Aboveground Biomass Prediction Research Articles

Evaluating machine learning approaches for aboveground biomass prediction in fragmented high-elevated forests using multi-sensor satellite data

Accurate aboveground biomass (AGB) estimations over large areas are essential for assessing carbon stocks and forest resources. This study evaluated machine learning approaches for AGB modeling in Pakistan's mountainous region of Diamir district using freely available Sentinel-1 and Sentinel-2 data and 171 field-measured AGB training points. Random Forest, Gradient Tree Boosting, CatBoost, LightGBM, and XGBoost algorithms were implemented and optimized. Models were developed using individual and combined datasets. Sentinel-2 optical data outperformed Sentinel-1 radar data, but the fusion of both sensors achieved the highest accuracy (R2 > 0.7, RMSE = 105.64 Mg/ha, MAE = 85.34 Mg/ha). Tree canopy height was the most informative predictor for this data, besides terrain variables and radar textures. The machine learning models significantly improved AGB estimates compared to traditional regression techniques, and gradient boosters outperformed Random Forest. This research demonstrates the potential of multi-sensor remote sensing data and advanced algorithms for forest biomass mapping in complex terrain, with modeling accuracies reaching root mean squared errors below 90 Mg/ha. The framework provides an effective solution for monitoring biomass using freely available satellite data. Further refinements include integrating higher-resolution optical data and additional field samples for better validation. This study contributes to remote sensing capabilities for assessing vegetation carbon stocks and dynamics.

Crop aboveground biomass monitoring model based on UAV spectral index reconstruction and Bayesian model averaging: A case study of film-mulched wheat and maize

Aboveground biomass (AGB) directly reflects crop carbon fixation capacity. Utilizing unmanned aerial vehicle (UAV) remote sensing for high-throughput AGB monitoring offers benefits to agricultural management and food production. This study aimed to develop a method for monitoring AGB in film-mulched crops using UAV multispectral data. A three-year field experiment was conducted using wheat (Exps. 1 and 2 in 2021–2022, and Exps. 5 and 6 in 2022–2023) and maize (Exps. 3 and 4) with varying cropping patterns. Spectral indices (SIs) of different feature reflectances (e.g., vegetation, soil, and mulch pixels) were reconstructed using least absolute shrinkage and selection operator methods. The Bayesian model averaging (BMA) method integrated seven independent machine learning algorithms to construct a Bayesian ensemble model (BEM). The results demonstrated that spectral index reconstruction (SIR) enhanced the correlation between SIs and AGB, while BMA reduced model uncertainty and improved AGB prediction accuracy. The BEM model, based on Exps. 1 and 2 datasets, achieved determination coefficients of 0.78–0.84 and root mean square error (RMSE) of 0.81–0.89 t ha−1. Model validation across years and crops demonstrated the generalization of the SIR + BEM method for AGB estimation. The RMSEs of the model inter-annual transport predictions were 0.83 and 0.93 t ha−1 in Exps. 5 and 6, respectively, and RMSE of the predictions between crops was 1.43 t ha−1 for the maize dataset (Exps. 3 + 4). The approach proposed in this study has great potential for monitoring growth status of film-mulched crops at the farm scale and providing guidance for precision agriculture management.

Estimation aboveground biomass in subtropical bamboo forests based on an interpretable machine learning framework

Forest biomass is an essential indicator of forest ecosystem carbon cycle and global climate change research, and traditional machine learning cannot explain the mechanism of feature variable impact on forest aboveground biomass (AGB). Therefore, we proposed an interpretable bamboo forest AGB prediction method based on Shaply Additive exPlanation (SHAP) and XGBoost model to explain the impact mechanism of feature variables on AGB. The bamboo forest AGB is estimated using the monthly and annual scale leaf area index (LAI), enhanced vegetation index (EVI), ratio vegetation index (RVI), precipitation (Pre), maximum temperature (Tmax), minimum temperature (Tmin) and solar radiation (Rad) data. The results showed that the method could be effectively predict AGB, and precipitation more important than temperature. The framework revealed the threshold effect, exceeded the threshold value, the impacts of LAI_Ann, EVI_Ann, and Pre_11 on AGB were stable. The SHAP interaction value between LAI_Ann and EVI_Ann decreased with increasing EVI_Ann and LAI_Ann. By contrast, when Pre_11 increased, the SHAP interaction value between LAI_Ann and Pre_11 increased with increasing LAI_Ann. The framework could also be easily implemented, providing an interpretable machine learning model of forest AGB.

