground biomass stored in the basin. The basin was classified into five major land use classes - Agricultural land, Forest land, Built-up areas, Water bodies, and Grasslands. Image classification was performed using the semi-automatic classification plug-in tool and a maximum likelihood classification algorithm in QGIS software. The accuracy of supervised image classification was determined using a confusion error matrix with twenty-five reference data sets as ground truth points.

Prediction of Vegetation Indices and Biophysical Parameters

Based on their performance on AGB estimation in previous studies (Pandit et al., 2018), here we selected Normalized Difference Vegetation Index (NDVI), Soil-adjusted Vegetation Index (SAVI), Green-NDVI (GNDVI), and Enhanced Vegetation Indices (EVI) to estimate AGB (Table 1).

Additionally biophysical variables describe the spatial distribution of vegetation state and dynamics and, thus, are useful for biomass estimation (Widlowski et al., 2004). Therefore, we here computed leaf area index (LAI) (Kaasalainen et al., 2015), Fractional vegetation cover (FVC), fraction of photosynthetic active radiation (FAPAR), and leaf chlorophyll content(CAB) ( Baloloy et al., 2018), were generated from the atmospherically corrected and re-sampled (10m) Sentinel-2 data using SNAP tool (Louis et al., 2016). The implementation in SNAP to estimate biophysical variables is based on neural networks that were trained against forward simulations of a radiative transfer models (Muhe S. & Argaw, M. 2022). We are aware that the data used to train this algorithm mostly represents short vegetation such as crops and grasses and is not accurate to estimate e.g. LAI in forests. Hence, the derived LAI cannot be taken as true forest LAI but we here use it rather as additional data-derived feature to estimate AGB.

The pixel values for each variable derived from the Sentinel-2 image were extracted using zonal statistics in QGIS software. The locations of each field plot points were used as references to match the pixels.

Table 1

Vegetation Indices and Biophysical Variables calculated from Sentinel-2 data and used in this study.
Indices/Variable	Indices/Variable	Description
Vegetation Indices	NDVI	(Band 8 - Band 4)/(Band 8 + Band 4)
	GNDVI	(B8 – B3)/(B8 + B3
	SAVI	(Band 8- Band 4/Band 8 + Band 4 + 0.5)*1.5
	EVI	2.5*((NIR − R)/ (1 + NIR + 6R − 7.5 Blue))
Biophysical Variables	LAI	Leaf Area Index
	FAPAR	Fraction of absorbed photo synthetically active radiation
	FVC	Fractional vegetation cover
	CAB	Leaf chlorophyll content

3.1. Correlation and Regression Analysis

We performed Pearson’s correlation analysis between all features and AGB to assess their strength and potential predictive capability for AGB estimation. We used a pair-wise product-moment correlation test to determine the correlation between observed AGB and predictor variables from Sentinel-2 images, such as vegetation indices (NDVI, SAVI, GNDVI, SR) and biophysical variables (LAI, FAPAR, FVC, CAB) using SPSS version 21 software. The predictor variables showing significant correlation with estimated AGB were considered important and selected for further regression analysis. We then conducted multiple regression analyses between estimated AGB and the selected significantly correlated predictor variables to develop a prediction model for AGB estimation. Regression analysis was chosen as it is the most commonly used method for estimating AGB of forests (Hamdan et al., 2014; Hirata et al., 2014; Pham et al., 2017).

Out of 136 sample plots, 99 (72.8%) were randomly selected for training the multiple regression models, and 37 (27.2%) were used for testing. The evaluation of the model was based on the coefficient of determination (R²) value and the Root Mean Square Error (RMSE) between field-estimated AGB (AGB_e) and satellite-predicted AGB (AGB_p):

$$\:\varvec{R}\varvec{M}\varvec{S}\varvec{E}=\sqrt{\frac{\sum\:_{\varvec{i}=1}^{\varvec{n}}{\left(\varvec{A}\varvec{G}{\varvec{B}}_{\varvec{e},\varvec{i}}-\varvec{A}\varvec{G}{\varvec{B}}_{\varvec{p},\varvec{i}}\right)}^{2}}{\varvec{n}}}$$

……………………………

The regression model utilized the variables with the highest R² and lowest RMSE values to obtain the best model. Subsequently, the derived equation was employed to estimate forest AGB values.

4.1. Field measured Forest Above Ground Biomass Estimation

The study forest's above ground biomass was calculated using field data from eight selected forest patches with 136 sample plots and an allometric equation. The AGB per sample was then converted into AGB per hectare, measured in tons per hectare, to enable further correlation analysis with the sentinel prediction variables. The top three AGB values were estimated from Tara Gedam, Alemsaga, and Zegie natural forests, with values of 192.63 t/ha, 186.68 t/ha, and 175.8 t/ha, respectively. The lowest AGB value was recorded from Wanzaye forest at 3.13 t/ha, as shown in Table 2 below.

Table 2

Estimated minimum and maximum Forest AGB values from sample plots.
No	Name of Forest	Total No of Plots	Min. (t ha-1)	Max (t ha-1)	Mean (t ha-1)	Standard Deviation
1	Zege	15	85.7	175.8	130.1	21.81
2	Wanzaye	15	3.13	94.8	45.4	32.13
3	Alemesaga	20	57.25	186.68	121.49	36.66
4	Genede Weyen	25	92.71	154.6	122.4	16.12
5	Gezamen	6	47.9	105.4	71.1	1.61
6	Keranyo	20	46.12	105.76	73.47	15.69
7	Tara Gedam	20	21.11	192.63	104.43	46.72
8	Adet	15	41.97	98.72	67.71	19.61

4.2. Sentinel − 2 predictor variable computation

The selected Sentinel-2 predictor variables, such as vegetation indices (NDVI, GNDVI, SAVI, and EVI) and biophysical variables (LAI, FVC, FAPAR, and CAB), were calculated using QGIS and SNAP software. The pixel values for each predictor variable were computed, and the value ranges were as follows: NDVI: 0.56 to 0.82, GNDVI: 0.70 to 0.70, SAVI: 0.59 to 0.82, EVI: 0.19 to 0.69, LAI: -0.34 to 5.02, FVC: 0 to 0.87, FAPAR: 0.49 to 0.91, CAB: 40.90 to 298.99. These details are also shown in figure-2 in the below.

