Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Prediction of forest aboveground biomass using an integrated approach of space-based parameters, and forest inventory data

Prediction of forest aboveground biomass using an integrated approach of space-based parameters,... GEOLOGY, ECOLOGY, AND LANDSCAPES INWASCON https://doi.org/10.1080/24749508.2022.2139484 RESEARCH ARTICLE Prediction of forest aboveground biomass using an integrated approach of space-based parameters, and forest inventory data a,b a b b b,c Biswajit Das , Santanu Kumar Patnaik , Reetashree Bordoloi , Ashish Paul and Om Prakash Tripathi a b Department of Geography, Rajiv Gandhi University, Doimukh, India; Department of Forestry, North Eastern Regional Institute of Science and Technology (Deemed to Be University), Nirjuli, India; Department of Environmental Science, Mizoram University, Aizawl, India ABSTRACT ARTICLE HISTORY Received 11 March 2022 Forests contribute significantly in mitigating the effects of climate change by sequestering Accepted 19 October 2022 atmospheric carbon in biomass and soil. To comprehend carbon stock and sequestration, forest degradation, and climate change mitigation, precise calculation biomass is needed. KEYWORDS The present study in Arunachal Pradesh, Northeast India, used Landsat OLI spectral variables, Spectral variables; regression land surface temperature, and field inventory data to estimate aboveground biomass (AGB) analysis; biomass; Northeast and carbon stock in selected forest types. The stepwise multilinear regression model was India; Land use developed using Landsat-derived spectral data, land surface temperature, and soil moisture −1 −1 The stand density varies from 122 individual ha to 833 individual ha . The estimated AGB −1 −1 density varies from 2.64 t ha to 534.21 t ha among the sample plots. The mean land surface temperature was 14.41°C. The predictive integrated model showed that the mean AGB ranged −1 −1 from 9.45 t ha to 330 t ha and dense forest recorded the maximum mean AGB (239.34 t −1 ha ). Akaike information criteria (AIC) and Bayesian information criteria (BIC) were utilised to evaluate the predictive model, and the AIC (176.13) and BIC (184.58) were omparatively lower than other regression models.To evaluate the accuracy of the integrated model (IM), the linear regression was performed between the predicted and observed AGB. The findings of the study will be useful in formulating site-specific suitable management plan. 1. Introduction affected the carbon capture, storage, and retention in The forest ecosystem has an indispensable contribu- forest and soil carbon pools. Further, the landuse land tion to the global carbon (C) cycle and is regarded as cover (LULC) changes induced forest ecosystem has the largest carbon sink, which removes around 25– been modifying the forest structure that increases the 33% of the anthropogenic greenhouse gas (GHG) forest fragmentation, loss of biodiversity, biogeochem- emission from the atmosphere (Le Quéré et al., ical cycle alteration, and hydrological services 2016). Forest stores a considerable amount of forest (Armenteras et al., 2019; Ramachandra & Bharath, carbon (70–90%) in aboveground biomass (AGB) 2019). (S. Chen et al., 2019) and a large portion of atmo- As the Intergovernmental Panel on Climate spheric carbon dioxide (CO ) in terrestrial land and Change (IPCC) Fifth Assessment Report, defores- soil (Motlagh et al., 2018). Besides climate change tation stands second in the list of largest contri- mitigation, the forest nourishes numerous ecosystem butors of carbon dioxide (CO ) emissions (Ciais services and is an indispensable component of earth’s et al., 2014). Deforestation and land degradation energy cycle. The forest biomass is a crucial parameter account for 20–25% of anthropogenic carbon to describe the structure and function of the forest emissions (Pachauri & Reisinger, 2007; ecosystem (Y. Li et al., 2020). Forest covert the atmo- Ramachandra & Bharath, 2019) hence influencing spheric carbon and is stored in AGB, belowground regional climate patterns including hydrogeologi- biomass, dead organic matter, and soil organic matter cal regime alternation. The forest cover loss has (Ramachandra & Bharath, 2020). The forest ecosystem altered the local rainfall regime due to alternation captures about 40% of terrestrial carbon and 50% of in the pattern of thermodynamic and mesoscale net ecosystem productivity (McGarvey et al., 2015). circulation processes (Lawrence & Vandecar, Annually, about 30% of the global anthropogenic CO 2015), which have extreme weather consequences −1 emissions (2 Pg C yr ) from the atmosphere seques- afterwards. Further, the large and intact forest tered by the forest and soil (Achat et al., 2006; Lal, through their leaves, leaf area and canopy cover- 2005). The extent of land use change, anthropogenic age transferring sensible heat to latent heat, which pressure, disturbances, and climate change have alter the wind dynamics and hence rainfall events CONTACT Biswajit Das biswajitdas.19.1989@gmail.com Department of Geography, Rajiv Gandhi University, Rono Hills, Doimukh, Arunachal Pradesh, India © 2022 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group on behalf of the International Water, Air & Soil Conservation Society(INWASCON). This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 2 B. DAS ET AL. increased (Ramachandra & Bharath, 2020). The now been widely preferred in AGB estimation (Jiang deforestation causes lower evapotranspiration et al., 2021; C. Li et al., 2019; Lu, 2006; Ravindranath across the region, which has the consequence of et al., 2008; Tang et al., 2016). Several researcher have delayed rainfall with low periodicity of rainfall used data like MODIS (Pandey et al., 2019), Landsat TM with longer dry conditions (Debortoli et al., (Deka et al., 2015; Günlü et al., 2014), Landsat ETM+ 2017). The deforestation may lead to biophysical (Shen et al., 2020; Zheng et al., 2004), Landsat OLI alternation in plants through altered micro- (Bordoloi et al., 2022; Kashung et al., 2018), IKONOS climatic condition along with raising air and land (Thenkabail et al., 2004), WorldView-2 (Eckert, 2012), surface temperature (LST) (Ramachandra et al., RADAR (Gunawardena et al., 2015), Sentinel-1A (Y. Li 2018). The increase in temperature has an upward et al., 2020; Norovsuren et al., 2020) to study the for- trend in plant water demand leading to loss of est AGB. canopy, microflora, microclimate change as well Many authors have reported the significance of crucial as an increase in the number of forest fire variables to develop a model that estimates the forest (Ramachandra & Bharath, 2019). AGB (S. Chen et al., 2019; Yu et al., 2019). Normalized The forest biomass combustion and dead plant difference vegetation index (NDVI), Soil-adjusted vege- material decomposition due to deforestation and tation index (SAVI) and Enhanced vegetation index forest degradation have accelerated the atmo- (EVI), Atmospheric resistant vegetation index (ARVI) spheric GHG emission particularly in developing were most widely used approaches (Bordoloi et al., countries (Pearson et al., 2017; Ramankutty et al., 2022; Kashung et al., 2018; Shen et al., 2020). Besides 2007). The terrestrial carbon cycle and carbon above, the digital elevation model (DEM) derived topo- sequestration during the process of plant growth graphic variables like slope, aspect can also be utilized as remains least affected aspect of the global carbon variables for AGB estimation model. Some researchers cycle (Le Quéré et al., 2018). Reducing emissions have used environmental parameters like temperature, from deforestation and degradation (REDD) land surface temperature, soil moisture in forest para- devised mechanisms to implement the forest mon- meter estimation (Jiang et al., 2021; Zarco-Tejada et al., itoring system that needs primarily the forest car- 2018). The satellite-derived land surface temperature bon stock estimation (GOFC-GOLD, 2014; from the thermal infrared sensor (TIRS) of Landsat UNFCCC, 2009; UN-REDD, 2011) through inte- improves the accuracy in estimating leaf area index grated approach of field-based data and earth (LAI) (Neinavaz et al., 2019) while its use in AGB estima- observatory data. Besides forest degradation and tion needs to be evaluated. The use of large number of LULC change, various terrain parameters (slope, variables can be time-consuming, estimation erroneous aspect and elevation) also have significant influ - and render the applicability and interpretability of the ence on forest composition, micro climate, solar model (Yu et al., 2019). Hence, the suitable variables radiation and moisture condition at different hill combination selection is crucial for the AGB estimation. slopes of mountainous forests (Holland & Steyn, Numerous algorithms have been used by the researcher, 1975; Måren et al., 2015; Sharma et al., 2010; among them classical statistical linear regression algo- Yadav & Gupta, 2006). The forest carbon stock rithm was the most widely used algorithm due its simpli- estimation primarily depends on various carbon city nature (C. Li et al., 2019; Lu et al., 2014; Sarker & pool such as AGB, BGB, litter biomass, deadwood Nichol, 2011). The linear relationship analysis used RS and soil organic carbon (SOC) (IPCC, 2006). based spectral variables and the field inventory data. However, the AGB alone has contributed about Hence a stepwise multi-linear regression approach was 50% to the carbon stock (Goetz & Dubayah, 2011). used in the present study. The objective of the study is to There are two major approaches to estimate the forest estimate and predict the biomass stock of different land- AGB viz. destructive and non-destructive approach. The use sectors of the region in connection to climate change former is the most accurate and reliable method of AGB scenarios. The study will assess the region’s biomass stock estimation. However, it fails when there was data belong- using in situ measurements from sample plots and pixel- ing to large area coverage (Qureshi et al., 2012; Stovall based RS derived spectral variables, land surface tempera- et al., 2017). Therefore, the later is the most accepted ture and soil moisture to simulate the spatial biomass approach of AGB estimation using allometric equations density. (Brown & Lugo, 1992; Chave et al., 2005, 2014). In the last two decades, the remote sensing (RS) technology became the most preferred approach and enables researchers to 2. Materials and methods get large scale real-time synoptic view of vegetation con- 2.1 Study area ditions (Du et al., 2014; Tang et al., 2016). However, to estimate the more precise regional AGB, the RS technol- The present study was carried out in four western ogy depends on fine scale filed inventory data. Therefore, districts of Arunachal Pradesh (West Kameng, East integrating field inventory data and RS technology has Kameng, Papum pare and Lower Subansiri) which GEOLOGY, ECOLOGY, AND LANDSCAPES 3 were spread in the mountainous forest of eastern applied the imageries by following the radiometric Himalayas. The study region spreads over an area of correction method as per Landsat 8 user handbook 17,384 km and lies 26°52’ to 29°59’ N latitude and 92° (Zanter, 2019). Additionally, because of the pre- 01’ to 94°22’ E longitude. The climate of the state sence of complex terrain conditions along with varies sharply with changes in latitude and elevation. dense forest cover in the region, it was precau- The areas at very high elevation in the upper tionary to apply topographic illumination correc- Himalayas close to the Tibetan border enjoy an alpine tion, hence an improved cosine correction method climate while below this region it is temperate. The (Civco, 1989) was applied. The ASTER DEM data areas at the sub-Himalayan experience humid subtro- with 30 m resolution was used for topographic pical climate along with hot summers and mild win- correction. The DEM images were re-sampled to ters. The average mean maximum and minimum Landsat OLI pixel size by the nearest neighbour- temperature varied between 29.5°C and 17.7°C in sub- hood transformation (Wu et al., 2016). For the tropical regions and 21.4°C and 2.4°C in cold humid forest inventory process, unsupervised classifica - regions and the foothills and plains experience higher tion approach was done. The satellite image of temperatures. The area is rich in biodiversity includ- the study area categorized into nine land use ing the treasure of germplasm for species of food, land cover categories viz. dense forest, moderately medicinal, bamboo and cane, orchids, wild animal, dense forest, open forest, plantation, agriculture and other life forms. As there was a paucity of data land, settlement, sandy area, snow areas, and on carbon sequestration in the state, it became crucial water bodies (Figure 2). to carry out such studies to estimate the AGB, carbon pool and carbon sequestration of the region (Figure 1). 2.3 Forest inventory data collection The selection of permanent sample plot was based on 2.2 Satellite data acquisition forest density class. To find out the different forest In the present study, cloud-free Landsat OLI opti- density classes, the classified land use map was used, cal RS data for the year 2018 and 2019 and and it showed dense forest, moderately dense forest, Advanced space-borne thermal emission radio- open forest, plantation among the major land use meter (ASTER) elevation data were selected and sectors in major agro-climatic zones. In each of the downloaded from Earth Explorer (https://earthex zones and major land use sectors, sampling plots were plorer.usgs.gov). In Landsat OLI data, the blue, randomly selected in replicates. The forest inventory green, red, near-infrared band, and thermal infra- was done by establishing 61 number of sample plot of red band were used in the analysis. All the ima- 31.6 m x 31.6 m (0.1 ha) as per standard layout. The geries were co-registered to a common UTM sample quadrats were chosen on the basis of stratified (Universal Transversal Mercator) projection sys- random sampling approach and representative sites tem with WGS 84 datum (Zone 46). The top of covering dense to sparse vegetation cover. Tree struc- atmospheric (TOA) reflectance correction was tural parameters like species, girth at breast height Figure 1. Map shows the location of Sample plot in the study area. 4 B. DAS ET AL. Figure 2. Map shows the Land use land cover of the study area. (GBH), height of the all the trees ≥ 30 cm girth at 2.4 Satellite-based spectral variables breast height (at 1.37 m from the base) were measured. The Landsat OLI spectral variables (SV) (Table 1) The measuring tape, clinometer was used to measure include difference vegetation index (DVI) which is the GBH and the height of plant species respectively. a simple difference between near-infrared (NIR) Further, GPS records were also taken for each plot. and red (R) band ratio, normalized difference Forest inventory data were collected for a period of vegetation index (NDVI) which is the ratio of two years (2018–2019) and among 61 permanent NIR and R bands, atmospheric resistant vegetation plots, 14 number of plots sampled in dense forest indices (ARVI) which uses three-band ratio to (DF) and moderately dense forest (MDF) each, 20 minimize the atmospheric scattering effect. Soil plots in open forest (OF) and 13 plots in plantation adjusted vegetation index (SAVI) minimizes soil (PL). Based on the tree DBH and height, the plot based brightness