Forest aboveground biomass (AGB) plays an important role in the study of the carbon cycle and climate change in the global terrestrial ecosystem. AGB estimation based on remote sensing is an effective method for regional scale. In this study, Landsat 8 Operational Land Imager and Sentinel-1A data and China’s National Forest Continuous Inventory data in combination with three algorithms, either the linear regression (LR), random forest (RF), or the extreme gradient boosting (XGBoost), were used to estimate biomass of the subtropical forests in Hunan Province, China. XGBoost is a scalable tree boosting system that is widely used by data scientists and provides state-of-the-art results for many problems. It can process an entire dataset with billions of examples using a minimal amount of computational resources through the particular way of cache access patterns, data compression, and data fragmentation. The results include: (1) The combination of Landsat 8 and Sentinel-1A images as predictor variables in the XGBoost model provided the best AGB estimation. (2) In contrast to the LR method, the F-test results indicated that a significant improvement in AGB estimation was achieved with the RF and XGBoost algorithms. (3) The effect of parameter optimization was found to be more significant on XGBoost than on RF. (4) The XGBoost model is an effective method for AGB estimation and can reduce the problems of overestimation and underestimation. This research provides a new way of estimating AGB for the subtropical forest based on remote sensing through the synergy of different sensors datasets and modeling algorithms.
Climate change significantly influences changes in ecological phenomena and processes, such as species distribution and phenology, thus accelerating the rate of species extinction or prosperity. Climate change is considered to be one of the most important threats to global biodiversity in the 21st century and will pose significant challenges to biodiversity conservation in the future. The use of niche modelling to predict changes in the suitable distribution of species under climate change scenarios is becoming a hot topic of biological conservation. In this study, we use data from China’s National Forest Continuous Inventory as well as specimen collection data of Cunninghamia lanceolata (Lamb.) Hook to run optimized Maxent models to predict potential suitable distribution of the species in the present day, 2050s, and 2070s under different climate change scenarios in China. In the modeling process, the most important uncorrelated variables were chosen, and the sample-size-adjusted Akaike information criterion (AICc) was used to select the optimal combination of feature type and regularization multiplier. Variable selection reduced the number of variables used and the complexity of the model, and the use of the AICc reduced overfitting. Variables relating to precipitation were more important than temperature variables in predicting C. lanceolata distribution in the optimal model. The predicted suitable distribution areas of C. lanceolata were different for the different periods under different climate change scenarios, with the centroids showing a degree of northward movement. The suitable distribution area is predicted to become more fragmented in the future. Our results reveal the climate conditions required for the suitable distribution of C. lanceolata in China and the likely changes to its distribution pattern in the future, providing a scientific basis for the sustainable management, protection, and restoration of the suitable habitat of this economically important tree species in the context of climate change.
Forest biomass is a major store of carbon and plays a crucial role in the regional and global carbon cycle. Accurate forest biomass assessment is important for monitoring and mapping the status of and changes in forests. However, while remote sensing-based forest biomass estimation in general is well developed and extensively used, improving the accuracy of biomass estimation remains challenging. In this paper, we used China’s National Forest Continuous Inventory data and Landsat 8 Operational Land Imager data in combination with three algorithms, either the linear regression (LR), random forest (RF), or extreme gradient boosting (XGBoost), to establish biomass estimation models based on forest type. In the modeling process, two methods of variable selection, e.g., stepwise regression and variable importance-base method, were used to select optimal variable subsets for LR and machine learning algorithms (e.g., RF and XGBoost), respectively. Comfortingly, the accuracy of models was significantly improved, and thus the following conclusions were drawn: (1) Variable selection is very important for improving the performance of models, especially for machine learning algorithms, and the influence of variable selection on XGBoost is significantly greater than that of RF. (2) Machine learning algorithms have advantages in aboveground biomass (AGB) estimation, and the XGBoost and RF models significantly improved the estimation accuracy compared with the LR models. Despite that the problems of overestimation and underestimation were not fully eliminated, the XGBoost algorithm worked well and reduced these problems to a certain extent. (3) The approach of AGB modeling based on forest type is a very advantageous method for improving the performance at the lower and higher values of AGB. Some conclusions in this paper were probably different as the study area changed. The methods used in this paper provide an optional and useful approach for improving the accuracy of AGB estimation based on remote sensing data, and the estimation of AGB was a reference basis for monitoring the forest ecosystem of the study area.