4.2.1. Correlation Analysis between estimated AGB and Vegetation Indices

Normalized Difference Vegetation Index (NDVI) and Above Ground Biomass (AGB); The coefficient of correlation (R²) between the field-measured forest above-ground biomass (AGB) and the Sentinel-2 derived predictor variable (NDVI) was 0.52. This means that approximately 52% of the field data can be explained by the NDVI-based model, while 48% of the field data cannot be explained by this model, as shown in figure-3 below. Despite this, the correlation with NDVI is considered good, given its widespread use in vegetation mapping and biomass modeling. The correlation with NDVI is higher than that of GNDI and SAVI, but lower than that of EVI from vegetation indices. In terms of biophysical parameters, the correlation with NDVI is greater than that of FAPAR and CAB, but less than that of LAI and FVC.

Several studies have produced varying results regarding the correlation coefficient between above-ground biomass (AGB) and the Normalized Difference Vegetation Index (NDVI). In a study by Ali et al., (2018), NDVI was used as a predictor for biomass and showed a high correlation with field-measured AGB, with a coefficient of correlation (R² = 0.85). However, in contrast, Adan, (2017), reported that NDVI derived from Sentinel-2 is a broadband index and has saturation issues, especially in areas with higher biomass density. This resulted in a lower correlation coefficient (R² = 0.1). Similarly, Nuthammachot et al., (2018) reported a correlation coefficient of R² = 0.65 between Sentinel-2 derived NDVI and AGB. Additionally, Hamdan et al., (2014a) found a correlation coefficient of R² = 0.543 when predicting AGB.

Green Normalized Difference Vegetation Index (GNDVI) and Above Ground Biomass (AGB ); The correlation coefficient (R²) between the field-measured forest biomass (AGB) and the GNDVI predictor variable derived from Sentinel-2 was 0.47. This indicates that approximately 47% of the variability in the field data can be explained by the GNDVI-based model, while the remaining 53% is not accounted for by this model, as depicted in Fig. 4 in the appendix. Notably, this value is lower than all selected vegetation indices (NDVI, SAVI, EVI) and biophysical variables (LAI, FAPAR, FVC), but higher than that of CAB.

Furthermore, a study by Ali et al., (2018) reported a correlation coefficient (R²) of 0.49 for the GNDVI derived from Sentinel-2. Additionally, Théau et al., (2021) found a correlation coefficient value (R²) of 0.75 for GNDVI when compared with field-measured estimated forest AGB. GNDVI is more sensitive to variations in chlorophyll in plants than NDVI and has a higher saturation point. It can be utilized in dense canopies and in more advanced stages of plant development.

Soil Adjusted Vegetation Index (SAVI) and Above Ground Biomass (AGB); The coefficient of correlation (R²) between forest above-ground biomass (AGB) from field-measured data and the Sentinel-2 derived predictor variable (SAVI) was 0.48. This means that approximately 48% of the field data is explained by the SAVI-based model, while 52% of the field data is not explained by this model. This information is shown in Fig. 3 in below. The correlation result of SAVI is greater than only from (GNDVI) but less than (NDVI and EVI) from vegetation indices.

In the study by Torabzadeh et al., (2019), Sentinel-2 data was used for AGB estimation and reported a higher R² value of 0.86 and an RMSE of 10.69 t/ha. Similarly, Navarro et al., (2019) integrated UAV, Sentinel-1, and Sentinel-2 Data for Mangrove Plantation AGB monitoring in Senegal, using SAVI as a predictor variable, and reported a correlation coefficient of R² of 0.89 and RMSE of 2.35 Mg/ha. In contradiction, Hamdan et al., (2014b) integrated SPOT-5 and ALOS PALSAR images, used SAVI as a biomass predictor variable and reported a lower correlation coefficient value of R² of 0.013 and RMSE of 68.206 Mg/ha. Also, Sentinel − 2 derived SAVI was negatively correlated with R² of − 0.272( Li et al., 2020).

Enhanced Vegetation Index (EVI) and Above Ground Biomass (AGB); The coefficient of correlation (R²) between forest above-ground biomass (AGB) measured in the field and the predictor variable derived from Sentinel-2 (EVI) was 0.55. This means that approximately 55% of the field data is explained by the EVI-based model, while the remaining 45% is not accounted for by this model (see Fig. 4 in the appendix). The correlation value of EVI is higher than that of all other selected predictor variables, including vegetation indices and biophysical parameters. Several studies have deemed Sentinel-2-induced EVI to be a reliable predictor. For instance, Pandit et al., (2018),utilized Sentinel-2 data-derived EVI as a predictor variable to estimate forest AGB, reporting a higher correlation coefficient of 0.80 compared to other variables and a lower RMSE value of 25.72 t/ha. Similarly, Torabzadeh et al., (2019) used Sentinel-derived EVI as a predictive variable for biomass and obtained a strong correlation coefficient value (R²) of 0.81, along with a lower RMSE value of 11.8 t/ha.

4.2.2. Correlation Analysis between estimated AGB and Biophysical parameters

Leaf Area Index (LAI) and Above Ground Biomass (AGB); The coefficient of correlation (R²) between forest AGB from field measured data and sentinel-2 derived predictor variable (LAI) was 0.52, which meant that about 52% of field data have been explained by LAI based model, while 48% of field data are not explained by this model shown figure-4 below.. The correlation result of LAI is greater than from (FAPAR and CAB) and less than from (FVC) selected predictor variables of biophysical parameters.