effects in the satellite image, and AGB was calculated for trees having girth ≥ 30 cm at enhanced vegetation index (EVI) is the complex girth breast height (1.37 m) by applying the general- index and high sensitivity to high AGB region as ized allometric equation for the north-eastern region it eliminates the aerosol influence and canopy of India proposed by Nath et al. (2019), which is based background effect. Hence, these vegetation indices on stem diameter, height, and wood density of the were incorporated to analyze the best relationship species. between the AGB and vegetation indices. Table 1. Spectral variables used spatial mapping of biomass and carbon. Vegetation Indices Expression Author NDVI NDVI ¼ ðNIR RÞ=ðNIRþ RÞ Rouse et al. (1974) SAVI SAVI ¼ ½ðNIR RÞ=ðNIRþ Rþ LÞ�ð1þ LÞWhere L = 0.5 for intermediated vegetation cover Huete (1988) ARVI ARVI ¼ ðNIR RBÞ=ðNIRþ RBÞ Kaufman & Tanre, 1992 EVI EVI ¼ G� ðNIR RÞ=ðNIRþ C R C Bþ LÞWhere: C = 6; C = 7.5; L = 1; G = 2.5 Liu and Huete (1995) 1 2 1 2 GEOLOGY, ECOLOGY, AND LANDSCAPES 5 crucial variables. To determine the dry and wet edge 2.5 Forest biomass and carbon calculation from triangle method, for each dry and wet edge, the The non-destructive approach was used to estimate maximum and lowest LST values were chosen from the forest AGB. Though the Forest Survey of India the pixel-based NDVI. As per J. Chen et al. (2011), the (FSI), have developed many species-specific allometric pixels having NDVI between 0.2–0.3 were chosen for equations to estimate the tree AGB (FSI, 1996), how- dry edge calculation. The equation for TDVI is as ever, there was a paucity of the allometric equation for follows. many local species available in the study area. Hence, the generalized allometric equation developed by Nath LST LST min TDVI ¼ et al. (2019) for the northeast India region was used in LST LST max min the present study. To estimate forest carbon stock Where: a conversion factor of 0.55 was used (MacDicken, LST is the minimum surface temperature for min 1997). given pixel for wet edge and modelled using equation (LST = a + b *NDVI) in the triangle, where min min min a and b are regression parameters for the wet min min 2.6 Satellite-based LST and soil moisture edge (Table 2). estimation LST is the maximum surface temperature for max The cloud-free Landsat 8 satellite imageries of the year given pixel for wet edge and modelled using equation 2018 to 2019 covering the study area were acquired (LST = a + b *NDVI) in the triangle, where max max max from the Earth explorer. The Landsat 8 satellite has on a and b are regression parameters for the dry max max boarded two different sensors, the operational land edge (Table 2). The linear regression analysis was imager (OLI) and the Thermal infrared sensor carried out between the plot-based volumetric moist- (TIRS) having different resolution. The T1-level ima- ure content and the plot-based TDVI values to get the geries having the radiometric correction and geo- best fit model for S estimation of the study area. metric correction were acquired. However, two prerequisite processes, the radiometric calibration and atmospheric correction were applied as per 2.7 Development of forest AGB model and Landsat 8 manual (Zanter, 2019) to the all the ima- evaluation geries to get high quality and accurate image informa- tion. The automated algorithm imageries (Avdan & The RS-based SVs, and plot based AGB were taken Jovanovska, 2016; Sobrino et al., 2004) was used in into account to model the forest AGB of the study LST retrieval from Landsat 8. Among the available area. In the present study, the satellite data derived bands, Band 10 (TIR), Band 4 (RED), and Band 5 spectral variables like NDVI, SAVI, ARVI, EVI and (NIR) were used in LST retrieval. The TIR band was physical variables like LST and S were evaluated to used to derive brightness temperature, whereas the find out their linear relationships with plot based AGB RED and NIR band was used in NDVI determination. of each land use sector (Table 2). The sampled plot’s The brightness temperature (BT) was obtained from locations were overlaid on different SVs, gridded LST, the TIRS band using the thermal constant provided in and S data to derive corresponding values of each the image metadata file. Following the NDVI, the land sampled plot. The four most widely used vegetation surface emissivity (LSE) was calculated, which was the indices (VIs) associated with RS based change detec- most essential aspect of LST retrieval. tion and biomass estimation were used. The tested VIs For the global climate change and large-scale includes NDVI, which was the ratio of complementary hydrological cycles, the soil moisture (S ) is regarded reflectance between the red and NIR spectral bands as important influencing factors (J. Chen et al., 2011). due to chlorophyll pigments and leaf cellular structure The retrieval of S was done by using the temperature respectively (Powell et al., 2010), the SAVI, although vegetation dryness index (TVDI). The TVDI derived similar to the NDVI but have an improvement in by using triangle method (Przeździecki et al., 2017; terms of the soil brightness correction factor (Huete, Sandholt et al., 2002), which utilized the NDVI and 1988) and the ARVI, which was self-correcting indices LST as influencing factor for S retrieval. To deter- for both soil and atmosphere reflectance and mini- mine TVDI, dry edge (low evapotranspiration) and mizes both soil and atmospheric noises (Kaufman & wet edge (maximum evapotranspiration) are the two Tanre, 1992). Table 2. TDVI retrieval parameter for Soil moisture retrieval. Parameter Slope Intercept Dry Edge a = −53.73 b = 35.56 min min Wet Edge a = 6.96 b = 10.034 max max 6 B. DAS ET AL. As the current study area have complex land- among the sample plots. Among the forest types, form feature along with different agro-ecological plantation has recorded maximum tree density of −1 zone, it is quite difficult to characterize the precise 609 individual ha followed by the dense forest (563 −1 estimation of AGB. Therefore, a statistical correla- individual ha ), moderately dense forest (517 indivi- −1 −1 tion was developed between LST, S and VIs with dual ha ) and open forest (343 individual ha ). The m, the field-based AGB of each land-use sector. The AGB density varies among the different forest types, −1 −1 plot based AGB was assessed against the satellite and it ranges from 2.64 t ha to 534.21 t ha among derived pixel based selected variables of the the sample plots. The dense forest have recorded the −1 sampled location. To assess the same, the linear maximum AGB of 279.41 t ha followed by the mod- −1 relationship between the variables and AGB was erately dense forest (161.16 t ha ), plantation (88.99 t −1 −1 performed to get best relationship of the model. ha ) and open forest (61.86 t ha ) (Table 3). A multiple regression analysis approach was used for modelling the AGB, which was further 3.2 LST and Soil moisture estimation validated. For the AGB modelling, researchers have used numer- The mean LST for the study area was 14.41°C, how- ous approaches like linear regression approach (Calvao & ever it varied among the different forest types. The Palmeirim, 2004; Ren & Zhou, 2014), multi-regression mean LST observed for DF was 10.19°C, MDF (14.34° approach (Askar et al., 2018; Eckert, 2012; Sarker & C), OF (16.16°C), and PL (16.32°C). The S of the Nichol, 2011), non-linear regression (Santos et al., study area was estimated based on regression model 2003), non-parametric model (Urbazaev et al., 2016), developed between TDVI and field based volumetric artificial neural network (Dong et al., 2020). In the pre- soil moisture content of the sample plot. The devel- sent study, we have applied stepwise multi-linear regres- oped model (S = −103.31*TDVI+147.38, R = 0.70) sion analysis for assessing the relationship between AGB, showed that the mean S of the study area was 30.64%, spectral variables, LST and S . The multi-linear regres- however it ranged between 1.41 and 84.56% (Figure 3). sion approach some time may lead to multicollinearity and overfitting in analysis hence to overcome this most widely used statistical parameter like Pearson correlation 3.3 Optimum variable selection (r), root mean square error (RMSE), coefficient of deter- The statistical analysis of individual SVs with the AGB mination (R ) and p – level were applied. The best fit was done by linear regression analysis and it was model for AGB prediction was done based on high r and evident that the correlation coefficient (R ) obtained high R values. A step wise linear regression method was between NDVI and AGB was 0.60 with RMSE of employed for modelling AGB using Origin software. −1 2 36.50 t ha . Similarly, for the SAVI (R = 0.61, Further to confirm the reliability of the model, R , −1 2 RMSE = 35.52 t ha ), ARVI (R = 0.65, RMSE = RMSE, multicollinearity of variables, the variance infla - −1 2 32.37 t ha ), and for EVI (R = 0.60, RMSE = tion factor {VIF = VIFj = 1/ (1-R )} were determined. To −1 36.53 t ha ). However, the linear regression analysis indicate multicollinearity problem, VIF value greater of LST with AGB showed R = 0.61, RMSE = 35.97 t than 10 (Askar et al., 2018; Sarker & Nichol, 2011) were −1 2 ha and that of S with the AGB resulted R = 0.62, determined. Further, Akaike information criteria (AIC; −1 RMSE = 34.63 t ha (Table 4). Akaike, 1974) and Bayesian information criteria (BIC; Aho et al., 2014) were used in the suitable model selection process. The selection of model was done based on 3.4 Modelling of AGB observed smallest AIC and BIC values. The stepwise multi-linear regression analysis was used to select appropriate variables for predictive AGB 3. Result model development. This was done by anlysing the resultant R , RMSE and VIF observed from linear 3.1 Tree density, aboveground biomass, and relationship analysis individual variables with AGB. carbon stock The selection of the variable’s combination was From the study, it was observed that the mean density based on the minimum RMSE of the model. The −1 −1 varies from 122 individual ha to 833 individual ha statistical relationship of individual SVs, then Table 3. Observed AGB of the sample plot in the study area. Forest Type Plot number Mean density Mean AGB Range Standard Deviation DF 14 563 279.41 156.06–534.21 107.84 MDF 14 517 161.16 96.56–235.69 43.99 OF 20 343 61.86 2.64–162.70 43.15 PL 13 609 88.99 21.07–221-62 58.67 Total 61 482 147.85 2.64–534.21 107.03 GEOLOGY, ECOLOGY, AND LANDSCAPES 7 Figure 3. Land surface temperature and soil moisture of the study area. Table 4. Statistical analysis of stepwise multilinear regression of AGB with variables. Variables R r RMSE p-value VIF NDVI 0.60 0.77 36.51 0.00 2.52 SAVI 0.61 0.78 35.52 0.00 2.59 ARVI 0.64 0.80 32.36 0.0002 2.84 EVI 0.60 0.77 36.53 0.00 2.52 LST 0.61 0.78 35.97 0.00 2.56 S 0.62 0.79 34.63 0.00 2.66 Integrated Model (IM) 0.82 0.90 17.09 0.00 5.67 IM = SAVI + ARVI + LST + S −1 combining all the SVs, and finally combining the LST predicted AGB ranged between 9.45 t ha and 330 t −1 and S with the SVs is presented in Table 4. ha . The DF have maximum mean predicted AGB −1 The linear regression analysis of field based AGB density (239.34 t ha ) followed by MDF (160.49 t −1 −1 −1 and individual variables of the study area is shown in ha ), PL (88.47 t ha ) and OF (62.29 t ha ). Table 4. The correlation coefficient R ranged between Further AIC and BIC analysis validated the resultant 0.60 and 0.64 whereas the Pearson’s r varied between model by showing the lower AIC (176.13) and BIC 0.77 and 0.80. Among the selected SVs, the ARVI have (184.58) value (Table 5). The AGB density map was shown better correlation (r = 0.80, R = 0.64). It was prepared by considering the predicted AGB of the evident that all the selected variables have shown sig- study area. nificant and positive correlation with AGB. Further, the predictive AGB model derived from stepwise 4. Discussion multi- linear regression technique has been expressed as follows: The current study focused on an integrated approach that used freely available optical satellite data-derived AGB ¼ 23:71� SAVI þ 15:62� ARVI 0:72� LST SVs, LST, S , field inventory measurements, and an þ 0:34� SM 4:23 empirical modelling approach to predict the spatial The developed predictive model derived using SVs, AGB of the study region. Although the satellite data- LST, and S have shown R = 0.82, p < 0.05. The based variables have a significant contribution to the −1 RMSE of the model was observed to be 17.09 t ha . carbon stock modelling process, however, there were certain factors which influenced the satellite data and the data acquisition process that further influence the 3.5 Spatial AGB and carbon density mapping carbon stock modelling. The study region have very To evaluate the accuracy of the integrated model (IM), complex landform types ranging from flat to undulat- the linear regression analysis was performed between ing terrain, as well as different land use sectors ranging the predicted AGB and observed AGB. It was observed from grassland to different forest density class, and that there was strong relationship between field based there surface reflectance is extremely complex, limit- observed AGB and predicted AGB and gave a strong ing the predictive model’s competence. Further, the coefficient of determination (R = 0.82, RMSE = use of multi-date satellite data due to unavailability of −1 13.43 t ha ) (Figure 4). The AGB and carbon density cloud free satellite data and have different sun and map was produced from the developed model between sensor arrangement at the time data acquisition pro- AGB, satellite derived spectral variables, LST and S cess also doubling the effect on capability of carbon (Figure 5). From the study, it was predicted that, the stock process. The variables selection and their 8 B. DAS ET AL. y = 0.8257x + 2.4535 R² = 0.8239 45.00 40.00 35.00 30.00 25.00 20.00 15.00 10.00 5.00 0.00 0.00 10.00 20.00 30.00 40.00 50.00 60.00 Observed AGB( Tonnes/0.1 ha) Figure 4. Relationship between Predicted AGB and Observed AGB. Table 5. AIC for the developed regression analysis of AGB with variables. Variables NDVI SAVI ARVI EVI LST SM IM AIC 219.41 217.74 212.07 219.45 218.51 216.22 176.13 BIC 221.52 219.86 214.19 221.56 220.62 218.33 184.58 IM = integrated Model characteristics were crucial in AGB predictive model variables. Presently, the available variables for the and accuracy of the predictive model is solely depen- AGB predictive models includes terrain factor, envir- dent on the inherent characteristics of selected onmental factors, textural feature and SVs (Lu et al., Figure 5. Predicted AGB density map of the study area. Predicted AGB (Tonnes/0.1 ha) GEOLOGY, ECOLOGY, AND LANDSCAPES 9 2014; Zarco-Tejada et al., 2018). The spectral variables reported by Haripriya (2002) for the Indian forest. include vegetation indices, which were most preferred Y. Li et al. (2020) reported that the mean AGB in −1 variables for AGB modelling. The incorporation tex- range of 16.76 to 173.72 t ha for the subtropical ture along with terrain factors for some specific area forest of Chenzhou City of China. The mean AGB −1 significantly enhances the accuracy of the AGB model predicted for plantation was 88.47 t ha . Devagiri (Gleason & Im, 2011). Further, the different forest et al. (2013) estimated comparably slightly higher. cover type has different LST performance, which influ - The mean predicted AGB for the tropical plantations −1 ences the