Enhancing Aboveground Biomass Prediction through Integration of the SCDR Paradigm into the U-Like Hierarchical Residual Fusion Model.

Deep learning methodologies employed for biomass prediction often neglect the intricate relationships between labels and samples, resulting in suboptimal predictive performance. This paper introduces an advanced supervised contrastive learning technique, termed Improved Supervised Contrastive Deep Regression (SCDR), which is adept at effectively capturing the nuanced relationships between samples and labels in the feature space, thereby mitigating this limitation. Simultaneously, we propose the U-like Hierarchical Residual Fusion Network (BioUMixer), a bespoke biomass prediction network tailored for image data. BioUMixer enhances feature extraction from biomass image data, facilitating information exchange and fusion while considering both global and local features within the images. The efficacy of the proposed method is validated on the Pepper_Biomass dataset, which encompasses over 600 original images paired with corresponding biomass labels. The results demonstrate a noteworthy enhancement in deep regression tasks, as evidenced by performance metrics on the Pepper_Biomass dataset, including RMSE = 252.18, MAE = 201.98, and MAPE = 0.107. Additionally, assessment on the publicly accessible GrassClover dataset yields metrics of RMSE = 47.92, MAE = 31.74, and MAPE = 0.192. This study not only introduces a novel approach but also provides compelling empirical evidence supporting the digitization and precision improvement of agricultural technology. The research outcomes align closely with the identified problem and research statement, underscoring the significance of the proposed methodologies in advancing the field of biomass prediction through state-of-the-art deep learning techniques.

Above‐ground biomass retrieval with multi‐source data: Prediction and applicability analysis in Eastern Mongolia

AbstractGrassland aboveground biomass (AGB) is a key variable to measure grassland productivity, and accurate assessment of AGB is important for optimizing grassland resource management and understanding carbon, water, and energy fluxes. Current approaches on large scales such as the Mongolian Steppe Ecosystem often combine field measurements with optical and/or synthetic aperture radar (SAR) data. Meanwhile, especially the representativeness of the field measurements for large‐scale analysis have seldom been accounted for. Therefore, we provide the first remotely sensed AGB product for central and Eastern Mongolia which (1) uses random forest (RF), (2) is fully validated against over 600 field samples, and (3) applies a novel method, dissimilarity index (DI), to derive the area of applicability of the model with respect to the training data. Therefore, different remote sensing data sources such as multi‐scale and multi‐temporal optical images—Worldview 2 (WV2), Sentinel 2 (S2), and Landsat 8 (L8) in combination with SAR data are tested for their suitability to provide an area‐wide estimation on large scale. The results showed that the AGB prediction by combining Sentinel 1 (S1) and S2 using RF had the highest accuracy. Furthermore, the model was applicable to at least 72.61% of the steppe area. Areas where the model was not applicable are mostly distributed along the edges of grassland. This study demonstrates the potential of combining Sentinel‐derived indices and machine learning to provide a reliable AGB prediction for grassland for extremely large ecosystems with strong climatic gradients.

Computation of prediction intervals for forest aboveground biomass predictions using generalized linear models in a large-extent boreal forest region

Abstract Remotely sensed data have an important application for estimation of forest variables, e.g. height, volume, and aboveground biomass (AGB). The increased use of remotely sensed data implemented along with model-based inference has shown improved efficiency in prediction and mapping of such forest variables. In this study, plot-level airborne laser scanning data and Swedish National Forest Inventory field reference data were used to predict AGB using generalized linear models (GLMs) assuming Gamma and Tweedie distributions for the field observed AGB. The GLMs were selected considering the convenience of not correcting transformation bias as it is required in other regression models with transformed response variable. To overcome the challenge in providing reliable uncertainty estimates for the estimated forest variable map products at individual pixel-scale, we focused on computing 95% prediction intervals (PIs) for Gamma and Tweedie GLMs with a square root link function. The relative uncertainties were computed as the ratio between the half-width of the PIs and the predicted AGBs. The AGB-airborne laser scanning models were developed with root mean square error values of 22.6 Mgha−1 (26%) and 21.7 Mgha−1 (25%), respectively, for the Gamma and Tweedie GLMs. Two methods were applied to compute PIs for the Gamma GLM, one using the R package ‘ciTools’ and another derived through asymptotic theory. It was found that the 95% PIs computed using ‘ciTools’ had the most accurate coverage probability in comparison to the other method. An extended version of these PIs was also utilized for the Tweedie GLMs. The range of PIs associated with the prediction of AGB were narrower for lower predicted AGB values compared with the length of higher predicted AGB values. Comparing the two fitted models, the Gamma GLM showed lower relative uncertainties for the lower range of predicted AGBs, whereas the Tweedie GLM showed lower relative uncertainties for the higher range of predicted AGBs. Overall, the Tweedie GLM provided a better model fit for AGB predictions.