Various studies reported that LAI was a good predictor variable for forest AGB estimation, since it showed strong coefficient of correlation with field observed AGB. On the research (Muhe & Argaw, 2022) LAI induced from sentinel data was used as biomass predictive variable and reported that have a correlation coefficient R² value of 0.74. Similarly (Baloloy et al., 2018; Srinet et al., 2018) reported that LAI higher showed a correlation coefficient values R² of 0.71 and R² of 0.92 respectively. Hence LAI can be considered as a good predictor variable of forest AGB as it showed significant correlation coefficient with field measured AGB.

Fractional Vegetation Cover (FVC) and Above Ground Biomass (AGB); The coefficient of correlation (R²) between forest AGB from field measured data and sentinel-2 derived predictor variable (FVC) was 0.53, which meant that about 53% of field data have been explained by FVC based model, while 47% of field data are not explained by this model shown Fig. 5 in the appendix. The correlation result of FVC is greater than from all selected biophysical predictor variables and vegetation indices except EVI.

On the study (Kumar et al., 2020) image based FVC estimates were compared with ground estimated FVC, and reported that FVC provided the best approximation to the ground FVC among different vegetation indices. Similar result were reported when FVC correlated with field AGB with the coefficient value of R²of 0.45 ( Baloloy et al., 2018). This observation proves that canopy and understory mixing was partly reduced by using higher resolution data. Unlike these two parameters, Chlorophyll values derived for each pixel may not have been significantly affected by undergrowth vegetation, thus, fewer errors were introduced.

Fraction of Absorbed Photo synthetically Active Radiation (FAPAR) and Above Ground Biomass (AGB); The coefficient of correlation (R²) between forest AGB from field measured data and sentinel-2 derived predictor variable (FAPAR) was 0.503, which meant that about 50.3% of field data have been explained by FAPAR based model, while 49.7% of field data are not explained by this model shown Fig. 5 in the appendix. The correlation result of FAPAR is greater than from GNDVI, SAVI and CAB but less than from NDVI, EVI, LAI and FVC biophysical and vegetation indices predictor variables.

FAPAR based estimation model has problems of anisotropic characteristics caused by sun − target − sensor geometry, the vegetation clumping effect, and the hot-spot effect. Hence they are not considered as good models for forest biomass estimation (Tao et al., 2016).

Leaf chlorophyll content (CAB) and Above Ground Biomass (AGB); The coefficient of correlation (R²) between forest AGB from field measured data and sentinel-2 derived predictor variable (CAB) was 0.45, which meant that about only 45% of field data have been explained by CAB based model, while 55% of field data are not explained by this model as shown Fig. 5 in the appendix. The correlation result of CAB is less from all selected biophysical and vegetation indices predictor variables. Similarly (Baloloy et al., 2018), reported that sentinel-2 derived CAB showed a correlation coefficient of R² of 0.45, when compared with field estimated Biomass.

4.3. Forest AGB Prediction Model Development

A stepwise linear regression analysis was used to identify reliable predictor variables for the development of a model to estimate above-ground biomass (AGB) in the forest. Sentinel predictor variables were then extracted and used to develop the AGB estimation model through stepwise linear regression analysis using SPSS software. The correlation coefficient results of the extracted sentinel-derived variables indicated a strong relationship with the estimated AGB, so all extracted sentinel predictor variables were used for the model development. The criterion probability for entering a variable into the model was set at less than 5% (p < 0.05).

Three variables, namely EVI, FAPAR, and LAI, met the criteria with p-values of 0.002, 0.034, and 0.002, respectively, as shown in Table 3, were used in the forest AGB model development. The remaining five variables were excluded from the model development due to their low coefficient of determination, as detailed in Table 3 below. Based on the regression results, the equation for biomass prediction was developed using the coefficients of EVI, FAPAR, and LAI as inputs.

Table 3

Linear regression result of the observed AGB with selected sentinel-2 predictor variables.
Coefficients
Model		Unstandardized Coefficients		Standardized Coefficients	t	Sig.	95.0% Confidence Interval for B
Model		B	Std. Error	Beta	t	Sig.	Lower Bound	Upper Bound
1	(Constant)	17.830	26.764		0.666	0.506	-35.130	70.791
	NDVI	27.827	93.693	.074	0.297	0.767	-157.575	213.229
	SAVI	182.597	103.827	.498	1.759	0.081	-22.858	388.052
	GNDVI	-233.906	124.849	− .421	-1.874	0.063	-480.960	13.148
	EVI	390.438	123.473	.740	3.162	0.002	146.107	634.769
	FVC	15.519	167.043	.039	0.093	0.926	-315.030	346.067
	FAPAR	-362.610	168.966	− .972	-2.146	0.034	-696.963	-28.257
	LAI	123.242	38.168	1.222	3.229	0.002	47.714	198.771
	CAB	− .682	.367	− .365	-1.856	0.066	-1.409	0.045
Dependent Variable: Measured AGB

This equation was designed to estimate the above-ground biomass of the study basin using the extracted prediction variables from Sentinel-2.

$$\:\varvec{P}\varvec{r}\varvec{e}\varvec{d}\varvec{i}\varvec{c}\varvec{t}\varvec{e}\varvec{d}\:\varvec{A}\varvec{G}\varvec{B}=\left(390.48\varvec{*}\varvec{E}\varvec{V}\varvec{I}\right)-\:\left(362.610\varvec{*}\varvec{F}\varvec{A}\varvec{P}\varvec{A}\varvec{R}\right)-\left(123.242\varvec{*}\varvec{L}\varvec{A}\varvec{I}\right)+17.830$$

…

Whereby; EVI is Enhanced Vegetation Index, FAPAR is Fraction of Photo synthetically Active Radiation and LAI is Leaf Area Index.

Though these prediction indices can be extracted from any Sentinel-2 image as well as used to predict forest above ground biomass by employing the developed biomass prediction equation.

Using the developed biomass estimation formula, AGB values of each sample forest stands were predicted. As indicated the table below (Table-4) highest values of forest AGB was predicted from Alemesaga, Tara Gedam and Zege forest stands with the values 268.32, 264.42 and 203.17 t/ha respectively.