vegetation growth (Neinavaz et al., 2019). (185.95 t ha ) showed higher mean AGB (118.19 t −1 The present predictive model used the Landsat OLI ha ) for teak plantation of Karnataka. Similarly, Sahu data derived SVs, LST, and S . Among the used SVs, et al. (2016) had also reported higher AGB of 181.2 t −1 although the NDVI was the most preferred VI in the ha for Mangrove plantation of Odisha. present study, however, it showed a comparably lower correlation with the field-based AGB, this might be 5. Pros and cons attributed due to the high saturation problem at high AGB stock areas. The same was reported that the The incorporation of numerous variables and fac- saturation was caused due to high reflectance in NIR tors in predictive AGB model may lead to uncer- in comparison to low reflectance in red band, which tainty and caused estimation error. The selected could result inferior correlation (Askar et al., 2018). variables significantly influence the estimation Due to the correction in soil brightness factor, the model performances (S. Chen et al., 2019). Beside SAVI minimized the soil reflectance over the canopy this, the sensor type, approach of sampling strategy reflectance properties, and have showed better perfor- and estimation model (Lu et al., 2014) were also mance than the NDVI (Huete et al., 1994; Vidhya major factor for AGB estimation model. Presently, et al., 2014). Among the selected SVs, the ARVI mini- optical data like Sntinel-2, Landsat-8, and SPOT mizes the atmospheric aerosols brightness effect of the images have major contribution in large scale earth’s surface feature reflectance properties, and AGB estimation (Y. Chen et al., 2019; Lu et al., hence have the better correlation (R = 0.64) with the 2014; Zhou et al., 2020). Among all the optical AGB, than the other SVs (Huete & Liu, 1994; Tanré data, Landsat series have been providing data with et al., 1990). medium resolution with great revisit time. The The multi-linear regression approach using the Landsat-8 onboard TIRS sensor effectively used in SVs, LST and S was applied to AGB model and LST retrieval, which significantly enhances the AGB estimation. The incorporation of the LST and accuracy of AGB model (Gleason & Im, 2011; S in the predictive model showed the improvement Kim et al., 2014). However, optical data fails to in accuracy, which was further evaluated by the provide the canopy information and have higher decrease in RMSE (Table 4). The performance of the saturation mainly in area having dense forest predictive model further assessed through the AIC cover (Zhao et al., 2016; Zhu & Liu, 2015). The value obtained from the statistical analysis of selected researcher like Tian et al. (2019) and Cao et al. variables (Table 4). The AIC (Bordoloi et al., 2022) (2018) have successfully used Airborne laser scan- and BIC (Yu et al., 2019) index showed the better ning (ALS) and terrestrial laser scanning (TLS) in competence in linear indices for the predictive AGB small scale AGB estimation with excellent accuracy. model. Hence, both AIC and BIC tested in the present However, the inherent character of the sensors and study and observed lower AIC and BIC value for the resolution makes it, difficult in large scale study. IM among the individual model using each variable, The linear approaches are quite simple and can can be compared with AIC and BIC value of above depict the relationships smoothly among the selected study. Hence, from the present analysis, it was variables; however, it fails in reflecting the relation- observed that using satellite data derived VIs as pre- ships between the AGB and selected variables under dictor variable quite helpful in AGB prediction due to complex forest conditions (S. Chen et al., 2019). their sensitivity of minimizing the terrain, atmo- Further, the elimination of collinearity among the sphere, and soil brightness effect on surface reflectance variables are prerequisite to ensure the significance properties (Lu et al., 2014). The RMSE thus obtained and interpretability of variables, as these may lead to using Landsat data from the study can be compared deletion of highly related variables with AGB (S. Chen with the value reported by (B. Li et al., 2018). et al., 2019). The mean AGB predicted from the present study −1 was 131.04 t ha which can be compared with the 6. Conclusion findings of studies that used Landsat OLI for the different region (B. Li et al., 2018; López-Serrano To study the regional and global carbon cycle mon- et al., 2020; Suhardiman et al., 2018). The mean AGB itoring, forest dynamic management, the mapping of −1 can also be compared with mean AGB (92 t ha ) AGB spatial distribution among the different forest 10 B. DAS ET AL. types are crucial. The present study mainly empha- Bordoloi, R., Das, B., Tripathi, O. P., Sahoo, U. K., Nath, A. J., Deb, S., Das, D. J., Gupta, A., Devi, N. B., sized the utilization of integrated approach which Charturvedi, S. S., Tiwari, B. K., Paul, A., & Tajo, L. combined the non-destructive, satellite data derived (2022). Satellite based integrated approaches to modelling SVs, LST, S and empirical modelling algorithms to spatial carbon stock and carbon sequestration potential of predict the spatial distribution of AGB of the Study different land uses of Northeast India. Environmental and area. From the study, it was evident that the using of Sustainability Indicators, 13, 100166. https://doi.org/10. 1016/j.indic.2021.100166 optical satellite derived SVs, LST and S as dependant Brown, S., & Lugo, A. E. (1992). Aboveground biomass variables smoothen the AGB prediction process of the estimates for tropical moist forests of the Brazilian study area. Further, the use of linear method to select Amazon. Interciencia. Caracas, 17(1), 8–18. http://phi the optimum variables may render the accuracy of the lip.inpa.gov.br/publ_livres/Other%20side-outro%20lado/ AGB model. Hence, non-linear method may be used Brown%20%26%20Lugo%20biomass/Brown%20and% to get better performance. Besides, optical data, the use 20Lugo%201992-Aboveground%20biomass%20esti mates.pdf SAR, LiDAR satellite data will have an advantage in Calvao, T., & Palmeirim, J. M. (2004). Mapping mediterra- determining forest structural parameter than the opti- nean scrub with satellite imagery: Biomass estimation and cal data, and hence have the better capability to predict spectral behaviour. International Journal of Remote the AGB. The use of deep learning techniques which Sensing, 25(16), 3113–3126. https://doi.org/10.1080/ includes artificial neural network, fuzzy logic, and Cao, L., Pan, J., Li, R., Li, J., & Li, Z. (2018). Integrating cellular automata, also has better prospects to model airborne LiDAR and optical data to estimate forest above- the AGB. The proposed integrated approach will have ground biomass in arid and semi-arid regions of China. significant role for the planner and decision makers to Remote Sensing, 10(4), 532. https://doi.org/10.3390/ achieve the carbon stock management along with their rs10040532 climate change mitigation at the regional level as well Chave, J., Andalo, C., Brown, S., Cairns, M. A., as national and global level. Chambers, J. Q., Eamus, D., Foster, H., Fromard, F., Higuchi, N., Kira, T., Lescure, J. P., Nelson, B. W., Ogawa, H., Puig, H., Riera, B., & Yamakura, T. (2005). Tree allometry and improved estimation of carbon stocks Disclosure statement and balance in tropical forests. Oecologia, 145(1), 87–99. https://doi.org/10.1007/s00442-005-0100-x No potential conflict of interest was reported by the Chave, J., Réjou-Méchain, M., Búrquez, A., Chidumayo, E., author(s). Colgan, M. S., Delitti, W. B., Vieilledent, G., Eid, T., Fearnside, P. M., Goodman, R. C., Henry, M., Martínez- Yrízar, A., Mugasha, W. A., Muller-Landau, H. C., ORCID Mencuccini, M., Nelson, B. W., Ngomanda, A., Nogueira, E. M., Ortiz-Malavassi, E., . . . Vieilledent, G. Biswajit Das http://orcid.org/0000-0003-0016-0082 (2014). Improved allometric models to estimate the aboveground biomass of tropical trees. Global Change Biology, 20(10), 3177–3190. https://doi.org/10.1111/gcb. References Achat, D. L., Fortin, M., Landmann, G., Ringeval, B., & Chen, S., Feng, Z., Chen, P., Ullah Khan, T., & Lian, Y. Augusto, L. (2015). Forest soil carbon is threatened by (2019). Non destructive estimation of the above-ground biomass of multiple tree species in boreal forests of China intensive biomass harvesting. Scientific Reports, 5(1), using terrestrial laser scanning. Forests, 10(11), 936. 15991. https://doi.org/10.1038/srep15991 https://doi.org/10.3390/f10110936 Aho, K., DErryberry, D., & Peterson, T. (2014). Model selec- Chen, Y., Li, L., Lu, D., & Li, D. (2019). Exploring bamboo tion for ecologist: The worldviews of AIC and BIC. Ecology, forest aboveground biomass estimation using Sentinel-2 95(3), 631–636. https://doi.org/10.1890/13-1452.1 data. Remote Sensing , 11(1), 7. https://doi.org/10.3390/ Akaike, H. (1974). A new look at the statistical model rs11010007 identification. EEE Transactions on Automatic Control, Chen, J., Wang, C., Jiang, H., Mao, L., & Yu, Z. (2011). 19(6), 716–723. https://doi.org/10.1109/TAC.1974. Estimating soil moisture using Temperature/Vegetation Dryness Index (TVDI) in the Huang-Huai-Hai (HHH) Armenteras, D., Murcia, U., Gonza´ Lez, T. M., Baro´ plain. International Journal of Remote Sensing, 32(4), N, O. J., & Arias, J. E. (2019). Scenarios of land use and 1165–1177. https://doi.org/10.1080/01431160903527421 land cover change for NW Amazonia: Impact on forest Ciais, P., Sabine, C., Bala, G., Bopp, L., Brovkin, V., intactness. Global Ecology and Conservation, 17, e00567. Canadell, J., Chhabra, A., DeFries, R., Galloway, J., https://doi.org/10.1016/j.gecco.2019.e00567 Heimann, M., Jones, C., Quéré, Le. C., Myneni, R.B., Askar, N., Phairuang, N., Wicaksono, W., Piao, S., & Thornton, P. (2014). Carbon and other bio- Sayektiningsih, T., & Sayektiningsih, T. (2018). geochemical cycles. In T. F. Stocker, D. Qin, G. Plattner, Estimating Aboveground biomass on private forest K. M. Tignor, S. K. Allen, J. Boschung, A. Nauels, Y. Xia, using sentinel-2 imagery”. Journal of Sensors, 6745629, V. Bex, & P. M. Midgley (Eds.), Climate change 2013: The 1–11. https://doi.org/10.1155/2018/6745629 physical science basis. Contribution of working group I to Avdan, U., & Jovanovska, G. (2016). Algorithm for auto- the fifth assessment report of the intergovernmental panel mated mapping of land surface temperature using on climate change (pp. 465–570). United Kingdom and LANDSAT 8 satellite data. Journal of Sensors, 2016, New York, NY, USA: Cambridge University Press 1480307. https://doi.org/10.1155/2016/1480307 GEOLOGY, ECOLOGY, AND LANDSCAPES 11 Civco, D. L. (1989). Topographic normalization of Landsat Holland, P. G., & Steyn, D. G. (1975). Vegetational Thematic Mapper digital imagery. Photogrammetry and responses to latitudinal variations in slope angle and Engineering Remote Sensing, 55(9), 1303–1309. aspect. Journal of Biogeography, 2(3), 179–183. https:// Debortoli, N. S., Dubreuil, V., Hirota, M., Filho, S. R., doi.org/10.2307/3037989 Lindoso, D., & Nabucet, J. (2017). Detecting deforestation Huete, A. R. (1988). A soil-adjusted vegetation index impacts in Southern Amazonia rainfall using rain gauges. (SAVI). Remote Sensing of Environment, 25(3), 295–309. International Journal of Climatology, 37(6), 2889–2900. https://doi.org/10.1016/0034-4257(88)90106-X https://doi.org/10.1002/joc.4886 Huete, A., Justice, C., & Liu, H. (1994). Development of Deka, J., Yumnam, J., Mahanta, P., & Tripathi, O. P. (2015). vegetation and soil indices for MODIS-EOS. Remote Improvement in estimation of above ground biomass of Sensing of Environment, 49(3), 224–234. https://doi.org/ albizia lebbeck using fraction reflectance of landsat TM 10.1016/0034-4257(94)90018-3 data. International Journal of Plant and Environment, 1 Huete, A. R., & Liu, H. Q. (1994). An error and sensitivity (1). https://doi.org/10.18811/ijpen.v1i1.7118 analysis of the atmospheric-and soil-correcting variants Devagiri, G. M., Money, S., Singh, S., Dadhawal, V. K., of the NDVI for the MODIS-EOS. IEEE Transactions on Patil, P., Khaple, A., Devakumar, A. S., & Hubballi, S. Geoscience and Remote Sensing, 32(4), 897–905. https:// (2013). Assessment of above ground biomass and carbon doi.org/10.1109/36.298018 pool in different vegetation types of south western part of IPCC. (2006). Forest Land. IPCC guidelines for national Karnataka, India using spectral modeling. Tropical greenhouse gas inventories. In H. S. Eggleston, Ecology, 54(2), 149–165. L. Buendia, K. Miwa, T. Ngara, & K. Tanabe (Eds.), Dong, L., Du, H., Han, N., Li, X., Zhu, D., Mao, F., Prepared by the national greenhouse gas inventories pro- Zhang, M., Zheng, J., Liu, H., Huang, Z., & He, S. gramme. Japan: IGES. (2020). Application of convolutional neural network on Jiang, F., Kutia, M., Ma, K., Chen, S., Long, J., & Sun, H. lei bamboo above-ground-biomass (AGB) estimation (2021). Estimating the aboveground biomass of conifer- using Worldview-2. Remote Sensing, 12(6), 958. https:// ous forest in Northeast China using spectral variables, doi.org/10.3390/rs12060958 land surface temperature and Soil moisture. Science of Du, L., Zhou, T., Zou, Z., Zhao, X., Huang, K., & Wu, H. the Total Environment, 785, 147335. https://doi.org/10. (2014). Mapping forest biomass using remote sensing and 1016/j.scitotenv.2021.147335 national forest inventory in China. Forests, 5(6), Kashung, Y., Das, B., Deka, S., Bordoloi, R., Paul, A., & 1267–1283. https://doi.org/10.3390/f5061267 Tripathi, O. P. (2018). Geospatial technology based diver- Eckert, S. (2012). Improved forest biomass and carbon esti- sity and above ground biomass assessment of woody mations using texture measures from worldView-2 satel- species of West Kameng district of Arunachal Pradesh. lite data. Remote Sensing, 4(4), 810–829. https://doi.org/ Forest Science and Technology, 14(2), 84–90. https://doi. 10.3390/rs4040810 org/10.1080/21580103.2018.1452797 FSI. (1996). Volume equations for forests of India, Nepal and Kaufman, Y. J., & Tanre, D. (1992). Atmospherically resis- Bhutan. forest survey of india, ministry of environment tant vegetation index (ARVI) for EOS-MODIS. IEEE and forests. Government of India. Transactions on Geoscience and Remote Sensing, 30(2), Gleason, C. J., & Im, J. (2011). A review of remote sensing of 261–270. https://doi.org/10.1109/36.134076 forest biomass and biofuel: Options for small-area Kim, D.