Estimation of Aboveground Biomass for Different Forest Types Using Data from Sentinel-1, Sentinel-2, ALOS PALSAR-2, and GEDI

Forest aboveground biomass (AGB) is integral to the global carbon cycle and climate change study. Local and regional AGB mapping is crucial for understanding global carbon stock dynamics. NASA’s global ecosystem dynamics investigation (GEDI) and combination of multi-source optical and synthetic aperture radar (SAR) datasets have great potential for local and regional AGB estimation and mapping. In this study, GEDI L4A AGB data and ground sample plots worked as true AGB values to explore their difference for estimating forest AGB using Sentinel-1 (S1), Sentinel-2 (S2), and ALOS PALSAR-2 (PALSAR) data, individually and in their different combinations. The effects of forest types and different true AGB values for validation were investigated in this study, as well. The combination of S1 and S2 performed best in forest AGB estimation with R2 ranging from 0.79 to 0.84 and RMSE ranging from 7.97 to 29.42 Mg/ha, with the ground sample plots used as ground truth data. While for GEDI L4A AGB product working as reference, R2 values range from 0.36 to 0.47 and RMSE values range from 31.41 to 37.50 Mg/ha. The difference between using GEDI L4A and ground sample plot as reference shows obvious dependence on forest types. In summary, optical dataset and its combination with SAR performed better in forest AGB estimation when the average AGB is less than 150 Mg/ha. The AGB predictions from GEDI L4A AGB product used as reference underperformed across the different forest types and study sites. However, GEDI can work as ground truth data source for forest AGB estimation in a certain level of estimation accuracy.

Estimating aboveground biomass of grassland in central Asia mountainous areas using unmanned aerial vehicle vegetation indices and image textures – A case study of typical grassland in Tajikistan

Climate change and human activities have severely impacted Central Asia's mountainous grasslands' health status, making monitoring aboveground biomass (AGB) crucial for grassland protection. However, a high-resolution and low-cost solution for AGB monitoring is lacking for Central Asia's grassland. This research proposes an unmanned aerial vehicle (UAV) based AGB monitoring framework using consumer-level cameras. Texture features from UAV RGB images and vegetation indices from UAV multispectral images are used to predict AGB. As one of the typical mountainous countries in Central Asia, Tajikistan is chosen as the study area. Four different grassland types were chosen to investigate the performance of applying UAV for AGB monitoring. Firstly, correlation analysis was performed to identify important features for AGB estimation. Subsequently, the ground-measured AGB and UAV image features were utilized in multiple linear regression (MLR) and generalized additive model (GAM) to develop an AGB prediction model. The results show that the angular second moment (ASM) of the green band and green normalized difference vegetation index (GNDVI) are the two most important features for AGB estimation in the study area. Furthermore, the GAM-based model demonstrated higher accuracy (R2 = 0.87, RMSE = 127.53 g/m2, rRMSE = 0.17) compared to MLR (R2 = 0.74, RMSE = 181.68 g/m2, rRMSE = 0.24), highlighting the nonlinear relationship between AGB and UAV image features. This research uses multispectral and RGB images to achieve a high accuracy and efficient AGB monitoring framework for Central Asia grasslands and provides a reference to validate satellite images for providing long-term and large-scale monitoring of grassland ecosystems and reasonable utilization of grassland resources.

Enhancing estimation of cover crop biomass using field-based high-throughput phenotyping and machine learning models.