Table 4

Minimum and maximum values of estimated and predicted forest AGB values in each sample forest.
No	Name of Forest	No of Plots	Minimum (t ha-1)		Maximum (t ha-1)		Mean (t ha-1)		Standard Deviation
			Estimated AGB	Predicted AGB	Estimated AGB	Predicted AGB	Estimated AGB	Predicted AGB	Estimated AGB	Predicted AGB
1	Zege	15	85.7	111.07	175.8	203.17	130.1	158.26	21.81	28.37
2	Wanzaye	15	3.13	40	94.8	125.62	45.4	73.55	32.13	29.29
3	Alemesaga	20	57.25	62.5	186.68	268.32	121.49	138.8	36.66	48.76
4	Genede Weyen	25	92.71	70.74	154.6	141.92	122.4	110.73	16.12	16.28
5	Gezamen	6	47.9	64.1	105.4	148.98	71.1	115.89	1.61	30.92
6	Keranyo	20	46.12	81.81	105.76	155.21	73.47	122.57	15.69	21.22
7	Tara Gedam	20	21.11	69.38	192.63	264.42	104.43	145.24	46.72	45.07
8	Adet	15	41.97	73.85	98.72	127.31	67.71	103.22	19.61	16.03

4.4. Model validation

The validation of the model was based on comparing the field-estimated forest aboveground biomass (AGB) with the predicted AGB. The forest AGB for 37 test sample plots was calculated using specific predictor variables in SPSS software and compared with their respective estimated forest AGB values. The field-estimated AGB was considered the dependent variable, and the predicted AGB was considered the independent variable. The correlation between the estimated and predicted forest AGB showed a good fit, with a coefficient of determination (R²) of 0.59 and a root mean square error (RMSE) of 38.36 t/ha as shown in Fig. 5 below. This model was then used to map the forest AGB of the study area. Additionally, Puliti et al., (2021) found that Sentinel-2 models had a better fit (Adjusted R² = (0.61–0.64) compared to Landsat models (Adjusted R² = 0.46–0.56). Similarly, Vafaei et al., (2018) discovered that AGB models derived from the combination of Sentinel-2A and ALOS-2 PALSAR-2 data had the highest prediction accuracy using the random forest model with an R² of 0.62 and an RMSE of 43.13.

4.5. Forest Land use and AGB computation

Land use and land cover classification were used to determine the total forest area in the study basin, which is essential for mapping forest aboveground biomass (AGB). The classification results showed that approximately 238,698.2 hectares, or 20.94% of the study basin, were covered by forest ecosystems (see Fig. 6 below). Additionally, agricultural land, grassland, built-up areas, and water bodies covered approximately 633,558.51ha; 216,469.47ha, 38,694.84ha, and 11,651.25ha of the basin, respectively (refer to Fig. 7 in appendix-2). The computed total forest AGB per sample plot was approximately 92.012 tons per hectare, resulting in an estimated total forest AGB of 21,963,098.77 tons for the study basin.

An accuracy assessment was also conducted using a reference dataset to evaluate the classification results. Ground truth data, in polygon format, was obtained from Google satellite maps and used in the Semi-automatic classification tool within the QGIS environment. The overall classification accuracy was determined to be about 97.5%, as indicated in the Table 5 below.

Table 5

Confusion matrix for accuracy assessment.
V_Classified	1	2	3	4	5	Total
1(Forest land)	99192	94	6	192	949	100433
2(Water body)	28	109968	7	0	0	110003
3(Built up)	2	142	2120	84	423	2771
4(Agricultural land)	46	10	224	50782	923	51985
5(Grass land)	28	0	17	595	87492	88132
Total	99296	110214	2374	51653	89787	353324
PA [%]	99.7	83.3	91.3	99.5	91.7	PA [%]
UA [%]	98.7	99.9	76.5	97.6	99.2
Kappa hat	0.98	0.99	0.7	0.9	0.99	0.95

In this study, we used a combination of field-measured data and remote sensing data from Sentinel-2 to estimate the aboveground biomass (AGB) of the forest and create a comprehensive AGB map. Estimating forest AGB using only allometric models presents several challenges and may lead to inaccurate results due to flawed sampling techniques, imprecise measurements, and the complexities of measuring in different types of forest stands. Similarly, using satellite remote sensing data alone to estimate forest AGB is also challenging due to issues such as topography, soil conditions, and forest structure (Lu et al., 2016). To address these challenges, it is recommended to integrate both field-measured data and remote sensing data for biomass estimation. This integration significantly aids in quantifying, monitoring, and understanding forest biomass at various scales ( Li, Qu, & Hao, 2009; Mayamanikandan et al., 2017; Mitchard et al., 2009).

The highest and lowest estimated above-ground biomass (AGB) results were found in Tara Gedam and Wanzaye forest stands, with values of 192.63 and 3.03 t/ha, respectively. These results align with a study by Gedefaw et al.,( 2014), which reported higher forest AGB values of 588.17 t/ha. The disparity in values could be attributed to the dense forest canopy cover and dominance of evergreen and high canopy tree species in the Tara Gedam forest. This is consistent with the findings of Roy et al., (2021), which indicated that the majority of forest biomass falls within the range of 100–200 t/ha, primarily in dense forest regions.

To assess the relationship between estimated AGB and selected predictor variables derived from Sentinel-2 data, a Pearson’s correlation analysis was conducted. The variables included vegetation indices (e.g., NDVI, GNDVI, SAVI, EVI) and biophysical parameters (e.g., LAI, FAPAR, FVC, Cab). The results of the correlation analysis revealed a moderate to strong positive relationship with estimated AGB, as indicated by correlation coefficients ranging from 0.45 to 0.55 for Cab and EVI, respectively. These values align with the suggested correlation coefficient intervals by Xu et al., (2007). EVI exhibited a strong correlation coefficient value with an R2 of 0.55, attributed to its improved sensitivity to high biomass regions and reduced atmospheric sensitivity, as noted by Silleos et al., (2006).