-H., Sexton, J. O., Noojipady, P., Huang, C., applications. GIScience & Remote Sensing, 48(2), Anand, A., Channan, S., Feng, M., & Townshed, J. 141–170. https://doi.org/10.2747/1548-1603.48.2.141 (2014). Global, Landsat-based forest-cover change from Goetz, S. J., & Dubayah, R. O. (2011). Advances in remote 1990 to 2000. Remote Sensing of Environment, 155, sensing technology and implications for measuring and 178–193. https://doi.org/10.1016/j.rse.2014.08.017 monitoring forest carbon stocks and change. Carbon Lal, R. (2005). Forest soils and carbon sequestration. Forest Management, 2(3), 231–244. https://doi.org/10.4155/ Ecology and Management, 220(1–3), 242–258. https://doi. cmt.11.18 org/10.1016/j.foreco.2005.08.015 GOFC-GOLD., 2014. A sourcebook of methods and proce- Lawrence, D., & Vandecar, K. (2015). Effects of tropical dures for monitoring and reporting anthropogenic green- deforestation on climate and agriculture. Nature house gas emissions and removals associated with Climate Change, 5(1), 27–36. https://doi.org/10.1038/ deforestation, gains and losses of carbon stocks in forests nclimate2430 remaining forests, and forestation. GOFC-GOLD Report Le Quéré, C., Andrew, R. M., Canadell, J. G., Sitch, S., version COP18-1The Korsbakken, J. I., Peters, G. P., Manning, A. C., Gunawardena, A. R., Nissanka, S. P., Dayawansa, N. D. K., & Boden, T. A., Tans, P. P., Houghton, R. A., Keeling, R. F., Fernando, T. T. (2015). Estimation of above ground bio- Alin, S., Andrews, O. D., Anthoni, P., Barbero, L., Bopp, L., mass in Horton plains national park, Sri Lanka using Chevallier, F., Chini, L. P., , and Zaehle, S. (2016). Global Optical, thermal and RADAR remote sensing data. carbon budget 2016. Earth System Science Data, 8(2), Tropical Agricultural Research, 26(4), 608–623. https:// 605–649. https://doi.org/10.5194/essd-8-605-2016 doi.org/10.4038/tar.v26i4.8123 Le Quéré, C., Andrew, R. M., Friedlingstein, P., Sitch, S., Günlü, A., Ercanli, İ., Ba, E. Z., & Çak, G. (2014). Estimating Hauck, J., Pongratz, J., Pickers, P. A., Korsbakken, J. I., aboveground biomass using Landsat TM imagery : A case Peters, G. P., Canadell, J. G., Arneth, A., Arora, V. K., study of Anatolian crimean pine forests in Turkey. Barbero, L., Bastos, A., Bopp, L., Chevallier, F., Chini, L. P., Annals of Forest Research, 57(2), 289–298. https://doi. Ciais, P., Doney, S. C., . . . Zheng, B. (2018). Global carbon org/10.15287/afr.2014.278 budget 2018. Earth System Science Data, 10(4), 2141–2194. Haripriya, G. S. (2002). Biomass carbon of truncated dia- https://doi.org/10.5194/essd-10-2141-2018 meter classes in Indian forests. Forest Ecology and Li, C., Li, Y., & Li, M. (2019). Improving forest aboveground Management, 168(1–3), 1–13. https://doi.org/10.1016/ biomass (AGB) estimation by incorporating crown S0378-1127(01)00729-0 12 B. DAS ET AL. density and using landsat 8 OLI images of a subtropical coverage using Satellite data in small scale area, forest in western hunan in central China. Forests, 10(2), Mongolia. In IOP Conference Series: Earth and 104. https://doi.org/10.3390/f10020104 Environmental Science, 320(1), 012019. Li, Y., Li, C., Li, C., Liu, Z., & Elbe-Bürger, A. (2020). Forest Pachauri, R. K., & Reisinger, A. (2007). Contribution of aboveground biomass estimation using Landsat 8 and working groups I, II and III to the fourth assessment report Sentinel-1A data with machine learning algorithms. of the intergovernmental panel on climate change IPCC. Scientific Reports, 10(1), 9952. https://doi.org/10.1038/ Pandey, P. C., Koutsias, N., Petropoulos, G. P., s41598-019-56847-4 Srivastava, P. K., & Ben Dor, E. (2019). Land use/land Liu, H. Q., & Huete, A. (1995). A feedback based modification cover in view of earth observation: Data sources, input of the NDVI to minimize canopy background and atmo- dimensions, and classifiers—a review of the state of the spheric noise. IEEE Transactions on Geoscience and Remote art. Geocarto International, 36(9), 957–988. https://doi. Sensing, 33(2), 457–465. https://doi.org/10.1109/TGRS.1995. org/10.1080/10106049.2019.1629647 8746027 Pearson, T. R., Brown, S., Murray, L., & Sidman, G. (2017). Li, B., Wang, W., Bai, L., Chen, N., & Wang, W. (2018). Greenhouse gas emissions from tropical forest degradation: Estimation of aboveground vegetation biomass based on An underestimated source. Carbon Balance Management, 12 Landsat-8 OLI satellite images in the Guanzhong Basin, (1), 3. https://doi.org/10.1186/s13021-017-0072-2 China. International Journal of Remote Sensing, 40(10), Powell, S. L., Cohen, W. B., Healey, S. P., Kennedy, R. E., 3927–3947. https://doi.org/10.1080/01431161.2018.1553323 Moisen, G. G., Pierce, K. B., & Ohmann, J. L. (2010). López-Serrano, P. M., Cárdenas Domínguez, J. L., Corral- Quantification of live aboveground forest biomass Rivas, J. J., Jiménez, E., López-Sánchez, C. A., & Vega- dynamics with Landsat time-series and field inventory Nieva, D. J. (2020). Modeling of Aboveground biomass data: A comparison of empirical modeling approaches. with LANDSAT 8 OLI and machine learning in temperate Remote Sensing of Environment, 114(5), 1053–1068. forests. Forests, 11(1), 11. https://doi.org/10.3390/f11010011 https://doi.org/10.1016/j.rse.2009.12.018 Lu, D. (2006). The potential and challenge of remote sen- Przeździecki, K., Zawadzki, J., Cieszewski, C., & Bettinger, P. sing-based biomass estimation. International Journal of (2017). Estimation of soil moisture across broad land- Remote Sensing, 27(7), 1297–1328. https://doi.org/10. scapes of Georgia and South Carolina using the triangle 1080/01431160500486732 method applied to MODIS satellite imagery. Silva Lu, D., Chen, Q., Wang, G., Liu, L., Li, G., and Moran, E. Fennica, 51(4), 1683. https://doi.org/10.14214/sf.1683 (2014). A survey of remote sensing–based aboveground Qureshi, A., Hussain, S. A., Hussain, S. A., & Hussain, S. A. biomass estimation methods in forest ecosystems. (2012). A review of protocols used for assessment of carbon International Journal of Digital Earth, 9(1), 63–105. stock in forested landscapes. Environmental Science & Policy, https://doi.org/10.1080/17538947.2014.990526 16, 81–89. https://doi.org/10.1016/j.envsci.2011.11.001 MacDicken, K. G. (1997). A guide to monitoring carbon Ramachandra, T. V., & Bharath, S. (2019). Global warming storage in forestry and agroforestry projects. Winrock mitigation through carbon sequestrations in the Central International Institute for Agricultural Development, Western Ghats. Remote Sensing in Earth Systems Sciences, Forest Carbon Monitoring Programme. 2(1), 39–63. https://doi.org/10.1007/s41976-019-0010-z Måren, I. E., Karki, S., Prajapati, C., Yadav, R. K., & Ramachandra, T. V., & Bharath, S. (2020). Carbon seques- Shrestha, B. B. (2015). Facing north or south: Does tration potential of the forest ecosystems in the Western slope aspect impact forest stand characteristics and soil Ghats, a global biodiversity hotspot. Natural Resources properties in a semiarid trans-Himalayan valley? Journal Research, 29(4), 2753–2771. https://doi.org/10.1007/ of Arid Environments, 121, 112–123. https://doi.org/10. s11053-019-09588-0 1016/j.jaridenv.2015.06.004 Ramachandra, T. V., Bharath, S., & Gupta, N. (2018). McGarvey, J. C., Thompson, J. R., Epstein, H. E., & Modelling landscape dynamics with LST in protected Shugart, H. (2015). Carbon storage in old-growth forests areas of Western Ghats, Karnataka. Journal of of the Mid-Atlantic: Toward better understanding the Environmental Management, 206, 1253–1262. https:// eastern forest carbon sink. Ecology, 96(2), 311–317. doi.org/10.1016/j.jenvman.2017.08.001 https://doi.org/10.1890/14-1154.1 Ramankutty, N., Gibbs, H. K., Achard, F., DeFries, R., Motlagh, M. G., Kafaky, S. B., Mataji, A., & Akhavan, R. Foley, J. A., & Houghton, R. A. (2007). Challenges to (2018). Estimating and mapping forest biomass using estimating carbon emissions from tropical deforestation. regression models and Spot-6 images (case study: Global Change Biology, 13(1), 51–66. https://doi.org/10. Hyrcanian forests of north of Iran). Environmental 1111/j.1365-2486.2006.01272.x Monitoring and Assessment, 190, 352. https://doi.org/10. Ravindranath, N. H., Chaturvedi, R. K., & Murthy, I. K. 1007/s10661-018-6725-0 (2008). Forest conservation, afforestation and reforesta- Nath, A. J., Tiwari, B. K., Sileshi, G. W., Sahoo, U. K., tion in India: Implications for forest carbon stocks. Brahma, B., Deb, S., Devi, N., Das, A., Reang, D., Current Science, 95(2), 216–222. http://www.ias.ac.in/ Chaturvedi, S., Tripathi, O., Das, D., & Gupta, A. (2019). currsci/jul252008/216.pdf Allometric models for estimation of forest biomass in North Ren, H., & Zhou, G. (2014). Determination of green above- East India. Forests, 10(2), 103. https://doi.org/10.3390/ ground biomass in desert steppe using litter-soil-adjusted f10020103 vegetation index. European Journal of Remote Sensing, 47 Neinavaz, E., Darvishzadeh, R., Skidmore, A. K., & (1), 611–625. https://doi.org/10.5721/EuJRS20144734 Abdullah, H. (2019). Integration of Landsat-8 ther- Rouse, J. W., Haas, R. H., Hell, J. A., Deering, D. W., & mal and visible-short wave infrared data for improv- Harlan, J. C. (1974). Monitoring the vernal advancement of ing prediction accuracy of forest leaf area index. retrogradation (greenwave effect) of natural vegetation. In Remote Sensing, 11(4), 390. https://doi.org/10.3390/ M. D. Greenbelt (Ed.), NASA/GSFC, Type III, Final Report rs11040390 (pp. 371). Norovsuren, B., Tseveen, B., Batomunkuev, V., & Sahu, S. C., Kumar, M., & Ravindranath, N. H. (2016). Renchin, T. (2020). Estimation for forest biomass and Carbon stocks in natural and planted mangrove forests GEOLOGY, ECOLOGY, AND LANDSCAPES 13 of Mahanadi Mangrove Wetland, East Coast of India. UNFCCC. 2009. Reducing emissions from deforestation in devel- Current Science, 110(12), 2253–2260. https://doi.org/10. oping countries: Approaches to stimulate action. FCCC/ 18520/cs/v110/i12/2253-2260 SBSTA/2009/19/Add.1. FCCC/SBSTA/2009/19/Add.1. Sandholt, I., Rasmussen, K., & Andersen, J. (2002). A simple UN-REDD., 2011. The UN-REDD programme strategy interpretation of the surface temperature/vegetation 2011-2015. https://www.iisd.org/pdf/2011/redd_pro index space for assessment of surface moisture status. gramme_strategy_2011_2015_en.pdf Remote Sensing of Environment, 79(2–3), 213–224. Urbazaev, M., Thiel, C., Migliavacca, M., Reichstein, M., https://doi.org/10.1016/S0034-4257(01)00274-7 Rodriguez-Veiga, P., & Schmullius, C. (2016). Improved Santos, J. R., Freitas, C. C., Araujo, L. S., Dutra, L. V., Mura, multi-sensor satellite-based aboveground biomass estima- J. C., Gama, F. F., Soler, S. S., & Sant'Anna, S. J. (2003). tion by selecting temporally stable forest inventory plots Airborne P-band SAR applied to the aboveground bio- using NDVI time series. Forests, 7(8), 169. https://doi.org/ mass studies in the Brazilian tropical rainforest. Remote 10.3390/f7080169 Sensing of Environment, 87(4), 482–493. https://doi.org/ Vidhya, R., Vijayasekaran, D., Farook, M. A., Jai, S., Rohini, M., 10.1016/j.rse.2002.12.001 & Sinduja, A. (2014). Improved classification of mangroves Sarker, L. R., & Nichol, J. E. (2011). Improved forest biomass health status using hyperspectral remote sensing data. The estimates using ALOS AVNIR-2 texture indices. Remote International Archives of Photogrammetry, Remote Sensing Sensing of Environment, 115(4), 968–977. https://doi.org/ and Spatial Information Sciences, 40(8), 667. https://doi.org/ 10.1016/j.rse.2010.11.010 10.5194/isprsarchives-XL-8-667-2014 Sharma, C., Gairola, S., Ghildiyal, S. K., & Suyal, S. (2010). Wu, C., Shen, H., Shen, A., Deng, J., Gan, M., Zhu, J., Physical properties of soils in relation to forest composi- Xu, H., & Wang, K. (2016). Comparison of tion in moist temperate valley slopes of the Central machine-learning methods for above-ground biomass Western Himalaya. Journal of Forest Science, 26, 117– estimation based on Landsat imagery. Journal of 129. http://ocean.kisti.re.kr/downfile/volume/ifsknu/ Applied Remote Sensing, 10(3), 035010. https://doi. SRGHBV/2010/v26n2/SRGHBV_2010_v26n2_117.pdf org/10.1117/1.JRS.10.035010 Shen, G., Wang, Z., Liu, C., & Han, Y. (2020). Mapping Yadav, A. S., & Gupta, S. K. (2006). Effect of aboveground biomass and carbon in Shanghai’s urban micro-environment and human disturbance on the diver- forest using Landsat ETM+ and inventory data. Urban sity of woody species in the Sariska Tiger Project in India. Forestry & Urban Greening, 51, 126655. https://doi.org/ Forest Ecology and Management, 225, 178–189. https:// 10.1016/j.ufug.2020.126655 doi.org/10.1016/j.foreco.2005.12.058 Sobrino, J. A., Jim´enez-Mu˜noz, J. C., & Paolini, L. (2004). Yu, X., Ge, H., Lu, D., Zhang, M., Lai, Z., & Yao, R. (2019). Land surface temperature retrieval from LANDSAT Comparative study on variable selection approaches in TM5. Remote Sensing of Environment, 90(4), 434–440. establishment of remote sensing model for forest biomass https://doi.org/10.1016/j.rse.2004.02.003 estimation. Remote Sensing, 11(12), 1437. https://doi.org/ Stovall, A. E., Vorster, A. G., Anderson, R. S., Evangelista, P. H., 10.3390/rs11121437 & Shugart, H. H. (2017). Non-destructive aboveground bio- Zanter, K. (2019). Landsat 8 (L8) data users Handbook; mass estimation of coniferous trees using terrestrial LiDAR. LSDS-1574 v.5.0; Department of the Interior Remote Sensing of Environment, 200, 31–42. https://doi.org/ U.S. geological survey. EROS. 10.1016/j.rse.2017.08.013 Zarco-Tejada, P. J., Hornero, A., Hernández-Clemente, R., Suhardiman, A., Tampubolon, B. A., & Sumaryono, M. & Beck, P. S. A. (2018). Understanding the temporal (2018). Examining spectral properties of Landsat 8 OLI dimension of the red-edge spectral region for forest for predicting above-ground carbon of Labanan Forest, decline detection using high-resolution hyperspectral Berau. IOP Conference Series: Earth and Environmental and Sentinel-2a imagery. ISPRS Journal of Science. 144, 012064. Photogrammetry and Remote Sensing, 137, 134–148. Tang, X., Fehrmann, L., Guan, F., Forrester, D., Guisasola, R., https://doi.org/10.1016/j.isprsjprs.2018.01.017 & Kleinn, C. (2016). Inventory based estimation of forest Zhao, P., Lu, D., Wang, G., Wu, C., Huang, Y., & Yu, S. biomass in Shitai County, China: A comparison of five (2016). Examining spectral reflectance saturation in methods. Annals of Forest Research, 59(1), 269–280. Landsat imagery and corresponding solutions to improve https://doi.org/10.15287/afr.2016.574 forest aboveground biomass estimation. Remote Sensing, Tanré, D., Deroo, C., Duhaut, P., Herman, M., 8(6), 469. https://doi.org/10.3390/rs8060469 Morcrette, J. J., Perbos, J., & Deschamps, P. Y. (1990). Zheng, D., Rademacher, J., Chen, J., Crow, T., Bresee, M., Le Technical note Description of a computer code to simu- Moine, J., & Ryu, S. R. (2004). Estimating aboveground late the satellite signal in the solar spectrum: The 5S code. biomass using Landsat 7 ETM+ data across a managed land- International Journal of Remote Sensing, 11(4), 659–668. scape in northern Wisconsin, USA. Remote Sensing of https://doi.org/10.1080/01431169008955048 Environment, 93(3), 402–411. https://doi.org/10.1016/j.rse. Thenkabail, P. S., Stucky, N., Griscom, B. W., Ashton, M. S., 2004.08.008 Diels, J., Van der Meer, B., & Enclona, E. (2004). Biomass Zhou, J., Zhou, Z., Zhao, Q., Han, Z., Wang, P., Xu, J., & estimations and carbon stock calculations in the oil palm Dian, Y. (2020). Evaluation of different algorithms for esti- plantations of African derived savannas using IKONOS mating the growing stock volume of Pinus massoniana data. International Journal of Remote Sensing, 25(23), plantations using spectral and spatial information from 5447–5472. https://doi.org/10.1080/ a SPOT6 image. Forests, 11(5), 540. https://doi.org/10.3390/ 01431160412331291279 f11050540 Tian, J., Dai, T., Li, H., Liao, C., Teng, W., Hu, Q., Ma, W., & Zhu, X., & Liu, D. (2015). Improving forest aboveground Xu, Y. (2019). A novel tree height extraction approach for biomass estimation using seasonal Landsat NDVI individual trees by combining TLS and UAV time-series. ISPRS Journal of Photogrammetry and Remote image-based point cloud integration. Forests, 10(7), 537. Sensing, 102, 222–231. https://doi.org/10.1016/j.isprsjprs. https://doi.org/10.3390/f10070537 2014.08.014 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Geology Ecology and Landscapes Taylor & Francis

Prediction of forest aboveground biomass using an integrated approach of space-based parameters, and forest inventory data

Download PDF
13 pages

Loading next page...