Incorporating cover crops into cropping systems offers numerous potential benefits, including the reduction of soil erosion, suppression of weeds, decreased nitrogen requirements for subsequent crops, and increased carbon sequestration. The aboveground biomass (AGB) of cover crops strongly influences their performance in delivering these benefits. Despite the significance of AGB, a comprehensive field-based high-throughput phenotyping study to quantify AGB of multiple cover crops in the U.S. Midwest has not been found. This study presents a two-year field experiment carried out in Eastern Nebraska, USA, to estimate AGB of five different cover crop species [canola (Brassica napus L.), rye (Secale cereale L.), triticale (Triticale × Triticosecale L.), vetch (Vicia sativa L.), and wheat (Triticum aestivum L.)] using high-throughput phenotyping and Machine Learning (ML) models. Destructive AGB sampling was performed three times during each spring season in 2022 and 2023. An array of morphological, spectral, thermal, and environmental features from the sensors were utilized as feature inputs of ML models. Moderately strong linear correlations between AGB and the selected features were observed. Four ML models, namely Random Forests Regression (RFR), Support Vector Regression (SVR), Partial Least Squares Regression (PLSR), and Artificial Neural Network (ANN), were investigated. Among the four models, PLSR achieved the highest Coefficient of Determination (R2) of 0.84 and the lowest Root Mean Squared Error (RMSE) of 892 kg/ha (Normalized RMSE (NRMSE) = 8.87%), indicating that PLSR could be the most appropriate method for estimating AGB of multiple cover crop species. Feature importance analysis ranked spectral features like Normalized Difference Red Edge (NDRE), Solar-induced Fluorescence (SIF), Spectral Reflectance at 485 nm (R485), and Normalized Difference Vegetation Index (NDVI) as top model features using PLSR. When utilizing fewer feature inputs, ANN exhibited better prediction performance compared to other models. Using morphological and spectral parameters as input features alone led to a R2 of 0.80 and 0.77 for AGB prediction using ANN, respectively. This study demonstrated the feasibility of high-throughput phenotyping and ML techniques for accurately estimating AGB of multiple cover crop species. Further enhancement of model performance could be achieved through additional destructive sampling conducted across multiple locations and years.

Aboveground Biomass Prediction of Plots in the Natural Forests of Arid Mountains Based on Large Trees

While the use of large tropical trees to predict aboveground biomass (AGB) in forests has previously been studied, the applicability of this approach in arid regions remains unquantified. In the natural forests of arid mountains of Northwestern China, this study collected individual tree data from 105 plots across 11 sites through field measurements. The objective was to assess the feasibility of using large trees for predicting plot AGB in these natural forests of arid mountains. This entailed determining the contribution of large trees, based on which a plot AGB prediction model was constructed. This study also aimed to identify the optimal number of large trees needed for accurate AGB prediction. The findings indicate that within the natural forests of arid mountains, only seven large trees (approximately 12% of the trees in a plot) are necessary to account for over 50% of the plot AGB. By measuring 18 large trees within a plot, this study achieved a precise plot AGB estimation, resulting in a model rRMSE of 0.27. The regression fit R2 for the predicted AGB and the estimated AGB was 0.79, effectively aligning the predicted and measured AGB. In the Tianshan Mountains’ natural forests, the prediction model yielded further improvements with an rRMSE of 0.13 and a remarkable regression R2 of 0.92 between predicted and estimated AGB. However, due to variances in tree size distribution and tree species biomass, the Altai Mountains’ natural forest was found to be unsuitable for predicting plot AGB using large trees. This study establishes that large trees can effectively represent plot AGB in the natural forests of arid mountains. Employing forest surveys or remote sensing to collect data from a few large trees instead of the entire tree population enables accurate plot AGB prediction. This research serves as the initial quantification of large tree utilization for plot AGB prediction in the natural forests of arid mountains, carrying substantial implications for future arid forest inventories, carbon accounting, and the formulation of prudent conservation strategies.