In this study, we observed that three out of the four selected predictor variables had a strong correlation with an R² value of 0.5. These variables are LAI, FAPAR, and FVC. Additionally, two vegetation indices, namely EVI and NDVI, exhibited a correlation with a coefficient of determination (R²) greater than 0.5. This finding aligns with Taddese et al., (2020), who reported a strong correlation between biophysical variables and field-estimated biomass. Based on this, it can be concluded that biophysical variables and surface reflectance attributes serve as better predictor variables than vegetation indices. It is worth noting that Sentinel-2-induced variables have been identified as good predictors of forest AGB in several studies( Chen et al., 2018; Frampton et al., 2013). This could be attributed to the large spatial coverage and high spatial resolution of Sentinel-2-derived products, making biomass assessment more efficient than with other optical sensor data products (Putzenlechner et al., 2019). Variables such as EVI, FAPAR, and LAI with p-values less than 0.005, which exhibit a high coefficient of determination and a high level of significance, were utilized in the development of the biomass prediction model.

A mode was developed using stepwise linear regression analysis, with field-estimated Above Ground Biomass as the dependent variable and Sentinel-2 derived predictive parameters as the independent variables. This model was used to compute various forest AGB values for each sample plot in each forest patch. The highest predicted AGB result was 268.32 t/ha for Alemsaga natural forest, which was 81.64 t/ha higher than the estimated field AGB of 186.68 t/ha. This indicates that using sentinel-2-derived predictor variables improves AGB estimation, which is consistent with previous studies ( Chen et al., 2021; Li et al., 2020). The study shows that sentinel products have the potential for biomass estimation and can improve the accuracy of biomass estimation for national and regional forest inventories and modeling. Additionally, the increase in predicted AGB for Alemsaga forest is attributed to its high level of protection from disturbances, consistent with findings in a study by Muhe & Argaw (2022), which predicted 250 t/ha of biomass in the most inaccessible parts of the forest that are restricted from human activities. Overall, various research findings have shown that reserved forests tend to have high biomass and carbon values (Malunguja & Devi, 2022; Upadhaya et al., 2015).

Accurate estimates of forest above-ground biomass (AGB) are crucial for understanding the net productivity of a forest and for its effective management. AGB serves as the baseline data for assessing carbon stocks in forests. Consequently, the use of remotely sensed data for forest biomass estimation has become increasingly important due to advancements in spectral and spatial resolutions of various sensors. In a particular study, Sentinel-2-derived vegetation indices and biophysical parameters were used to estimate the AGB of a specific forest basin. Field-collected data was also utilized for AGB estimation through the use of an allometric equation. The findings indicated that certain biophysical parameters (FAPAR and LAI) and a vegetation index (EVI) showed a strong correlation with AGB. These variables were considered important for model development in predicting forest AGB. The research demonstrated that using Sentinel-2 derived parameters from optical remote sensing yielded good estimates of forest AGB. However, it also highlighted the need to incorporate RADAR and LiDAR remote sensing data products to obtain even more precise estimates of forest above-ground biomass.

AGB: Above Ground Biomass; RMSE: Root Mean Square Error; RADAR: Radio Detection and Ranging; LiDAR: Light Detection and Ranging; DBH: Diameter at Breast Height; GPS: Global Positioning System; UNFCCC: United Nations Framework Convention on Climate Change, WD: Wood Density; SNAP: Sentinel Application Platform; QGIS: Quantum GIS; UTM: Universal Transverse Mercator; VI: Vegetation Indices; BPV: Biophysical Variables; NDVI: Normalized Difference Vegetation Index; SAVI: Soil-adjusted Vegetation Index, GNDVI: Green-NDVI, EVI: Enhanced Vegetation Indices; LAI: Leaf Area Index; FAPAR: Fraction of absorbed photo synthetically active radiation; FVC: Fractional Vegetation Cover; CSV: Comma Separated Variable; CAB: Leaf Chlorophyll Content; ALOS: Advanced Land Observing Satellite.

Competing interests

The authors declare that, there is no any financial or non-financial competing interest.

Funding

This research didn’t received external funding.

Authors’ contributions

All authors made a valuable contribution. HK designed and wrote the methodology, collected field data and literature, carried out data analysis, and wrote the draft manuscript; MF refines the methodology, supports in collecting additional literature, reviewed, edited, and rewrites the manuscript; and WZ supports in refining methodology and collection of additional literature, coordinated fieldwork and field expanse and refining the manuscript. All authors agreed on the final draft.

Acknowledgement

Note applicable.

Ethics and consent to participate

Note applicable

Consent for publication

Note applicable

Availability of data and materials

The datasets used and analyzed during the current study are available from the corresponding author upon reasonable request.