 
/lp/taylor-francis/prediction-of-forest-aboveground-biomass-using-an-integrated-approach-JAXAqrETFX

References (70)

Publisher
Taylor & Francis
Copyright
© 2022 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group on behalf of the International Water, Air & Soil Conservation Society(INWASCON).
ISSN
2474-9508
DOI
10.1080/24749508.2022.2139484
Publisher site
See Article on Publisher Site

Abstract

GEOLOGY, ECOLOGY, AND LANDSCAPES INWASCON https://doi.org/10.1080/24749508.2022.2139484 RESEARCH ARTICLE Prediction of forest aboveground biomass using an integrated approach of space-based parameters, and forest inventory data a,b a b b b,c Biswajit Das , Santanu Kumar Patnaik , Reetashree Bordoloi , Ashish Paul and Om Prakash Tripathi a b Department of Geography, Rajiv Gandhi University, Doimukh, India; Department of Forestry, North Eastern Regional Institute of Science and Technology (Deemed to Be University), Nirjuli, India; Department of Environmental Science, Mizoram University, Aizawl, India ABSTRACT ARTICLE HISTORY Received 11 March 2022 Forests contribute significantly in mitigating the effects of climate change by sequestering Accepted 19 October 2022 atmospheric carbon in biomass and soil. To comprehend carbon stock and sequestration, forest degradation, and climate change mitigation, precise calculation biomass is needed. KEYWORDS The present study in Arunachal Pradesh, Northeast India, used Landsat OLI spectral variables, Spectral variables; regression land surface temperature, and field inventory data to estimate aboveground biomass (AGB) analysis; biomass; Northeast and carbon stock in selected forest types. The stepwise multilinear regression model was India; Land use developed using Landsat-derived spectral data, land surface temperature, and soil moisture −1 −1 The stand density varies from 122 individual ha to 833 individual ha . The estimated AGB −1 −1 density varies from 2.64 t ha to 534.21 t ha among the sample plots. The mean land surface temperature was 14.41°C. The predictive integrated model showed that the mean AGB ranged −1 −1 from 9.45 t ha to 330 t ha and dense forest recorded the maximum mean AGB (239.34 t −1 ha ). Akaike information criteria (AIC) and Bayesian information criteria (BIC) were utilised to evaluate the predictive model, and the AIC (176.13) and BIC (184.58) were omparatively lower than other regression models.To evaluate the accuracy of the integrated model (IM), the linear regression was performed between the predicted and observed AGB. The findings of the study will be useful in formulating site-specific suitable management plan. 1. Introduction affected the carbon capture, storage, and retention in The forest ecosystem has an indispensable contribu- forest and soil carbon pools. Further, the landuse land tion to the global carbon (C) cycle and is regarded as cover (LULC) changes induced forest ecosystem has the largest carbon sink, which removes around 25– been modifying the forest structure that increases the 33% of the anthropogenic greenhouse gas (GHG) forest fragmentation, loss of biodiversity, biogeochem- emission from the atmosphere (Le Quéré et al., ical cycle alteration, and hydrological services 2016). Forest stores a considerable amount of forest (Armenteras et al., 2019; Ramachandra & Bharath, carbon (70–90%) in aboveground biomass (AGB) 2019). (S. Chen et al., 2019) and a large portion of atmo- As the Intergovernmental Panel on Climate spheric carbon dioxide (CO ) in terrestrial land and Change (IPCC) Fifth Assessment Report, defores- soil (Motlagh et al., 2018). Besides climate change tation stands second in the list of largest contri- mitigation, the forest nourishes numerous ecosystem butors of carbon dioxide (CO ) emissions (Ciais services and is an indispensable component of earth’s et al., 2014). Deforestation and land degradation energy cycle. The forest biomass is a crucial parameter account for 20–25% of anthropogenic carbon to describe the structure and function of the forest emissions (Pachauri & Reisinger, 2007; ecosystem (Y. Li et al., 2020). Forest covert the atmo- Ramachandra & Bharath, 2019) hence influencing spheric carbon and is stored in AGB, belowground regional climate patterns including hydrogeologi- biomass, dead organic matter, and soil organic matter cal regime alternation. The forest cover loss has (Ramachandra & Bharath, 2020). The forest ecosystem altered the local rainfall regime due to alternation captures about 40% of terrestrial carbon and 50% of in the pattern of thermodynamic and mesoscale net ecosystem productivity (McGarvey et al., 2015). circulation processes (Lawrence & Vandecar, Annually, about 30% of the global anthropogenic CO 2015), which have extreme weather consequences −1 emissions (2 Pg C yr ) from the atmosphere seques- afterwards. Further, the large and intact forest tered by the forest and soil (Achat et al., 2006; Lal, through their leaves, leaf area and canopy cover- 2005). The extent of land use change, anthropogenic age transferring sensible heat to latent heat, which pressure, disturbances, and climate change have alter the wind dynamics and hence rainfall events CONTACT Biswajit Das biswajitdas.19.1989@gmail.com Department of Geography, Rajiv Gandhi University, Rono Hills, Doimukh, Arunachal Pradesh, India © 2022 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group on behalf of the International Water, Air & Soil Conservation Society(INWASCON). This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 2 B. DAS ET AL. increased (Ramachandra & Bharath, 2020). The now been widely preferred in AGB estimation (Jiang deforestation causes lower evapotranspiration et al., 2021; C. Li et al., 2019; Lu, 2006; Ravindranath across the region, which has the consequence of et al., 2008; Tang et al., 2016). Several researcher have delayed rainfall with low periodicity of rainfall used data like MODIS (Pandey et al., 2019), Landsat TM with longer dry conditions (Debortoli et al., (Deka et al., 2015; Günlü et al., 2014), Landsat ETM+ 2017). The deforestation may lead to biophysical (Shen et al., 2020; Zheng et al., 2004), Landsat OLI alternation in plants through altered micro- (Bordoloi et al., 2022; Kashung et al., 2018), IKONOS climatic condition along with raising air and land (Thenkabail et al., 2004), WorldView-2 (Eckert, 2012), surface temperature (LST) (Ramachandra et al., RADAR (Gunawardena et al., 2015), Sentinel-1A (Y. Li 2018). The increase in temperature has an upward et al., 2020; Norovsuren et al., 2020) to study the for- trend in plant water demand leading to loss of est AGB. canopy, microflora, microclimate change as well Many authors have reported the significance of crucial as an increase in the number of forest fire variables to develop a model that estimates the forest (Ramachandra & Bharath, 2019). AGB (S. Chen et al., 2019; Yu et al., 2019). Normalized The forest biomass combustion and dead plant difference vegetation index (NDVI), Soil-adjusted vege- material decomposition due to deforestation and tation index (SAVI) and Enhanced vegetation index forest degradation have accelerated the atmo- (EVI), Atmospheric resistant vegetation index (ARVI) spheric GHG emission particularly in developing were most widely used approaches (Bordoloi et al., countries (Pearson et al., 2017; Ramankutty et al., 2022; Kashung et al., 2018; Shen et al., 2020). Besides 2007). The terrestrial carbon cycle and carbon above, the digital elevation model (DEM) derived topo- sequestration during the process of plant growth graphic variables like slope, aspect can also be utilized as remains least affected aspect of the global carbon variables for AGB estimation model. Some researchers cycle (Le Quéré et al., 2018). Reducing emissions have used environmental parameters like temperature, from deforestation and degradation (REDD) land surface temperature, soil moisture in forest para- devised mechanisms to implement the forest mon- meter estimation (Jiang et al., 2021; Zarco-Tejada et al., itoring system that needs primarily the forest car- 2018). The satellite-derived land surface temperature bon stock estimation (GOFC-GOLD, 2014; from the thermal infrared sensor (TIRS) of Landsat UNFCCC, 2009; UN-REDD, 2011) through inte- improves the accuracy in estimating leaf area index grated approach of field-based data and earth (LAI) (Neinavaz et al., 2019) while its use in AGB estima- observatory data. Besides forest degradation and tion needs to be evaluated. The use of large number of LULC change, various terrain parameters (slope, variables can be time-consuming, estimation erroneous aspect and elevation) also have significant influ - and render the applicability and interpretability of the ence on forest composition, micro climate, solar model (Yu et al., 2019). Hence, the suitable variables radiation and moisture condition at different hill combination selection is crucial for the AGB estimation. slopes of mountainous forests (Holland & Steyn, Numerous algorithms have been used by the researcher, 1975; Måren et al., 2015; Sharma et al., 2010; among them classical statistical linear regression algo- Yadav & Gupta, 2006). The forest carbon stock rithm was the most widely used algorithm due its simpli- estimation primarily depends on various carbon city nature (C. Li et al., 2019; Lu et al., 2014; Sarker & pool such as AGB, BGB, litter biomass, deadwood Nichol, 2011). The linear relationship analysis used RS and soil organic carbon (SOC) (IPCC, 2006). based spectral variables and the field inventory data. However, the AGB alone has contributed about Hence a stepwise multi-linear regression approach was 50% to the carbon stock (Goetz & Dubayah, 2011). used in the present study. The objective of the study is to There are two major approaches to estimate the forest estimate and predict the biomass stock of different land- AGB viz. destructive and non-destructive approach. The use sectors of the region in connection to climate change former is the most accurate and reliable method of AGB scenarios. The study will assess the region’s biomass stock estimation. However, it fails when there was data belong- using in situ measurements from sample plots and pixel- ing to large area coverage (Qureshi et al., 2012; Stovall based RS derived spectral variables, land surface tempera- et al., 2017). Therefore, the later is the most accepted ture and soil moisture to simulate the spatial biomass approach of AGB estimation using allometric equations density. (Brown & Lugo, 1992; Chave et al., 2005, 2014). In the last two decades, the remote sensing (RS) technology became the most preferred approach and enables researchers to 2. Materials and methods get large scale real-time synoptic view of vegetation con- 2.1 Study area ditions (Du et al., 2014; Tang et al., 2016). However, to estimate the more precise regional AGB, the RS technol- The present study was carried out in four western ogy depends on fine scale filed inventory data. Therefore, districts of Arunachal Pradesh (West Kameng, East integrating field inventory data and RS technology has Kameng, Papum pare and Lower Subansiri) which GEOLOGY, ECOLOGY, AND LANDSCAPES 3 were spread in the mountainous forest of eastern applied the imageries by following the radiometric Himalayas. The study region spreads over an area of correction method as per Landsat 8 user handbook 17,384 km and lies 26°52’ to 29°59’ N latitude and 92° (Zanter, 2019). Additionally, because of the pre- 01’ to 94°22’ E longitude. The climate of the state sence of complex terrain conditions along with varies sharply with changes in latitude and elevation. dense forest cover in the region, it was precau- The areas at very high elevation in the upper tionary to apply topographic illumination correc- Himalayas close to the Tibetan border enjoy an alpine tion, hence an improved cosine correction method climate while below this region it is temperate. The (Civco, 1989) was applied. The ASTER DEM data areas at the sub-Himalayan experience humid subtro- with 30 m resolution was used for topographic pical climate along with hot summers and mild win- correction. The DEM images were re-sampled to ters. The average mean maximum and minimum Landsat OLI pixel size by the nearest neighbour- temperature varied between 29.5°C and 17.7°C in sub- hood transformation (Wu et al., 2016). For the tropical regions and 21.4°C and 2.4°C in cold humid forest inventory process, unsupervised classifica - regions and the foothills and plains experience higher tion approach was done. The satellite image of temperatures. The area is rich in biodiversity includ- the study area categorized into nine land use ing the treasure of germplasm for species of food, land cover categories viz. dense forest, moderately medicinal, bamboo and cane, orchids, wild animal, dense forest, open forest, plantation, agriculture and other life forms. As there was a paucity of data land, settlement, sandy area, snow areas, and on carbon sequestration in the state, it became crucial water bodies (Figure 2). to carry out such studies to estimate the AGB, carbon pool and carbon sequestration of the region (Figure 1). 2.3 Forest inventory data collection The selection of permanent sample plot was based on 2.2 Satellite data acquisition forest density class. To find out the different forest In the present study, cloud-free Landsat OLI opti- density classes, the classified land use map was used, cal RS data for the year 2018 and 2019 and and it showed dense forest, moderately dense forest, Advanced space-borne thermal emission radio- open forest, plantation among the major land use meter (ASTER) elevation data were selected and sectors in major agro-climatic zones. In each of the downloaded from Earth Explorer (https://earthex zones and major land use sectors, sampling plots were plorer.usgs.gov). In Landsat OLI data, the blue, randomly selected in replicates. The forest inventory green, red, near-infrared band, and thermal infra- was done by establishing 61 number of sample plot of red band were used in the analysis. All the ima- 31.6 m x 31.6 m (0.1 ha) as per standard layout. The geries were co-registered to a common UTM sample quadrats were chosen on the basis of stratified (Universal Transversal Mercator) projection sys- random sampling approach and representative sites tem with WGS 84 datum (Zone 46). The top of covering dense to sparse vegetation cover. Tree struc- atmospheric (TOA) reflectance correction was tural parameters like species, girth at breast height Figure 1. Map shows the location of Sample plot in the study area. 4 B. DAS ET AL. Figure 2. Map shows the Land use land cover of the study area. (GBH), height of the all the trees ≥ 30 cm girth at 2.4 Satellite-based spectral variables breast height (at 1.37 m from the base) were measured. The Landsat OLI spectral variables (SV) (Table 1) The measuring tape, clinometer was used to measure include difference vegetation index (DVI) which is the GBH and the height of plant species respectively. a simple difference between near-infrared (NIR) Further, GPS records were also taken for each plot. and red (R) band ratio, normalized difference Forest inventory data were collected for a period of vegetation index (NDVI) which is the ratio of two years (2018–2019) and among 61 permanent NIR and R bands, atmospheric resistant vegetation plots, 14 number of plots sampled in dense forest indices (ARVI) which uses three-band ratio to (DF) and moderately dense forest (MDF) each, 20 minimize the atmospheric scattering effect. Soil plots in open forest (OF) and 13 plots in plantation