Optimising carbon fixation through agroforestry: Estimation of aboveground biomass using multi-sensor data synergy and machine learning

As agricultural land expansion is the primary driver of deforestation, agroforestry could be an optimal land use strategy for climate change mitigation and reducing pressure on forests. Agroforestry is a promising method for carbon sequestration. With recent advancements in geospatial and data science technology, the ability to predict aboveground biomass (AGB) and assess ecosystem services in agroforestry is rapidly expanding. This study was conducted in the Belpada Block of Balangir, Odisha, a forest-dominated region of eastern India. We recorded species occurrence and measured plant parameters, including Circumference at Breast Height (CBH), height, and geolocation, in 196 plots (0.09 ha) in agroforestry intervention sites and noted the tree species. This study used Sentinel-1 and Sentinel-2 multi sensor data to achieve data synergy in AGB estimation. Three machine learning models were used: Random Forest (RF), Support Vector Machine (SVM), and Artificial Neural Network (ANN). The RF model exhibited the highest level of prediction accuracy (R2 = 0.69 and RMSE = 17.07 Mg/ha), followed by the ANN model (R2 = 0.63 and RMSE = 19.35 Mg/ha), SVM model (R2 = 0.54, RMSE = 21.97 Mg/ha. The spectral vegetation indices that are (Normalized Difference Vegetation Index (NDVI), Soil-Adjusted Vegetation Index (SAVI), Enhanced Vegetation Index (EVI), Modified Simple Ratio (MSR), Modified Soil-Adjusted Vegetation Index (MSAVI), Difference Vegetation Index (DVI), and SAR backscatter values, were found important variables for AGB prediction. The findings revealed that agroforestry interventions and plantations resulted in an average carbon stock increase of 15 Mg/ha over five years in the study area. The Plant Value Index (PVI), which indicates the importance of species in the local economy and biomass carbon storage, showed that Tectona grandis was the dominant species with the highest PVI value (88.35), followed by Eucalyptus globulus (56.87), Mangifera indica (53.75), and Azadirachta indica (15.45). This approach enables the expansion of monitoring efforts to assess carbon stock in agroforestry systems, thereby promoting effective management strategies.

Quantifying key vegetation parameters from Sentinel-3 and MODIS over the eastern Eurasian steppe with a Bayesian geostatistical model

Accurate estimation of grassland leaf area index (LAI), fractional vegetation cover (FVC), and aboveground biomass (AGB) is fundamental in grassland studies. The newly launched Ocean and Land Color Imager (OLCI) sensor onboard Sentinel-3 (S3) provides images with comparable spatial and spectral resolution with MODIS data. However, the use of S3 OLCI imageries for vegetation variable estimation is rarely evaluated. This study evaluated the potential of S3 OLCI and MODIS data for estimating grassland LAI, FVC, and AGB in the eastern Eurasian steppe. A Bayesian spatial model (Integrated Nested Laplace Approximation with Stochastic Partial Differential Equation, INLA-SPDE) was used to address spatial autocorrelation of in-situ observation data and to enhance our predictions. Our results showed that the models based on S3 OLCI data presented higher accuracy than models with MODIS data. The RMSEs decreased by 3.7–10.8 %, 3.7–7.5 %, and 1.6–14.2 % for LAI, FVC, and AGB predictions, respectively. Through combinations of multiple predictors, we confirmed the robustness of red edge bands for grassland variable estimation, the models employing red edge variables yielded 3.5 %, 3.2 %, and 0.4 % lower RMSEs than models with conventional visible and NIR bands for LAI, FVC, and AGB prediction, respectively. INLA-SPDE spatial model produced lower bias and higher prediction accuracy than random forest and random forests kriging method in most of the models; the INLA-SPDE predicted LAI and FVC maps also showed a better agreement with ground observations than MODIS and PROBA-V land products.

Unmanned Aerial System-Based Wheat Biomass Estimation Using Multispectral, Structural and Meteorological Data