Forest aboveground biomass (AGB) estimation modeling based on remote sensing is an important method for large-scale biomass estimation; the accuracy of the estimation models has been a topic of broad and current interest. In this study, we used permanent sample plot data and Landsat 8 Operational Land Imager (OLI) images of western Hunan. Remote-sensing-based models were developed for different vegetation types, and different crown density classes were incorporated. The linear model, linear dummy variable model, and linear mixed-effects model were used to determine the most effective and accurate method for remote-sensing-based AGB estimation. The results show that the adjusted coefficient of determination (R2adj) and root mean square error (RMSE) of the linear dummy model and linear mixed-effects model were significantly better than those of the linear model; the R2adj increased more than 0.16 and the RMSE decreased more than 2.12 for each vegetation type, and the F-test also showed significant differences between the linear model and linear dummy variable model and between the linear model and linear mixed-effects model. The accuracies of the AGB estimations of the linear dummy variable model and the linear mixed-effects model were significantly better than those of linear model in the thin and dense crown density classes. There were no significant differences in the AGB estimation performance between the linear dummy variable model and linear mixed-effects model; these two models were more flexible and more suitable than the linear model for remote-sensing-based AGB estimation. The results of this study provide a new approach for solving the low-accuracy estimations of linear models.
The accurate estimation of forest biomass is very significant to the study of regional ecosystems. China’s National Forest Continuous Inventory data and Landsat 8 data were used to establish a linear regression (LR) model for the estimation of forest aboveground biomass (AGB) based on forest type. The original results of the LR yielded upon conducting AGB estimation were insufficient. Therefore, an interpolation map of the residuals of the observed and predicted AGB was used to correct the error of the original LR model. There is a highly positive result of the accuracy of the corrected AGB map. First, the AGB estimation based on forest type could effectively improve its accuracy. For example, significant improvements were made in the estimations of broadleaf, coniferous, and mixed forests compared to that of the total vegetation conducted using the original LR model. Second, semivariance analysis should be conducted before spatial interpolation using the Kriging method to determine optimal semivariogram models and parameters. The optimal semivariogram model for broadleaf and total forests was the exponential model, while that for the coniferous and mixed forests was the spherical model. Third, combining Kriging interpolation predicted the AGB map effectively and reduced the under- and overestimation of AGB, although it did not fully eliminate this limitation. The R 2 values of broadleaf, coniferous, and mixed forests were improved to 0.897, 0.856, and 0.826, respectively. Overall, the methods used in this study provide an effective approach towards improving the accuracy of AGB estimations by reducing under- and overestimation based on remote sensing data and increasing the ability to monitor the forest ecosystem.
As the largest and most important natural terrestrial ecosystem, forest plays a crucial role in reducing the concentrations of greenhouse gases in the atmosphere, mitigating global warming, maintaining the global ecological balance, and promoting global biological evolution and community succession. The accurate and rapid assessment of forest biomass is highly significant for estimating the regional carbon budget and monitoring forest change. In this study, Landsat images and China’s National Forest Continuous Inventory data of 1999, 2004, 2009, and 2014 were used to establish extreme gradient boosting (XGBoost) models for forest aboveground biomass (AGB) estimation based on forest type in the Xiangjiang River Basin, Hunan Province, China. Kriging interpolation of the AGB residuals was used to correct the error of AGB estimation. Then, a new XGBoost model was established using the final corrected AGB maps and climate data to estimate the AGB under different climate scenarios during the 2050s and 2070s. The results indicated that AGB estimation using the XGBoost model with correction via Kriging interpolation of the AGB residuals can significantly improve the accuracy of AGB estimation. The total AGB of the study area increased over time from 1999 to 2014, indicating that the forest quality improved in the study area. Under the different climate scenarios, the total AGB during the 2050s and 2070s was predicted to decline continuously with increasing of greenhouse gas emissions, indicating that greenhouse gas emissions have a negative impact on forest growth. The results of this study can provide data support for evaluating the ecological function and value of forest ecosystems, and for formulating reasonable forest management measures to mitigate the effects of climate change.