adjusted vegetation index (SAVI) minimizes soil (PL). Based on the tree DBH and height, the plot based brightness effects in the satellite image, and AGB was calculated for trees having girth ≥ 30 cm at enhanced vegetation index (EVI) is the complex girth breast height (1.37 m) by applying the general- index and high sensitivity to high AGB region as ized allometric equation for the north-eastern region it eliminates the aerosol influence and canopy of India proposed by Nath et al. (2019), which is based background effect. Hence, these vegetation indices on stem diameter, height, and wood density of the were incorporated to analyze the best relationship species. between the AGB and vegetation indices. Table 1. Spectral variables used spatial mapping of biomass and carbon. Vegetation Indices Expression Author NDVI NDVI ¼ ðNIR RÞ=ðNIRþ RÞ Rouse et al. (1974) SAVI SAVI ¼ ½ðNIR RÞ=ðNIRþ Rþ LÞ�ð1þ LÞWhere L = 0.5 for intermediated vegetation cover Huete (1988) ARVI ARVI ¼ ðNIR RBÞ=ðNIRþ RBÞ Kaufman & Tanre, 1992 EVI EVI ¼ G� ðNIR RÞ=ðNIRþ C R C Bþ LÞWhere: C = 6; C = 7.5; L = 1; G = 2.5 Liu and Huete (1995) 1 2 1 2 GEOLOGY, ECOLOGY, AND LANDSCAPES 5 crucial variables. To determine the dry and wet edge 2.5 Forest biomass and carbon calculation from triangle method, for each dry and wet edge, the The non-destructive approach was used to estimate maximum and lowest LST values were chosen from the forest AGB. Though the Forest Survey of India the pixel-based NDVI. As per J. Chen et al. (2011), the (FSI), have developed many species-specific allometric pixels having NDVI between 0.2–0.3 were chosen for equations to estimate the tree AGB (FSI, 1996), how- dry edge calculation. The equation for TDVI is as ever, there was a paucity of the allometric equation for follows. many local species available in the study area. Hence, the generalized allometric equation developed by Nath LST LST min TDVI ¼ et al. (2019) for the northeast India region was used in LST LST max min the present study. To estimate forest carbon stock Where: a conversion factor of 0.55 was used (MacDicken, LST is the minimum surface temperature for min 1997). given pixel for wet edge and modelled using equation (LST = a + b *NDVI) in the triangle, where min min min a and b are regression parameters for the wet min min 2.6 Satellite-based LST and soil moisture edge (Table 2). estimation LST is the maximum surface temperature for max The cloud-free Landsat 8 satellite imageries of the year given pixel for wet edge and modelled using equation 2018 to 2019 covering the study area were acquired (LST = a + b *NDVI) in the triangle, where max max max from the Earth explorer. The Landsat 8 satellite has on a and b are regression parameters for the dry max max boarded two different sensors, the operational land edge (Table 2). The linear regression analysis was imager (OLI) and the Thermal infrared sensor carried out between the plot-based volumetric moist- (TIRS) having different resolution. The T1-level ima- ure content and the plot-based TDVI values to get the geries having the radiometric correction and geo- best fit model for S estimation of the study area. metric correction were acquired. However, two prerequisite processes, the radiometric calibration and atmospheric correction were applied as per 2.7 Development of forest AGB model and Landsat 8 manual (Zanter, 2019) to the all the ima- evaluation geries to get high quality and accurate image informa- tion. The automated algorithm imageries (Avdan & The RS-based SVs, and plot based AGB were taken Jovanovska, 2016; Sobrino et al., 2004) was used in into account to model the forest AGB of the study LST retrieval from Landsat 8. Among the available area. In the present study, the satellite data derived bands, Band 10 (TIR), Band 4 (RED), and Band 5 spectral variables like NDVI, SAVI, ARVI, EVI and (NIR) were used in LST retrieval. The TIR band was physical variables like LST and S were evaluated to used to derive brightness temperature, whereas the find out their linear relationships with plot based AGB RED and NIR band was used in NDVI determination. of each land use sector (Table 2). The sampled plot’s The brightness temperature (BT) was obtained from locations were overlaid on different SVs, gridded LST, the TIRS band using the thermal constant provided in and S data to derive corresponding values of each the image metadata file. Following the NDVI, the land sampled plot. The four most widely used vegetation surface emissivity (LSE) was calculated, which was the indices (VIs) associated with RS based change detec- most essential aspect of LST retrieval. tion and biomass estimation were used. The tested VIs For the global climate change and large-scale includes NDVI, which was the ratio of complementary hydrological cycles, the soil moisture (S ) is regarded reflectance between the red and NIR spectral bands as important influencing factors (J. Chen et al., 2011). due to chlorophyll pigments and leaf cellular structure The retrieval of S was done by using the temperature respectively (Powell et al., 2010), the SAVI, although vegetation dryness index (TVDI). The TVDI derived similar to the NDVI but have an improvement in by using triangle method (Przeździecki et al., 2017; terms of the soil brightness correction factor (Huete, Sandholt et al., 2002), which utilized the NDVI and 1988) and the ARVI, which was self-correcting indices LST as influencing factor for S retrieval. To deter- for both soil and atmosphere reflectance and mini- mine TVDI, dry edge (low evapotranspiration) and mizes both soil and atmospheric noises (Kaufman & wet edge (maximum evapotranspiration) are the two Tanre, 1992). Table 2. TDVI retrieval parameter for Soil moisture retrieval. Parameter Slope Intercept Dry Edge a = −53.73 b = 35.56 min min Wet Edge a = 6.96 b = 10.034 max max 6 B. DAS ET AL. As the current study area have complex land- among the sample plots. Among the forest types, form feature along with different agro-ecological plantation has recorded maximum tree density of −1 zone, it is quite difficult to characterize the precise 609 individual ha followed by the dense forest (563 −1 estimation of AGB. Therefore, a statistical correla- individual ha ), moderately dense forest (517 indivi- −1 −1 tion was developed between LST, S and VIs with dual ha ) and open forest (343 individual ha ). The m, the field-based AGB of each land-use sector. The AGB density varies among the different forest types, −1 −1 plot based AGB was assessed against the satellite and it ranges from 2.64 t ha to 534.21 t ha among derived pixel based selected variables of the the sample plots. The dense forest have recorded the −1 sampled location. To assess the same, the linear maximum AGB of 279.41 t ha followed by the mod- −1 relationship between the variables and AGB was erately dense forest (161.16 t ha ), plantation (88.99 t −1 −1 performed to get best relationship of the model. ha ) and open forest (61.86 t ha ) (Table 3). A multiple regression analysis approach was used for modelling the AGB, which was further 3.2 LST and Soil moisture estimation validated. For the AGB modelling, researchers have used numer- The mean LST for the study area was 14.41°C, how- ous approaches like linear regression approach (Calvao & ever it varied among the different forest types. The Palmeirim, 2004; Ren & Zhou, 2014), multi-regression mean LST observed for DF was 10.19°C, MDF (14.34° approach (Askar et al., 2018; Eckert, 2012; Sarker & C), OF (16.16°C), and PL (16.32°C). The S of the Nichol, 2011), non-linear regression (Santos et al., study area was estimated based on regression model 2003), non-parametric model (Urbazaev et al., 2016), developed between TDVI and field based volumetric artificial neural network (Dong et al., 2020). In the pre- soil moisture content of the sample plot. The devel- sent study, we have applied stepwise multi-linear regres- oped model (S = −103.31*TDVI+147.38, R = 0.70) sion analysis for assessing the relationship between AGB, showed that the mean S of the study area was 30.64%, spectral variables, LST and S . The multi-linear regres- however it ranged between 1.41 and 84.56% (Figure 3). sion approach some time may lead to multicollinearity and overfitting in analysis hence to overcome this most widely used statistical parameter like Pearson correlation 3.3 Optimum variable selection (r), root mean square error (RMSE), coefficient of deter- The statistical analysis of individual SVs with the AGB mination (R ) and p – level were applied. The best fit was done by linear regression analysis and it was model for AGB prediction was done based on high r and evident that the correlation coefficient (R ) obtained high R values. A step wise linear regression method was between NDVI and AGB was 0.60 with RMSE of employed for modelling AGB using Origin software. −1 2 36.50 t ha . Similarly, for the SAVI (R = 0.61, Further to confirm the reliability of the model, R , −1 2 RMSE = 35.52 t ha ), ARVI (R = 0.65, RMSE = RMSE, multicollinearity of variables, the variance infla - −1 2 32.37 t ha ), and for EVI (R = 0.60, RMSE = tion factor {VIF = VIFj = 1/ (1-R )} were determined. To −1 36.53 t ha ). However, the linear regression analysis indicate multicollinearity problem, VIF value greater of LST with AGB showed R = 0.61, RMSE = 35.97 t than 10 (Askar et al., 2018; Sarker & Nichol, 2011) were −1 2 ha and that of S with the AGB resulted R = 0.62, determined. Further, Akaike information criteria (AIC; −1 RMSE = 34.63 t ha (Table 4). Akaike, 1974) and Bayesian information criteria (BIC; Aho et al., 2014) were used in the suitable model selection process. The selection of model was done based on 3.4 Modelling of AGB observed smallest AIC and BIC values. The stepwise multi-linear regression analysis was used to select appropriate variables for predictive AGB 3. Result model development. This was done by anlysing the resultant R , RMSE and VIF observed from linear 3.1 Tree density, aboveground biomass, and relationship analysis individual variables with AGB. carbon stock The selection of the variable’s combination was From the study, it was observed that the mean density based on the minimum RMSE of the model. The −1 −1 varies from 122 individual ha to 833 individual ha statistical relationship of individual SVs, then Table 3. Observed AGB of the sample plot in the study area. Forest Type Plot number Mean density Mean AGB Range Standard Deviation DF 14 563 279.41 156.06–534.21 107.84 MDF 14 517 161.16 96.56–235.69 43.99 OF 20 343 61.86 2.64–162.70 43.15 PL 13 609 88.99 21.07–221-62 58.67 Total 61 482 147.85 2.64–534.21 107.03 GEOLOGY, ECOLOGY, AND LANDSCAPES 7 Figure 3. Land surface temperature and soil moisture of the study area. Table 4. Statistical analysis of stepwise multilinear regression of AGB with variables. Variables R r RMSE p-value VIF NDVI 0.60 0.77 36.51 0.00 2.52 SAVI 0.61 0.78 35.52 0.00 2.59 ARVI 0.64 0.80 32.36 0.0002 2.84 EVI 0.60 0.77 36.53 0.00 2.52 LST 0.61 0.78 35.97 0.00 2.56 S 0.62 0.79 34.63 0.00 2.66 Integrated Model (IM) 0.82 0.90 17.09 0.00 5.67 IM = SAVI + ARVI + LST + S −1 combining all the SVs, and finally combining the LST predicted AGB ranged between 9.45 t ha and 330 t −1 and S with the SVs is presented in Table 4. ha . The DF have maximum mean predicted AGB −1 The linear regression analysis of field based AGB density (239.34 t ha ) followed by MDF (160.49 t −1 −1 −1 and individual variables of the study area is shown in ha ), PL (88.47 t ha ) and OF (62.29 t ha ). Table 4. The correlation coefficient R ranged between Further AIC and BIC analysis validated the resultant 0.60 and 0.64 whereas the Pearson’s r varied between model by showing the lower AIC (176.13) and BIC 0.77 and 0.80. Among the selected SVs, the ARVI have (184.58) value (Table 5). The AGB density map was shown better correlation (r = 0.80, R = 0.64). It was prepared by considering the predicted AGB of the evident that all the selected variables have shown sig- study area. nificant and positive correlation with AGB. Further, the predictive AGB model derived from stepwise 4. Discussion multi- linear regression technique has been expressed as follows: The current study focused on an integrated approach that used freely available optical satellite data-derived AGB ¼ 23:71� SAVI þ 15:62� ARVI 0:72� LST SVs, LST, S , field inventory measurements, and an þ 0:34� SM 4:23 empirical modelling approach to predict the spatial The developed predictive model derived using SVs, AGB of the study region. Although the satellite data- LST, and S have shown R = 0.82, p < 0.05. The based variables have a significant contribution to the −1 RMSE of the model was observed to be 17.09 t ha . carbon stock modelling process, however, there were certain factors which influenced the satellite data and the data acquisition process that further influence the 3.5 Spatial AGB and carbon density mapping carbon stock modelling. The study region have very To evaluate the accuracy of the integrated model (IM), complex landform types ranging from flat to undulat- the linear regression analysis was performed between ing terrain, as well as different land use sectors ranging the predicted AGB and observed AGB. It was observed from grassland to different forest density class, and that there was strong relationship between field based there surface reflectance is extremely complex, limit- observed AGB and predicted AGB and gave a strong ing the predictive model’s competence. Further, the coefficient of determination (R = 0.82, RMSE = use of multi-date satellite data due to unavailability of −1 13.43 t ha ) (Figure 4). The AGB and carbon density cloud free satellite data and have different sun and map was produced from the developed model between sensor arrangement at the time data acquisition pro- AGB, satellite derived spectral variables, LST and S cess also doubling the effect on capability of carbon (Figure 5). From the study, it was predicted that, the stock process. The variables selection and their 8 B. DAS ET AL. y = 0.8257x + 2.4535 R² = 0.8239 45.00 40.00 35.00 30.00 25.00 20.00 15.00 10.00 5.00 0.00 0.00 10.00 20.00 30.00 40.00 50.00 60.00 Observed AGB( Tonnes/0.1 ha) Figure 4. Relationship between Predicted AGB and Observed AGB. Table 5. AIC for the developed regression analysis of AGB with variables. Variables NDVI SAVI ARVI EVI LST SM IM AIC 219.41 217.74 212.07 219.45 218.51 216.22 176.13 BIC 221.52 219.86 214.19 221.56 220.62 218.33 184.58 IM = integrated Model characteristics were crucial in AGB predictive model variables. Presently, the available variables for the and accuracy of the predictive model is solely depen- AGB predictive models includes terrain factor, envir- dent on the inherent characteristics of selected onmental factors, textural feature and SVs (Lu et al., Figure 5. Predicted AGB density map of the study area. Predicted AGB (Tonnes/0.1 ha) GEOLOGY, ECOLOGY, AND LANDSCAPES 9 2014; Zarco-Tejada et al., 2018). The spectral variables reported by Haripriya (2002) for the Indian forest. include vegetation indices, which were most preferred Y. Li et al. (2020) reported that the mean AGB in −1 variables for AGB modelling. The incorporation tex- range of 16.76 to 173.72 t ha for the subtropical ture along with terrain factors for some specific area forest of Chenzhou City of China. The mean AGB −1 significantly enhances the accuracy of the AGB model predicted for plantation was 88.47 t ha . Devagiri (Gleason & Im, 2011). Further, the different forest et al. (2013) estimated comparably slightly higher. cover type has different LST