Rapid estimation of above-ground biomass (AGB) with high accuracy is essential for monitoring crop growth status and predicting crop yield. Recently, remote sensing techniques using unmanned aerial systems (UASs) have exhibited great potential in obtaining structural information about crops and identifying spatial heterogeneity. However, methods of data fusion of different factors still need to be explored in order to enhance the accuracy of their estimates. Therefore, the objective of this study was to investigate the combined metrics of different variables (spectral, structural and meteorological factors) for AGB estimation of wheat using UAS multispectral data. UAS images were captured on two selected growing dates at a typical reclaimed cropland in the North China Plain. The spectral response was determined using the highly correlated vegetation index (VI). A structural metric, the canopy height model (CHM), was produced using UAS-based multispectral images. The measure of growing degree days (GDD) was selected as a meteorological proxy. Subsequently, a structurally–meteorologically weighted canopy spectral response metric (SM-CSRM) was derived by the pixel-level fusion of CHM, GDD and VI. Both correlation coefficient analysis and simple function fitting were implemented to explore the highest correlation between the measured AGB and each proposed metric. The optimal regression model was built for AGB prediction using leave-one-out cross-validation. The results showed that the proposed SM-CSRM generally improved the correlation between wheat AGB and various VIs and can be used for estimating the wheat AGB. Specifically, the combination of MERIS terrestrial chlorophyll index (MTCI), vegetation-masked CHM (mCHM) and normalized GDD (nGDD) achieved an optimal accuracy (R2 = 0.8069, RMSE = 0.1667 kg/m2, nRMSE = 19.62%) through the polynomial regression method. This improved the nRMSE by 3.44% compared to the predictor using MTCI × mCHM. Moreover, the pixel-level fusion method slightly enhanced the nRMSE by ~0.3% for predicted accuracy compared to the feature-level fusion method. In conclusion, this paper demonstrated that an SM-CSRM using pixel-level fusion with canopy spectral, structural and meteorological factors can obtain a good level of accuracy for wheat biomass prediction. This finding could benefit the assessment of reclaimed cropland or the monitoring of crop growth and field management in precision agriculture.

Improving maize growth and development simulation by integrating temperature compensatory effect under plastic film mulching into the AquaCrop model

Temperature compensatory effect, which quantifies the increase in cumulative air temperature from soil temperature increase caused by mulching, provides an effective method for incorporating soil temperature into crop models. In this study, compensated temperature was integrated into the AquaCrop model to investigate the capability of the compensatory effect to improve assessment of the promotion of maize growth and development by plastic film mulching (PM). A three-year experiment was conducted from 2014 to 2016 with two maize varieties (spring and summer) and two mulching conditions (PM and non-mulching (NM)), and the AquaCrop model was employed to reproduce crop growth and yield responses to changes in NM, PM, and compensated PM. A marked difference in soil temperature between NM and PM was observed before 50 days after sowing (DAS) during three growing seasons. During sowing–emergence and emergence–tasseling, the increase in air temperature was proportional to the compensatory coefficient, with spring maize showing a higher compensatory temperature than summer maize. Simulation results for canopy cover (CC) were generally in good agreement with the measurements, whereas predictions of aboveground biomass and grain yield under PM indicated large underestimates from 60 DAS to the end of maturity. Simulations of spring maize biomass and yield showed general increase based on temperature compensation, accompanied by improvement in modeling accuracy, with RMSEs decreasing from 2.5 to 1.6 t ha−1 and from 4.1 t to 3.4 t ha−1. Improvement in biomass and yield simulation was less pronounced for summer than for spring maize, implying that crops grown during low-temperature periods would benefit more from the compensatory effect. This study demonstrated the effectiveness of the temperature compensatory effect to improve the performance of the AquaCrop model in simulating maize growth under PM practices.

Aboveground Biomass Prediction of Arid Shrub-Dominated Community Based on Airborne LiDAR through Parametric and Nonparametric Methods

Aboveground biomass (AGB) of shrub communities in the desert is a basic quantitative characteristic of the desert ecosystem and an important index to measure ecosystem productivity and monitor desertification. An accurate and efficient method of predicting the AGB of a shrub community is essential for studying the spatial patterns and ecological functions of the desert region. Even though there are several entries in the literature on the AGB prediction of desert shrub communities using remote sensing data, the applicability and accuracy of airborne LiDAR data and prediction methods have not been well studied. We first extracted the elevation, density and intensity variables based on the airborne LiDAR, and then sample plot-level AGB prediction models were constructed using the parametric regression (nonlinear regression) and nonparametric methods (Random Forest, Support Vector Machine, K-Nearest Neighbor, Gradient Boosting Machine, and Multivariate adaptive regression splines). We evaluated accuracies of all the AGB prediction models we developed based on the fit statistics. Results showed that: (1) the elevation, density and intensity variables obtained from LiDAR point cloud data effectively predicted the AGB of the desert shrub community at a sample plot level, (2) the kappa coefficient of nonlinear mixed-effects (NLME) model obtained was 0.6977 with an improvement by 13% due to the random effects included into the model, and (3) the nonparametric model, such as Support Vector Machine showed the best fit statistics (R2 = 0.8992), which is 28% higher than the NLME-model, and effectively reduced the heteroscedasticity. The AGB prediction model presented in this paper, which is based on the airborne LiDAR data and machine learning algorithm, will provide a valuable tool to the managers and researchers for evaluating desert ecosystem productivity and monitoring desertification.