Aboal, J. R., Arévalo, J. R., & Fernández, Á. (2005). Allometric relationships of different tree species and stand above ground biomass in the Gomera laurel forest (Canary Islands). Flora-Morphology, Distribution, Functional Ecology of Plants, 200(3), 264-274.
Adan, M. S. (2017). Integrating Sentinel-2 derived vegetation indices and terrestrial laser scanner to estimate above-ground biomass/Carbon in Ayer Hitam tropical forest Malaysia. University of Twente,
Addo-Fordjour, P., & Rahmad, Z. B. (2013). Development of allometric equations for estimating above-ground liana biomass in tropical primary and secondary forests, Malaysia. International Journal of Ecology, 2013.
Alamgir, M., & Al-Amin, M. (2008). Allometric models to estimate biomass organic carbon stock in forest vegetation. Journal of forestry research, 19(2), 101.
Ali, A., Ullah, S., Bushra, S., Ahmad, N., Ali, A., & Khan, M. A. Quantifying forest carbon stocks by integrating satellite images and forest inventory data Quantifizierung der Kohlenstoffvorräte in Wäldern durch die Integration von Satellitenbildern und Waldinventurdaten.
Baloloy, A., Blanco, A., Candido, C., Argamosa, R., Dumalag, J., Dimapilis, L., & Paringit, E. (2018). Estimation of mangrove forest aboveground biomass using multispectral bands, vegetation indices and biophysical variables derived from optical satellite imageries: rapideye, planetscope and sentinel-2. ISPRS Annals of Photogrammetry, Remote Sensing & Spatial Information Sciences, 4(3).
Baloloy, A., Blanco, A., Candido, C., Argamosa, R. L., Dumalag, J. C., Dimapilis, L. C., & Paringit, E. (2018). Estimation of mangrove forest aboveground biomass using multispectral bands, vegetation indices and biophysical variables derived from optical satellite imageries: rapideye, planetscope and sentinel-2. ISPRS Annals of Photogrammetry, Remote Sensing & Spatial Information Sciences, 4(3).
Baloloy, A. B., Blanco, A. C., Candido, C. G., Argamosa, R. J. L., Dumalag, J. B. L. C., Dimapilis, L. L. C., & Paringit, E. C. (2018). Estimation of mangrove forest aboveground biomass using multispectral bands, vegetation indices and biophysical variables derived from optical satellite imageries: rapideye, planetscope and sentinel-2. ISPRS Annals of Photogrammetry, Remote Sensing & Spatial Information Sciences, 4(3).
Bonan, G. B. (2008). Forests and climate change: forcings, feedbacks, and the climate benefits of forests. science, 320(5882), 1444-1449.
Brown, S. (2002). Measuring carbon in forests: current status and future challenges. Environmental pollution, 116(3), 363-372.
Brown, S., & Lugo, A. E. (1990). Tropical secondary forests. Journal of tropical ecology, 6(1), 1-32.
Chave, J., Coomes, D., Jansen, S., Lewis, S. L., Swenson, N. G., & Zanne, A. E. (2009). Towards a worldwide wood economics spectrum. Ecology letters, 12(4), 351-366.
Chave, J., Réjou‐Méchain, M., Búrquez, A., Chidumayo, E., Colgan, M. S., Delitti, W. B., . . . Goodman, R. C. (2014). Improved allometric models to estimate the aboveground biomass of tropical trees. Global change biology, 20(10), 3177-3190.
Chen, Y., Guerschman, J., Shendryk, Y., Henry, D., & Harrison, M. T. (2021). Estimating pasture biomass using Sentinel-2 imagery and machine learning. Remote Sensing, 13(4), 603.
Deo, R. K., Russell, M. B., Domke, G. M., Andersen, H.-E., Cohen, W. B., & Woodall, C. W. (2017). Evaluating site-specific and generic spatial models of aboveground forest biomass based on Landsat time-series and LiDAR strip samples in the Eastern USA. Remote Sensing, 9(6), 598.
Dixon, R. K., Solomon, A., Brown, S., Houghton, R., Trexier, M., & Wisniewski, J. (1994). Carbon pools and flux of global forest ecosystems. science, 263(5144), 185-190.
Fantu, W., Nuruddin, A. A., Haris, F. A., & Ab Malik, A. R. (2007). Above-ground biomass allometric equations for selected plantation-grown Eucalyptus species in Ethiopia. Ethiopian Journal of Natural Resources.
FAO. (2010). Global Forest Resources Assessment Main Report. Roma. Food and Agriculture Organization of the United Nations.
Frampton, W. J., Dash, J., Watmough, G., & Milton, E. J. (2013). Evaluating the capabilities of Sentinel-2 for quantitative estimation of biophysical variables in vegetation. ISPRS journal of photogrammetry and remote sensing, 82, 83-92.
Gedefaw, M., Soromessa, T., & Belliethathan, S. (2014). Forest carbon stocks in woody plants of Tara Gedam forest: Implication for climate change mitigation. Science, Technology and Arts Research Journal, 3(1), 101-107.
Gobakken, T., & Næsset, E. (2009). Assessing effects of positioning errors and sample plot size on biophysical stand properties derived from airborne laser scanner data. Canadian Journal of Forest Research, 39(5), 1036-1052.
Hamdan, O., Aziz, H. K., & Hasmadi, I. M. (2014). L-band ALOS PALSAR for biomass estimation of Matang Mangroves, Malaysia. Remote Sensing of Environment, 155, 69-78.
Hamdan, O., Hasmadi, I. M., & Aziz, H. K. (2014a). Combination of SPOT-5 and ALOS PALSAR images in estimating aboveground biomass of lowland Dipterocarp forest. In IOP Conference Series: Earth and Environmental Science (Vol. 18, pp. 012016): IOP Publishing.
Hamdan, O., Hasmadi, I. M., & Aziz, H. K. (2014b). Combination of SPOT-5 and ALOS PALSAR images in estimating aboveground biomass of lowland Dipterocarp forest. Paper presented at the IOP Conference Series: Earth and Environmental Science.
Han, L., Yang, G., Dai, H., Xu, B., Yang, H., Feng, H., . . . Yang, X. (2019). Modeling maize above-ground biomass based on machine learning approaches using UAV remote-sensing data. Plant Methods, 15(1), 1-19.
Hirata, Y., Tabuchi, R., Patanaponpaiboon, P., Poungparn, S., Yoneda, R., & Fujioka, Y. (2014). Estimation of aboveground biomass in mangrove forests using high-resolution satellite data. Journal of Forest Research, 19(1), 34-41.
Houghton, R. A. (2003). Why are estimates of the terrestrial carbon balance so different? Global change biology, 9(4), 500-509.
IPCC, I. (2006). Guidelines for national greenhouse gas inventories. Prepared by the National Greenhouse Gas Inventories Programme. Eggleston HS, Buendia L, Miwa K, Ngara T, Tanabe K, editors. Published: IGES, Japan.
Kaasalainen, S., Holopainen, M., Karjalainen, M., Vastaranta, M., Kankare, V., Karila, K., & Osmanoglu, B. (2015). Combining lidar and synthetic aperture radar data to estimate forest biomass: status and prospects. Forests, 6(1), 252-270.
Kairo, J., Bosire, J., Langat, J., Kirui, B., & Koedam, N. (2009). Allometry and biomass distribution in replanted mangrove plantations at Gazi Bay, Kenya. Aquatic Conservation: Marine and Freshwater Ecosystems, 19(S1), S63-S69.
Kumar, P., Krishna, A. P., Rasmussen, T. M., & Pal, M. K. (2020). An Approach for Fraction of Vegetation Cover Estimation in Forest Above-Ground Biomass Assessment Using Sentinel-2 Images. Paper presented at the International Conference on Computer Vision and Image Processing.
Li, L., Zhou, X., Chen, L., Chen, L., Zhang, Y., & Liu, Y. (2020). Estimating urban vegetation biomass from Sentinel-2A image data. Forests, 11(2), 125.
Li, M., Qu, J. J., & Hao, X. (2009). Estimating aboveground biomass for different forest types based on Landsat TM measurements. Paper presented at the 2009 17th International Conference on Geoinformatics.
Lillesand, T. M., Kiefer, R. W., & Chipman, J. (2000). Remote sensing and image interpretation. John Willey & Sons. New York, 724.