performance, which influ - The mean predicted AGB for the tropical plantations −1 ences the vegetation growth (Neinavaz et al., 2019). (185.95 t ha ) showed higher mean AGB (118.19 t −1 The present predictive model used the Landsat OLI ha ) for teak plantation of Karnataka. Similarly, Sahu data derived SVs, LST, and S . Among the used SVs, et al. (2016) had also reported higher AGB of 181.2 t −1 although the NDVI was the most preferred VI in the ha for Mangrove plantation of Odisha. present study, however, it showed a comparably lower correlation with the field-based AGB, this might be 5. Pros and cons attributed due to the high saturation problem at high AGB stock areas. The same was reported that the The incorporation of numerous variables and fac- saturation was caused due to high reflectance in NIR tors in predictive AGB model may lead to uncer- in comparison to low reflectance in red band, which tainty and caused estimation error. The selected could result inferior correlation (Askar et al., 2018). variables significantly influence the estimation Due to the correction in soil brightness factor, the model performances (S. Chen et al., 2019). Beside SAVI minimized the soil reflectance over the canopy this, the sensor type, approach of sampling strategy reflectance properties, and have showed better perfor- and estimation model (Lu et al., 2014) were also mance than the NDVI (Huete et al., 1994; Vidhya major factor for AGB estimation model. Presently, et al., 2014). Among the selected SVs, the ARVI mini- optical data like Sntinel-2, Landsat-8, and SPOT mizes the atmospheric aerosols brightness effect of the images have major contribution in large scale earth’s surface feature reflectance properties, and AGB estimation (Y. Chen et al., 2019; Lu et al., hence have the better correlation (R = 0.64) with the 2014; Zhou et al., 2020). Among all the optical AGB, than the other SVs (Huete & Liu, 1994; Tanré data, Landsat series have been providing data with et al., 1990). medium resolution with great revisit time. The The multi-linear regression approach using the Landsat-8 onboard TIRS sensor effectively used in SVs, LST and S was applied to AGB model and LST retrieval, which significantly enhances the AGB estimation. The incorporation of the LST and accuracy of AGB model (Gleason & Im, 2011; S in the predictive model showed the improvement Kim et al., 2014). However, optical data fails to in accuracy, which was further evaluated by the provide the canopy information and have higher decrease in RMSE (Table 4). The performance of the saturation mainly in area having dense forest predictive model further assessed through the AIC cover (Zhao et al., 2016; Zhu & Liu, 2015). The value obtained from the statistical analysis of selected researcher like Tian et al. (2019) and Cao et al. variables (Table 4). The AIC (Bordoloi et al., 2022) (2018) have successfully used Airborne laser scan- and BIC (Yu et al., 2019) index showed the better ning (ALS) and terrestrial laser scanning (TLS) in competence in linear indices for the predictive AGB small scale AGB estimation with excellent accuracy. model. Hence, both AIC and BIC tested in the present However, the inherent character of the sensors and study and observed lower AIC and BIC value for the resolution makes it, difficult in large scale study. IM among the individual model using each variable, The linear approaches are quite simple and can can be compared with AIC and BIC value of above depict the relationships smoothly among the selected study. Hence, from the present analysis, it was variables; however, it fails in reflecting the relation- observed that using satellite data derived VIs as pre- ships between the AGB and selected variables under dictor variable quite helpful in AGB prediction due to complex forest conditions (S. Chen et al., 2019). their sensitivity of minimizing the terrain, atmo- Further, the elimination of collinearity among the sphere, and soil brightness effect on surface reflectance variables are prerequisite to ensure the significance properties (Lu et al., 2014). The RMSE thus obtained and interpretability of variables, as these may lead to using Landsat data from the study can be compared deletion of highly related variables with AGB (S. Chen with the value reported by (B. Li et al., 2018). et al., 2019). The mean AGB predicted from the present study −1 was 131.04 t ha which can be compared with the 6. Conclusion findings of studies that used Landsat OLI for the different region (B. Li et al., 2018; López-Serrano To study the regional and global carbon cycle mon- et al., 2020; Suhardiman et al., 2018). The mean AGB itoring, forest dynamic management, the mapping of −1 can also be compared with mean AGB (92 t ha ) AGB spatial distribution among the different forest 10 B. DAS ET AL. types are crucial. The present study mainly empha- Bordoloi, R., Das, B., Tripathi, O. P., Sahoo, U. K., Nath, A. J., Deb, S., Das, D. J., Gupta, A., Devi, N. B., sized the utilization of integrated approach which Charturvedi, S. S., Tiwari, B. K., Paul, A., & Tajo, L. combined the non-destructive, satellite data derived (2022). Satellite based integrated approaches to modelling SVs, LST, S and empirical modelling algorithms to spatial carbon stock and carbon sequestration potential of predict the spatial distribution of AGB of the Study different land uses of Northeast India. Environmental and area. From the study, it was evident that the using of Sustainability Indicators, 13, 100166. https://doi.org/10. 1016/j.indic.2021.100166 optical satellite derived SVs, LST and S as dependant Brown, S., & Lugo, A. E. (1992). Aboveground biomass variables smoothen the AGB prediction process of the estimates for tropical moist forests of the Brazilian study area. Further, the use of linear method to select Amazon. Interciencia. Caracas, 17(1), 8–18. http://phi the optimum variables may render the accuracy of the lip.inpa.gov.br/publ_livres/Other%20side-outro%20lado/ AGB model. Hence, non-linear method may be used Brown%20%26%20Lugo%20biomass/Brown%20and% to get better performance. Besides, optical data, the use 20Lugo%201992-Aboveground%20biomass%20esti mates.pdf SAR, LiDAR satellite data will have an advantage in Calvao, T., & Palmeirim, J. M. (2004). Mapping mediterra- determining forest structural parameter than the opti- nean scrub with satellite imagery: Biomass estimation and cal data, and hence have the better capability to predict spectral behaviour. International Journal of Remote the AGB. The use of deep learning techniques which Sensing, 25(16), 3113–3126. https://doi.org/10.1080/ includes artificial neural network, fuzzy logic, and Cao, L., Pan, J., Li, R., Li, J., & Li, Z. (2018). Integrating cellular automata, also has better prospects to model airborne LiDAR and optical data to estimate forest above- the AGB. The proposed integrated approach will have ground biomass in arid and semi-arid regions of China. significant role for the planner and decision makers to Remote Sensing, 10(4), 532. https://doi.org/10.3390/ achieve the carbon stock management along with their rs10040532 climate change mitigation at the regional level as well Chave, J., Andalo, C., Brown, S., Cairns, M. A., as national and global level. Chambers, J. Q., Eamus, D., Foster, H., Fromard, F., Higuchi, N., Kira, T., Lescure, J. P., Nelson, B. W., Ogawa, H., Puig, H., Riera, B., & Yamakura, T. (2005). Tree allometry and improved estimation of carbon stocks Disclosure statement and balance in tropical forests. Oecologia, 145(1), 87–99. https://doi.org/10.1007/s00442-005-0100-x No potential conflict of interest was reported by the Chave, J., Réjou-Méchain, M., Búrquez, A., Chidumayo, E., author(s). Colgan, M. S., Delitti, W. B., Vieilledent, G., Eid, T., Fearnside, P. M., Goodman, R. C., Henry, M., Martínez- Yrízar, A., Mugasha, W. A., Muller-Landau, H. C., ORCID Mencuccini, M., Nelson, B. W., Ngomanda, A., Nogueira, E. M., Ortiz-Malavassi, E., . . . Vieilledent, G. Biswajit Das http://orcid.org/0000-0003-0016-0082 (2014). Improved allometric models to estimate the aboveground biomass of tropical trees. Global Change Biology, 20(10), 3177–3190. https://doi.org/10.1111/gcb. References Achat, D. L., Fortin, M., Landmann, G., Ringeval, B., & Chen, S., Feng, Z., Chen, P., Ullah Khan, T., & Lian, Y. Augusto, L. (2015). Forest soil carbon is threatened by (2019). Non destructive estimation of the above-ground biomass of multiple tree species in boreal forests of China intensive biomass harvesting. Scientific Reports, 5(1), using terrestrial laser scanning. Forests, 10(11), 936. 15991. https://doi.org/10.1038/srep15991 https://doi.org/10.3390/f10110936 Aho, K., DErryberry, D., & Peterson, T. (2014). Model selec- Chen, Y., Li, L., Lu, D., & Li, D. (2019). Exploring bamboo tion for ecologist: The worldviews of AIC and BIC. Ecology, forest aboveground biomass estimation using Sentinel-2 95(3), 631–636. https://doi.org/10.1890/13-1452.1 data. Remote Sensing , 11(1), 7. https://doi.org/10.3390/ Akaike, H. (1974). A new look at the statistical model rs11010007 identification. EEE Transactions on Automatic Control, Chen, J., Wang, C., Jiang, H., Mao, L., & Yu, Z. (2011). 19(6), 716–723. https://doi.org/10.1109/TAC.1974. Estimating soil moisture using Temperature/Vegetation Dryness Index (TVDI) in the Huang-Huai-Hai (HHH) Armenteras, D., Murcia, U., Gonza´ Lez, T. M., Baro´ plain. International Journal of Remote Sensing, 32(4), N, O. J., & Arias, J. E. (2019). Scenarios of land use and 1165–1177. https://doi.org/10.1080/01431160903527421 land cover change for NW Amazonia: Impact on forest Ciais, P., Sabine, C., Bala, G., Bopp, L., Brovkin, V., intactness. Global Ecology and Conservation, 17, e00567. Canadell, J., Chhabra, A., DeFries, R., Galloway, J., https://doi.org/10.1016/j.gecco.2019.e00567 Heimann, M., Jones, C., Quéré, Le. C., Myneni, R.B., Askar, N., Phairuang, N., Wicaksono, W., Piao, S., & Thornton, P. (2014). Carbon and other bio- Sayektiningsih, T., & Sayektiningsih, T. (2018). geochemical cycles. In T. F. Stocker, D. Qin, G. Plattner, Estimating Aboveground biomass on private forest K. M. Tignor, S. K. Allen, J. Boschung, A. Nauels, Y. Xia, using sentinel-2 imagery”. Journal of Sensors, 6745629, V. Bex, & P. M. Midgley (Eds.), Climate change 2013: The 1–11. https://doi.org/10.1155/2018/6745629 physical science basis. Contribution of working group I to Avdan, U., & Jovanovska, G. (2016). Algorithm for auto- the fifth assessment report of the intergovernmental panel mated mapping of land surface temperature using on climate change (pp. 465–570). United Kingdom and LANDSAT 8 satellite data. Journal of Sensors, 2016, New York, NY, USA: Cambridge University Press 1480307. https://doi.org/10.1155/2016/1480307 GEOLOGY, ECOLOGY, AND LANDSCAPES 11 Civco, D. L. (1989). Topographic normalization of Landsat Holland, P. G., & Steyn, D. G. (1975). Vegetational Thematic Mapper digital imagery. Photogrammetry and responses to latitudinal variations in slope angle and Engineering Remote Sensing, 55(9), 1303–1309. aspect. Journal of Biogeography, 2(3), 179–183. https:// Debortoli, N. S., Dubreuil, V., Hirota, M., Filho, S. R., doi.org/10.2307/3037989 Lindoso, D., & Nabucet, J. (2017). Detecting deforestation Huete, A. R. (1988). A soil-adjusted vegetation index impacts in Southern Amazonia rainfall using rain gauges. (SAVI). Remote Sensing of Environment, 25(3), 295–309. International Journal of Climatology, 37(6), 2889–2900. https://doi.org/10.1016/0034-4257(88)90106-X https://doi.org/10.1002/joc.4886 Huete, A., Justice, C., & Liu, H. (1994). Development of Deka, J., Yumnam, J., Mahanta, P., & Tripathi, O. P. (2015). vegetation and soil indices for MODIS-EOS. Remote Improvement in estimation of above ground biomass of Sensing of Environment, 49(3), 224–234. https://doi.org/ albizia lebbeck using fraction reflectance of landsat TM 10.1016/0034-4257(94)90018-3 data. International Journal of Plant and Environment, 1 Huete, A. R., & Liu, H. Q. (1994). An error and sensitivity (1). https://doi.org/10.18811/ijpen.v1i1.7118 analysis of the atmospheric-and soil-correcting variants Devagiri, G. M., Money, S., Singh, S., Dadhawal, V. K., of the NDVI for the MODIS-EOS. IEEE Transactions on Patil, P., Khaple, A., Devakumar, A. S., & Hubballi, S. Geoscience and Remote Sensing, 32(4), 897–905. https:// (2013). Assessment of above ground biomass and carbon doi.org/10.1109/36.298018 pool in different vegetation types of south western part of IPCC. (2006). Forest Land. IPCC guidelines for national Karnataka, India using spectral modeling. Tropical greenhouse gas inventories. In H. S. Eggleston, Ecology, 54(2), 149–165. L. Buendia, K. Miwa, T. Ngara, & K. Tanabe (Eds.), Dong, L., Du, H., Han, N., Li, X., Zhu, D., Mao, F., Prepared by the national greenhouse gas inventories pro- Zhang, M., Zheng, J., Liu, H., Huang, Z., & He, S. gramme. Japan: IGES. (2020). Application of convolutional neural network on Jiang, F., Kutia, M., Ma, K., Chen, S., Long, J., & Sun, H. lei bamboo above-ground-biomass (AGB) estimation (2021). Estimating the aboveground biomass of conifer- using Worldview-2. Remote Sensing, 12(6), 958. https:// ous forest in Northeast China using spectral variables, doi.org/10.3390/rs12060958 land surface temperature and Soil moisture. Science of Du, L., Zhou, T., Zou, Z., Zhao, X., Huang, K., & Wu, H. the Total Environment, 785, 147335. https://doi.org/10. (2014). Mapping forest biomass using remote sensing and 1016/j.scitotenv.2021.147335 national forest inventory in China. Forests, 5(6), Kashung, Y., Das, B., Deka, S., Bordoloi, R., Paul, A., & 1267–1283. https://doi.org/10.3390/f5061267 Tripathi, O. P. (2018). Geospatial technology based diver- Eckert, S. (2012). Improved forest biomass and carbon esti- sity and above ground biomass assessment of woody mations using texture measures from worldView-2 satel- species of West Kameng district of Arunachal Pradesh. lite data. Remote Sensing, 4(4), 810–829. https://doi.org/ Forest Science and Technology, 14(2), 84–90. https://doi. 10.3390/rs4040810 org/10.1080/21580103.2018.1452797 FSI. (1996). Volume equations for forests of India, Nepal and Kaufman, Y. J., & Tanre, D. (1992). Atmospherically resis- Bhutan. forest survey of india, ministry of environment tant vegetation index (ARVI) for EOS-MODIS. IEEE and forests. Government of India. Transactions on Geoscience and Remote Sensing, 30(2), Gleason, C. J., & Im, J. (2011). A review of remote sensing of 261–270. https://doi.org/10.1109/36.134076 forest biomass and biofuel: Options for small-area Kim, D.