Forest Aboveground Biomass Prediction by Integrating Terrestrial Laser Scanning Data, Landsat 8 OLI-Derived Forest Canopy Density and Spectral Indices

Forest Aboveground Biomass Prediction by Integrating Terrestrial Laser Scanning Data, Landsat 8 OLI-Derived Forest Canopy Density and Spectral Indices

Relationship between plant species diversity and aboveground biomass in alpine grasslands on the Qinghai–Tibet Plateau: Spatial patterns and the factors driving them

Alpine grasslands are important ecosystems on the Qinghai–Tibet Plateau and are extremely sensitive to climate change. However, the spatial responses of plant species diversity and biomass in alpine grasslands to environmental factors under the background of global climate change have not been thoroughly characterized. In this study, a random forest model was constructed using grassland ground monitoring data with satellite remote sensing data and environmental variables to characterize the plant species diversity and aboveground biomass of grasslands in the Three-River Headwaters Region within the Qinghai–Tibet Plateau and analyze spatial variation in the relationship between the plant species diversity and aboveground biomass and their driving factors. The results show that (1) the selection of characteristic variables can effectively improve the accuracy of random forest models. The stepwise regression variable selection method was the most effective approach, with an R2 of 0.60 for the plant species diversity prediction model and 0.55 for the aboveground biomass prediction model, (2) The spatial distribution patterns of the plant species diversity and aboveground biomass in the study area were similar, they were both high in the southeast and low in the northwest and gradually decreased from east to west. The relationship between the plant species diversity and aboveground biomass varied spatially, they were mostly positively correlated (67.63%), but they were negatively correlated in areas with low and high values of plant species diversity and aboveground biomass, and (3) Analysis with geodetector revealed that longitude, average annual precipitation, and elevation were the main factors driving variation in the plant species diversity and aboveground biomass relationship. We characterized plant species diversity and aboveground biomass, as well as their spatial relationships, over a large spatial scale. Our data will aid biodiversity monitoring and grassland conservation management, as well as future studies aimed at clarifying the relationship between biodiversity and ecosystem functions.

Forest total and component biomass retrieval via GA-SVR algorithm and quad-polarimetric SAR data

A reliable evaluation of biomass is a vital prerequisite for realizing the international goal of “emission peak and carbon neutrality”. It is critical to estimate the components of forest biomass, for ecosystem management. Additionally, working on components we may solve the saturation problems in AGB estimation using remote sensing features. In our previous works we proposed GA-SVR (Genetic algorithms and support vector regression) algorithm with polarimetric SAR (Synthetic Aperture Rader) to retrieve total forest Above Ground Biomass (AGB) estimation in our previous works, however, the potential of GA-SVR algorithm applied in component AGB estimation especially using combination of multi-frequency polarimetric SAR features deserves further exploration. In this study, we use quad-polarimetric SAR data at C- and L- bands, extracting the backscatter coefficients and polarimetric features derived from four polarization decomposition methods (Yamaguchi 3-component decomposition, Freeman 2-component decomposition, H/A/alpha decomposition, and TSVM decomposition) as the input to the GA-SVR for forest component AGB estimation. The effectiveness of 66 polarimetric features derived from C-, L-band at each test site was evaluated for forest component AGB prediction at two test sites. The outcomes demonstrated that the GA-SVR attained high estimation accuracy according to the values of coefficient of determination R2, root mean square error, relative root mean square error, mean deviation, mean absolute deviation, mean percentage error, and mean absolute percentage error. The highest attained values of them were 0.77, 1.01 Mg/ha, 23.02%, −0.07 Mg/ha, 0.71 Mg/ha, 0.15%, and 18.42%, respectively. The study reconfirmed the robustness of GA-SVR algorithm and effectiveness of polarimetric SAR features extracted from four decomposition methods for forest total and AGB estimation. It also revealed that the capability of combining C- band L-band SAR polarimetric features for improving forest total and component AGB relies on the difference of forest structures.