Losi, C. J., Siccama, T. G., Condit, R., & Morales, J. E. (2003). Analysis of alternative methods for estimating carbon stock in young tropical plantations. Forest ecology and management, 184(1-3), 355-368.
Lu, D. (2006). The potential and challenge of remote sensing‐based biomass estimation. International journal of remote sensing, 27(7), 1297-1328.
Lu, D., Chen, Q., Wang, G., Liu, L., Li, G., & Moran, E. (2016). A survey of remote sensing-based aboveground biomass estimation methods in forest ecosystems. International Journal of Digital Earth, 9(1), 63-105.
Malunguja, G. K., & Devi, A. (2022). Quantitative Assessment and Predicting the Effects of Soil Pollutants on Herbaceous Biomass Production in Reserved Forests. Proceedings of the National Academy of Sciences, India Section B: Biological Sciences, 1-16.
Mascaro, J., Detto, M., Asner, G. P., & Muller-Landau, H. C. (2011). Evaluating uncertainty in mapping forest carbon with airborne LiDAR. Remote Sensing of Environment, 115(12), 3770-3774.
Mayamanikandan, T., Jha, C., Das, I., Amminedu, E., & Reddy, C. S. (2017). Forest Biomass Estimation in tropical deciduous forests of Western Ghats using remote sensing data and GIS.
McRoberts, R. E., Tomppo, E. O., & Czaplewski, R. L. (2015). Sampling designs for national forest assessments. Knowledge Reference for National Forest Assessments; FAO: Rome, Italy, 23-40.
Mitchard, E. T., Saatchi, S. S., Woodhouse, I. H., Nangendo, G., Ribeiro, N., Williams, M., . . . Meir, P. (2009). Using satellite radar backscatter to predict above‐ground woody biomass: A consistent relationship across four different African landscapes. Geophysical Research Letters, 36(23).
Muhe, S., & Argaw, M. (2022). Estimation of above-ground biomass in tropical afro-montane forest using Sentinel-2 derived indices. Environmental Systems Research, 11(1), 1-22.
Muller-Wilm, U., Louis, J., Richter, R., Gascon, F., & Niezette, M. (2013, September). Sentinel-2 level 2A prototype processor: Architecture, algorithms and first results. In Proceedings of the ESA Living Planet Symposium, Edinburgh, UK (pp. 9-13).
Navarro, J. A., Algeet, N., Fernández-Landa, A., Esteban, J., Rodríguez-Noriega, P., & Guillén-Climent, M. L. (2019). Integration of UAV, Sentinel-1, and Sentinel-2 data for mangrove plantation aboveground biomass monitoring in Senegal. Remote Sensing, 11(1), 77.
Nuthammachot, N., Phairuang, W., Wicaksono, P., & Sayektiningsih, T. (2018). Estimating aboveground biomass on private forest using Sentinel-2 imagery. Journal of Sensors, 2018.
Pandit, S., Tsuyuki, S., & Dube, T. (2018). Estimating above-ground biomass in sub-tropical buffer zone community forests, Nepal, using Sentinel 2 data. Remote Sensing, 10(4), 601.
Peltier, R., Njiti, C. F., Ntoupka, M., Manlay, R., Henry, M., & Morillon, V. (2007). Evaluation du stock de carbone et de la productivité en bois d'un parc à karités du Nord-Cameroun. BOIS & FORETS DES TROPIQUES, 294(294), 39-50.
Pham, T. D., Yoshino, K., & Bui, D. T. (2017). Biomass estimation of Sonneratia caseolaris (l.) Engler at a coastal area of Hai Phong city (Vietnam) using ALOS-2 PALSAR imagery and GIS-based multi-layer perceptron neural networks. GIScience & Remote Sensing, 54(3), 329-353.
Powell, S. L., Cohen, W. B., Healey, S. P., Kennedy, R. E., Moisen, G. G., Pierce, K. B., & Ohmann, J. L. (2010). Quantification of live aboveground forest biomass dynamics with Landsat time-series and field inventory data: A comparison of empirical modeling approaches. Remote Sensing of Environment, 114(5), 1053-1068.
Putzenlechner, B., Castro, S., Kiese, R., Ludwig, R., Marzahn, P., Sharp, I., & Sanchez-Azofeifa, A. (2019). Validation of Sentinel-2 fAPAR products using ground observations across three forest ecosystems. Remote Sensing of Environment, 232, 111310.
Queen, J. P., Quinn, G. P., & Keough, M. J. (2002). Experimental design and data analysis for biologists: Cambridge university press.
Reyes, G., Brown, S., Chapman, J., & Lugo, A. (1992). Wood densities of tropical tree species. USDA-FS-Southern Forest Experimental Station, New Orleans, Louisiana. General Technical Report SO, 88.
Rossiter, D. G. (2001). Assessing the thematic accuracy of area-class soil maps. Soil Science Division, ITC. Enschede, Holland. Waiting publication.
Roy, S., Mudi, S., Das, P., Ghosh, S., Shit, P. K., Bhunia, G. S., & Kim, J. (2021). Estimating Above Ground Biomass (AGB) and Tree Density using Sentinel-1 Data. In Spatial Modeling in Forest Resources Management (pp. 259-280): Springer.
Segura, M. A., Acuña, L. M., & Andrade, H. J. (2018). ALLOMETRIC MODELS TO ESTIMATE ABOVEGROUND BIOMASS OF SMALL TREES IN WET TROPICAL FORESTS OF COLOMBIAN PACIFIC AREA. Revista Árvore, 42(2).
Silleos, N. G., Alexandridis, T. K., Gitas, I. Z., & Perakis, K. (2006). Vegetation indices: advances made in biomass estimation and vegetation monitoring in the last 30 years. Geocarto International, 21(4), 21-28.
Sola, I., González-Audícana, M., & Álvarez-Mozos, J. (2016). Multi-criteria evaluation of topographic correction methods. Remote Sensing of Environment, 184, 247-262.
Srinet, R., Nandy, S., & Patel, N. (2018). Remote Sensing based Estimation of Forest Biophysical Variables using Machine Learning Algorithm: Friedrich-Schiller-Universität Jena.
Strassburg, B. B., Kelly, A., Balmford, A., Davies, R. G., Gibbs, H. K., Lovett, A., . . . Turner, R. K. (2010). Global congruence of carbon storage and biodiversity in terrestrial ecosystems. Conservation Letters, 3(2), 98-105.
Taddese, H., Asrat, Z., Burud, I., Gobakken, T., Ørka, H. O., Dick, Ø. B., & Næsset, E. (2020). Use of Remotely Sensed Data to Enhance Estimation of Aboveground Biomass for the Dry Afromontane Forest in South-Central Ethiopia. Remote Sensing, 12(20), 3335.
Tao, X., Liang, S., He, T., & Jin, H. (2016). Estimation of fraction of absorbed photosynthetically active radiation from multiple satellite data: Model development and validation. Remote Sensing of Environment, 184, 539-557.
Tetemke, B. A., Birhane, E., Rannestad, M. M., & Eid, T. (2019). Allometric Models for Predicting Aboveground Biomass of Trees in the Dry Afromontane Forests of Northern Ethiopia. Forests, 10(12), 1114.
Théau, J., Lauzier-Hudon, É., Aubé, L., & Devillers, N. (2021). Estimation of forage biomass and vegetation cover in grasslands using UAV imagery. PloS one, 16(1), e0245784.
Torabzadeh, H., Moradi, M., & Fatehi, P. (2019). Estimating aboveground biomass in zagros forest, Iran, using sentinel-2 data. The International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences, 42, 1059-1063.
Upadhaya, K., Thapa, N., & Barik, S. K. (2015). Tree diversity and biomass of tropical forests under two management regimes in Garo hills of north-eastern India. Tropical Ecology, 56(2), 257-268.
Widlowski, J., Pinty, B., Gobron, N., Verstraete, M., Diner, D., & Davis, A. (2004). Canopy structure parameters derived from multi-angular remote sensing data for terrestrial carbon studies. Climatic Change, 67(2-3), 403-415.
Xu, W., Chang, C., Hung, Y. S., Kwan, S., & Fung, P. C. W. (2007). Order statistics correlation coefficient as a novel association measurement with applications to biosignal analysis. IEEE Transactions on Signal Processing, 55(12), 5552-5563.
Zhao, H., Li, Z., Zhou, G., Qiu, Z., & Wu, Z. (2019). Site-Specific Allometric Models for Prediction of Above-and Belowground Biomass of Subtropical Forests in Guangzhou, Southern China. Forests, 10(10), 862.