-H., Sexton, J. O., Noojipady, P., Huang, C., applications. GIScience & Remote Sensing, 48(2), Anand, A., Channan, S., Feng, M., & Townshed, J. 141–170. https://doi.org/10.2747/1548-1603.48.2.141 (2014). Global, Landsat-based forest-cover change from Goetz, S. J., & Dubayah, R. O. (2011). Advances in remote 1990 to 2000. Remote Sensing of Environment, 155, sensing technology and implications for measuring and 178–193. https://doi.org/10.1016/j.rse.2014.08.017 monitoring forest carbon stocks and change. Carbon Lal, R. (2005). Forest soils and carbon sequestration. Forest Management, 2(3), 231–244. https://doi.org/10.4155/ Ecology and Management, 220(1–3), 242–258. https://doi. cmt.11.18 org/10.1016/j.foreco.2005.08.015 GOFC-GOLD., 2014. A sourcebook of methods and proce- Lawrence, D., & Vandecar, K. (2015). Effects of tropical dures for monitoring and reporting anthropogenic green- deforestation on climate and agriculture. Nature house gas emissions and removals associated with Climate Change, 5(1), 27–36. https://doi.org/10.1038/ deforestation, gains and losses of carbon stocks in forests nclimate2430 remaining forests, and forestation. GOFC-GOLD Report Le Quéré, C., Andrew, R. M., Canadell, J. G., Sitch, S., version COP18-1The Korsbakken, J. I., Peters, G. P., Manning, A. C., Gunawardena, A. R., Nissanka, S. P., Dayawansa, N. D. K., & Boden, T. A., Tans, P. P., Houghton, R. A., Keeling, R. F., Fernando, T. T. (2015). Estimation of above ground bio- Alin, S., Andrews, O. D., Anthoni, P., Barbero, L., Bopp, L., mass in Horton plains national park, Sri Lanka using Chevallier, F., Chini, L. P., , and Zaehle, S. (2016). Global Optical, thermal and RADAR remote sensing data. carbon budget 2016. Earth System Science Data, 8(2), Tropical Agricultural Research, 26(4), 608–623. https:// 605–649. https://doi.org/10.5194/essd-8-605-2016 doi.org/10.4038/tar.v26i4.8123 Le Quéré, C., Andrew, R. M., Friedlingstein, P., Sitch, S., Günlü, A., Ercanli, İ., Ba, E. Z., & Çak, G. (2014). Estimating Hauck, J., Pongratz, J., Pickers, P. A., Korsbakken, J. I., aboveground biomass using Landsat TM imagery : A case Peters, G. P., Canadell, J. G., Arneth, A., Arora, V. K., study of Anatolian crimean pine forests in Turkey. Barbero, L., Bastos, A., Bopp, L., Chevallier, F., Chini, L. P., Annals of Forest Research, 57(2), 289–298. https://doi. Ciais, P., Doney, S. C., . . . Zheng, B. (2018). Global carbon org/10.15287/afr.2014.278 budget 2018. Earth System Science Data, 10(4), 2141–2194. Haripriya, G. S. (2002). Biomass carbon of truncated dia- https://doi.org/10.5194/essd-10-2141-2018 meter classes in Indian forests. Forest Ecology and Li, C., Li, Y., & Li, M. (2019). Improving forest aboveground Management, 168(1–3), 1–13. https://doi.org/10.1016/ biomass (AGB) estimation by incorporating crown S0378-1127(01)00729-0 12 B. DAS ET AL. density and using landsat 8 OLI images of a subtropical coverage using Satellite data in small scale area, forest in western hunan in central China. Forests, 10(2), Mongolia. In IOP Conference Series: Earth and 104. https://doi.org/10.3390/f10020104 Environmental Science, 320(1), 012019. Li, Y., Li, C., Li, C., Liu, Z., & Elbe-Bürger, A. (2020). Forest Pachauri, R. K., & Reisinger, A. (2007). Contribution of aboveground biomass estimation using Landsat 8 and working groups I, II and III to the fourth assessment report Sentinel-1A data with machine learning algorithms. of the intergovernmental panel on climate change IPCC. Scientific Reports, 10(1), 9952. https://doi.org/10.1038/ Pandey, P. C., Koutsias, N., Petropoulos, G. P., s41598-019-56847-4 Srivastava, P. K., & Ben Dor, E. (2019). Land use/land Liu, H. Q., & Huete, A. (1995). A feedback based modification cover in view of earth observation: Data sources, input of the NDVI to minimize canopy background and atmo- dimensions, and classifiers—a review of the state of the spheric noise. IEEE Transactions on Geoscience and Remote art. Geocarto International, 36(9), 957–988. https://doi. Sensing, 33(2), 457–465. https://doi.org/10.1109/TGRS.1995. org/10.1080/10106049.2019.1629647 8746027 Pearson, T. R., Brown, S., Murray, L., & Sidman, G. (2017). Li, B., Wang, W., Bai, L., Chen, N., & Wang, W. (2018). Greenhouse gas emissions from tropical forest degradation: Estimation of aboveground vegetation biomass based on An underestimated source. Carbon Balance Management, 12 Landsat-8 OLI satellite images in the Guanzhong Basin, (1), 3. https://doi.org/10.1186/s13021-017-0072-2 China. International Journal of Remote Sensing, 40(10), Powell, S. L., Cohen, W. B., Healey, S. P., Kennedy, R. E., 3927–3947. https://doi.org/10.1080/01431161.2018.1553323 Moisen, G. G., Pierce, K. B., & Ohmann, J. L. (2010). López-Serrano, P. M., Cárdenas Domínguez, J. L., Corral- Quantification of live aboveground forest biomass Rivas, J. J., Jiménez, E., López-Sánchez, C. A., & Vega- dynamics with Landsat time-series and field inventory Nieva, D. J. (2020). Modeling of Aboveground biomass data: A comparison of empirical modeling approaches. with LANDSAT 8 OLI and machine learning in temperate Remote Sensing of Environment, 114(5), 1053–1068. forests. Forests, 11(1), 11. https://doi.org/10.3390/f11010011 https://doi.org/10.1016/j.rse.2009.12.018 Lu, D. (2006). The potential and challenge of remote sen- Przeździecki, K., Zawadzki, J., Cieszewski, C., & Bettinger, P. sing-based biomass estimation. International Journal of (2017). Estimation of soil moisture across broad land- Remote Sensing, 27(7), 1297–1328. https://doi.org/10. scapes of Georgia and South Carolina using the triangle 1080/01431160500486732 method applied to MODIS satellite imagery. Silva Lu, D., Chen, Q., Wang, G., Liu, L., Li, G., and Moran, E. Fennica, 51(4), 1683. https://doi.org/10.14214/sf.1683 (2014). A survey of remote sensing–based aboveground Qureshi, A., Hussain, S. A., Hussain, S. A., & Hussain, S. A. biomass estimation methods in forest ecosystems. (2012). A review of protocols used for assessment of carbon International Journal of Digital Earth, 9(1), 63–105. stock in forested landscapes. Environmental Science & Policy, https://doi.org/10.1080/17538947.2014.990526 16, 81–89. https://doi.org/10.1016/j.envsci.2011.11.001 MacDicken, K. G. (1997). A guide to monitoring carbon Ramachandra, T. V., & Bharath, S. (2019). Global warming storage in forestry and agroforestry projects. Winrock mitigation through carbon sequestrations in the Central International Institute for Agricultural Development, Western Ghats. Remote Sensing in Earth Systems Sciences, Forest Carbon Monitoring Programme. 2(1), 39–63. https://doi.org/10.1007/s41976-019-0010-z Måren, I. E., Karki, S., Prajapati, C., Yadav, R. K., & Ramachandra, T. V., & Bharath, S. (2020). Carbon seques- Shrestha, B. B. (2015). Facing north or south: Does tration potential of the forest ecosystems in the Western slope aspect impact forest stand characteristics and soil Ghats, a global biodiversity hotspot. Natural Resources properties in a semiarid trans-Himalayan valley? Journal Research, 29(4), 2753–2771. https://doi.org/10.1007/ of Arid Environments, 121, 112–123. https://doi.org/10. s11053-019-09588-0 1016/j.jaridenv.2015.06.004 Ramachandra, T. V., Bharath, S., & Gupta, N. (2018). McGarvey, J. C., Thompson, J. R., Epstein, H. E., & Modelling landscape dynamics with LST in protected Shugart, H. (2015). Carbon storage in old-growth forests areas of Western Ghats, Karnataka. Journal of of the Mid-Atlantic: Toward better understanding the Environmental Management, 206, 1253–1262. https:// eastern forest carbon sink. Ecology, 96(2), 311–317. doi.org/10.1016/j.jenvman.2017.08.001 https://doi.org/10.1890/14-1154.1 Ramankutty, N., Gibbs, H. K., Achard, F., DeFries, R., Motlagh, M. G., Kafaky, S. B., Mataji, A., & Akhavan, R. Foley, J. A., & Houghton, R. A. (2007). Challenges to (2018). Estimating and mapping forest biomass using estimating carbon emissions from tropical deforestation. regression models and Spot-6 images (case study: Global Change Biology, 13(1), 51–66. https://doi.org/10. Hyrcanian forests of north of Iran). Environmental 1111/j.1365-2486.2006.01272.x Monitoring and Assessment, 190, 352. https://doi.org/10. Ravindranath, N. H., Chaturvedi, R. K., & Murthy, I. K. 1007/s10661-018-6725-0 (2008). Forest conservation, afforestation and reforesta- Nath, A. J., Tiwari, B. K., Sileshi, G. W., Sahoo, U. K., tion in India: Implications for forest carbon stocks. Brahma, B., Deb, S., Devi, N., Das, A., Reang, D., Current Science, 95(2), 216–222. http://www.ias.ac.in/ Chaturvedi, S., Tripathi, O., Das, D., & Gupta, A. (2019). currsci/jul252008/216.pdf Allometric models for estimation of forest biomass in North Ren, H., & Zhou, G. (2014). Determination of green above- East India. Forests, 10(2), 103. https://doi.org/10.3390/ ground biomass in desert steppe using litter-soil-adjusted f10020103 vegetation index. European Journal of Remote Sensing, 47 Neinavaz, E., Darvishzadeh, R., Skidmore, A. K., & (1), 611–625. https://doi.org/10.5721/EuJRS20144734 Abdullah, H. (2019). Integration of Landsat-8 ther- Rouse, J. W., Haas, R. H., Hell, J. A., Deering, D. W., & mal and visible-short wave infrared data for improv- Harlan, J. C. (1974). Monitoring the vernal advancement of ing prediction accuracy of forest leaf area index. retrogradation (greenwave effect) of natural vegetation. In Remote Sensing, 11(4), 390. https://doi.org/10.3390/ M. D. Greenbelt (Ed.), NASA/GSFC, Type III, Final Report rs11040390 (pp. 371). Norovsuren, B., Tseveen, B., Batomunkuev, V., & Sahu, S. C., Kumar, M., & Ravindranath, N. H. (2016). Renchin, T. (2020). Estimation for forest biomass and Carbon stocks in natural and planted mangrove forests GEOLOGY, ECOLOGY, AND LANDSCAPES 13 of Mahanadi Mangrove Wetland, East Coast of India. UNFCCC. 2009. Reducing emissions from deforestation in devel- Current Science, 110(12), 2253–2260. https://doi.org/10. oping countries: Approaches to stimulate action. FCCC/ 18520/cs/v110/i12/2253-2260 SBSTA/2009/19/Add.1. FCCC/SBSTA/2009/19/Add.1. Sandholt, I., Rasmussen, K., & Andersen, J. (2002). A simple UN-REDD., 2011. The UN-REDD programme strategy interpretation of the surface temperature/vegetation 2011-2015. https://www.iisd.org/pdf/2011/redd_pro index space for assessment of surface moisture status. gramme_strategy_2011_2015_en.pdf Remote Sensing of Environment, 79(2–3), 213–224. Urbazaev, M., Thiel, C., Migliavacca, M., Reichstein, M., https://doi.org/10.1016/S0034-4257(01)00274-7 Rodriguez-Veiga, P., & Schmullius, C. (2016). Improved Santos, J. R., Freitas, C. C., Araujo, L. S., Dutra, L. V., Mura, multi-sensor satellite-based aboveground biomass estima- J. C., Gama, F. F., Soler, S. S., & Sant'Anna, S. J. (2003). tion by selecting temporally stable forest inventory plots Airborne P-band SAR applied to the aboveground bio- using NDVI time series. Forests, 7(8), 169. https://doi.org/ mass studies in the Brazilian tropical rainforest. Remote 10.3390/f7080169 Sensing of Environment, 87(4), 482–493. https://doi.org/ Vidhya, R., Vijayasekaran, D., Farook, M. A., Jai, S., Rohini, M., 10.1016/j.rse.2002.12.001 & Sinduja, A. (2014). Improved classification of mangroves Sarker, L. R., & Nichol, J. E. (2011). Improved forest biomass health status using hyperspectral remote sensing data. The estimates using ALOS AVNIR-2 texture indices. Remote International Archives of Photogrammetry, Remote Sensing Sensing of Environment, 115(4), 968–977. https://doi.org/ and Spatial Information Sciences, 40(8), 667. https://doi.org/ 10.1016/j.rse.2010.11.010 10.5194/isprsarchives-XL-8-667-2014 Sharma, C., Gairola, S., Ghildiyal, S. K., & Suyal, S. (2010). Wu, C., Shen, H., Shen, A., Deng, J., Gan, M., Zhu, J., Physical properties of soils in relation to forest composi- Xu, H., & Wang, K. (2016). Comparison of tion in moist temperate valley slopes of the Central machine-learning methods for above-ground biomass Western Himalaya. Journal of Forest Science, 26, 117– estimation based on Landsat imagery. Journal of 129. http://ocean.kisti.re.kr/downfile/volume/ifsknu/ Applied Remote Sensing, 10(3), 035010. https://doi. SRGHBV/2010/v26n2/SRGHBV_2010_v26n2_117.pdf org/10.1117/1.JRS.10.035010 Shen, G., Wang, Z., Liu, C., & Han, Y. (2020). Mapping Yadav, A. S., & Gupta, S. K. (2006). Effect of aboveground biomass and carbon in Shanghai’s urban micro-environment and human disturbance on the diver- forest using Landsat ETM+ and inventory data. Urban sity of woody species in the Sariska Tiger Project in India. Forestry & Urban Greening, 51, 126655. https://doi.org/ Forest Ecology and Management, 225, 178–189. https:// 10.1016/j.ufug.2020.126655 doi.org/10.1016/j.foreco.2005.12.058 Sobrino, J. A., Jim´enez-Mu˜noz, J. C., & Paolini, L. (2004). Yu, X., Ge, H., Lu, D., Zhang, M., Lai, Z., & Yao, R. (2019). Land surface temperature retrieval from LANDSAT Comparative study on variable selection approaches in TM5. Remote Sensing of Environment, 90(4), 434–440. establishment of remote sensing model for forest biomass https://doi.org/10.1016/j.rse.2004.02.003 estimation. Remote Sensing, 11(12), 1437. https://doi.org/ Stovall, A. E., Vorster, A. G., Anderson, R. S., Evangelista, P. H., 10.3390/rs11121437 & Shugart, H. H. (2017). Non-destructive aboveground bio- Zanter, K. (2019). Landsat 8 (L8) data users Handbook; mass estimation of coniferous trees using terrestrial LiDAR. LSDS-1574 v.5.0; Department of the Interior Remote Sensing of Environment, 200, 31–42. https://doi.org/ U.S. geological survey. EROS. 10.1016/j.rse.2017.08.013 Zarco-Tejada, P. J., Hornero, A., Hernández-Clemente, R., Suhardiman, A., Tampubolon, B. A., & Sumaryono, M. & Beck, P. S. A. (2018). Understanding the temporal (2018). Examining spectral properties of Landsat 8 OLI dimension of the red-edge spectral region for forest for predicting above-ground carbon of Labanan Forest, decline detection using high-resolution hyperspectral Berau. IOP Conference Series: Earth and Environmental and Sentinel-2a imagery. ISPRS Journal of Science. 144, 012064. Photogrammetry and Remote Sensing, 137, 134–148. Tang, X., Fehrmann, L., Guan, F., Forrester, D., Guisasola, R., https://doi.org/10.1016/j.isprsjprs.2018.01.017 & Kleinn, C. (2016). Inventory based estimation of forest Zhao, P., Lu, D., Wang, G., Wu, C., Huang, Y., & Yu, S. biomass in Shitai County, China: A comparison of five (2016). Examining spectral reflectance saturation in methods. Annals of Forest Research, 59(1), 269–280. Landsat imagery and corresponding solutions to improve https://doi.org/10.15287/afr.2016.574 forest aboveground biomass estimation. Remote Sensing, Tanré, D., Deroo, C., Duhaut, P., Herman, M., 8(6), 469. https://doi.org/10.3390/rs8060469 Morcrette, J. J., Perbos, J., & Deschamps, P. Y. (1990). Zheng, D., Rademacher, J., Chen, J., Crow, T., Bresee, M., Le Technical note Description of a computer code to simu- Moine, J., & Ryu, S. R. (2004). Estimating aboveground late the satellite signal in the solar spectrum: The 5S code. biomass using Landsat 7 ETM+ data across a managed land- International Journal of Remote Sensing, 11(4), 659–668. scape in northern Wisconsin, USA. Remote Sensing of https://doi.org/10.1080/01431169008955048 Environment, 93(3), 402–411. https://doi.org/10.1016/j.rse. Thenkabail, P. S., Stucky, N., Griscom, B. W., Ashton, M. S., 2004.08.008 Diels, J., Van der Meer, B., & Enclona, E. (2004). Biomass Zhou, J., Zhou, Z., Zhao, Q., Han, Z., Wang, P., Xu, J., & estimations and carbon stock calculations in the oil palm Dian, Y. (2020). Evaluation of different algorithms for esti- plantations of African derived savannas using IKONOS mating the growing stock volume of Pinus massoniana data. International Journal of Remote Sensing, 25(23), plantations using spectral and spatial information from 5447–5472. https://doi.org/10.1080/ a SPOT6 image. Forests, 11(5), 540. https://doi.org/10.3390/ 01431160412331291279 f11050540 Tian, J., Dai, T., Li, H., Liao, C., Teng, W., Hu, Q., Ma, W., & Zhu, X., & Liu, D. (2015). Improving forest aboveground Xu, Y. (2019). A novel tree height extraction approach for biomass estimation using seasonal Landsat NDVI individual trees by combining TLS and UAV time-series. ISPRS Journal of Photogrammetry and Remote image-based point cloud integration. Forests, 10(7), 537. Sensing, 102, 222–231. https://doi.org/10.1016/j.isprsjprs. https://doi.org/10.3390/f10070537 2014.08.014

Journal

Geology Ecology and Landscapes – Taylor & Francis

Published: Oct 31, 2022

Keywords: Spectral variables; regression analysis; biomass; Northeast India; Land use

There are no references for this article.