Harnessing data assimilation and spatial autocorrelation for forest inventory

Spatially explicit uncertainties in forest above-ground biomass predictions for population units are underestimated if spatial structure in the form of residual spatial autocorrelation and heteroscedasticity is ignored. Methods that consider the spatial structure of biomass model residuals are needed to comprehensively estimate, as well as to effectively reduce, the uncertainty in biomass predictions, for pursuing higher levels of precision for measurement, reporting and verification of forest carbon stocks. The objectives of the study were threefold: (1) to demonstrate a spatial data assimilation (DA) procedure that harnesses small-footprint airborne LiDAR, the best linear unbiased predictor (BLUP) and the spatial structure of biomass model residuals to reduce prediction variances of individual tree biomass and plot-level biomass density; (2) to derive a variance estimator that decomposes the variance into components associated with corresponding error sources; and (3) to compare prediction variances for three methods used to calibrate a height-based allometric model for tree biomass: ordinary least squares (OLS), generalized least squares (GLS), and spatial DA using the BLUP. Five major conclusions are drawn. First, for individual tree biomass predictions, spatial DA decreased prediction variance by 40% and 20% relative to OLS and GLS. Because the decrease in residual variability accounted for 98% of the decrease in prediction variance in total, the assimilation effect was the largest for reducing residual variability. Second, for biomass density predictions, DA decreased prediction variance by 3% and 49% relative to OLS and GLS, with the largest decrease in residual covariance. Accumulated gain in precision for individual tree predictions in the DA procedure was offset by precision loss caused by residual covariance and the variance associated with omission and commission errors while predicting individual tree biomass from the LiDAR data. Third, OLS, which assumed no spatial structure, underestimated prediction variance for LiDAR-predicted biomass density by 48%. Fourth, from the perspective of prediction accuracy, DA reduced the RMSE for individual tree biomass predictions by 11% and 14% and reduced the RMSE for biomass density predictions by 28% and 33% relative to OLS and GLS. Fifth, the omission/commission difference model was effective for correcting the systematic prediction error in the LiDAR-predicted biomass density. Overall, the proposed spatial DA procedure demonstrated great potential for reducing the uncertainty in forest biomass predictions, thereby facilitating more efficient biomass inventories. The procedure can be generalized to other dependent variables of interest given their correlations with new information from LiDAR.

A Proposed Ensemble Feature Selection Method for Estimating Forest Aboveground Biomass from Multiple Satellite Data

Feature selection (FS) can increase the accuracy of forest aboveground biomass (AGB) prediction from multiple satellite data and identify important predictors, but the role of FS in AGB estimation has not received sufficient attention. Here, we aimed to quantify the degree to which FS can benefit forest AGB prediction. To this end, we extracted a series of features from Landsat, Phased Array L-band Synthetic Aperture Radar (PALSAR), and climatic and topographical information, and evaluated the performance of four state-of-the-art FS methods in selecting predictive features and improving the estimation accuracy with selected features. We then proposed an ensemble FS method that takes inro account the stability of an individual FS algorithm with respect to different training datasets used; the heterogeneity or diversity of different FS methods; the correlations between features and forest AGB; and the multicollinearity between the selected features. We further investigated the performance of the proposed stability-heterogeneity-correlation-based ensemble (SHCE) method for AGB estimation. The results showed that selected features by SHCE provided a more accurate prediction of forest AGB than existing state-of-the-art FS methods, with R2 = 0.66 ± 0.01, RMSE = 14.35 ± 0.12 Mg ha−1, MAE = 9.34 ± 0.09 Mg ha−1, and bias = 1.67 ± 0.11 Mg ha−1 at 90 m resolution. Boruta yielded comparable prediction accuracy of forest AGB, but could not identify the importance of features, which led to a slightly greater bias than the proposed SHCE method. SHCE not only ranked selected features by importance but provided feature subsets that enabled accurate AGB prediction. Moreover, SHCE provides a flexible framework to combine FS results, which will be crucial in many scenarios, particularly the wide-area mapping of land-surface parameters from various satellite datasets.