No competing interests reported.

AppendixAdditionalfiles.xlsx

Download PDF

Editorial decision: Revision requested
08 Aug, 2024
Reviews received at journal
08 Aug, 2024
Reviews received at journal
07 Aug, 2024
Reviews received at journal
06 Aug, 2024
Reviews received at journal
05 Aug, 2024
Reviews received at journal
05 Aug, 2024
Reviewers agreed at journal
05 Aug, 2024
Reviewers agreed at journal
04 Aug, 2024
Reviewers agreed at journal
03 Aug, 2024
Reviewers agreed at journal
02 Aug, 2024
Reviewers agreed at journal
01 Aug, 2024
Reviewers agreed at journal
01 Aug, 2024
Reviewers agreed at journal
01 Aug, 2024
Reviewers agreed at journal
31 Jul, 2024
Reviewers agreed at journal
31 Jul, 2024
Reviewers agreed at journal
30 Jul, 2024
Reviewers agreed at journal
30 Jul, 2024
Reviewers agreed at journal
30 Jul, 2024
Reviewers agreed at journal
30 Jul, 2024
Reviewers agreed at journal
30 Jul, 2024
Reviewers agreed at journal
30 Jul, 2024
Reviewers invited by journal
29 Jul, 2024
Editor assigned by journal
27 Jul, 2024
Submission checks completed at journal
27 Jul, 2024
First submitted to journal
26 Jul, 2024

You are reading this latest preprint version

Above Ground Forest Biomass Estimation Using Sentinel-2 Data in the Upper Blue Nile Basin, North-Western Ethiopia

Status:

Version 1

Abstract

Figures

1. Introduction

2. Materials and Methods

2.1. Description of the Study Area

2.2. Field data collection

2.3. Remote sensing data Preprocessing

3. Data Analysis

3.1. Correlation and Regression Analysis

4. Results

4.1. Field measured Forest Above Ground Biomass Estimation

4.2. Sentinel − 2 predictor variable computation

4.2.1. Correlation Analysis between estimated AGB and Vegetation Indices

4.2.2. Correlation Analysis between estimated AGB and Biophysical parameters

4.3. Forest AGB Prediction Model Development

4.4. Model validation

4.5. Forest Land use and AGB computation

5. Discussion

6. Conclusion

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1