Quantifying the spatial pattern of large-scale forest biomass can provide a general picture of the carbon stocks within a region and is of great scientific and political importance. The combination of the advantages of remote sensing data and field survey data can reduce uncertainty as well as demonstrate the spatial distribution of forest biomass. In this study, the seventh national forest inventory statistics (for the period 2004–2008) and the spatially explicit MODIS Land Cover Type product (MCD12C1) were used together to quantitatively estimate the spatially-explicit distribution of forest biomass in China (with a resolution of 0.05°, ~5600 m). Our study demonstrated that the calibrated forest cover proportion maps allow proportionate downscaling of regional forest biomass statistics to forest cover pixels to produce a relatively fine-resolution biomass map. The total stock of forest biomass in China was 11.9 Pg with an average of 76.3 Mg ha⁻¹ during the study period; the high values were located in mountain ranges in northeast, southwest and southeast China and were strongly correlated with forest age and forest density.

The Delmarva Peninsula in the eastern United States is partially characterized by thousands of small, forested, depressional wetlands that are highly sensitive to weather variability and climate change, but provide critical ecosystem services. Due to the relatively small size of these depressional wetlands and their occurrence under forest canopy cover, it is very challenging to map their inundation status based on existing remote sensing data and traditional classification approaches. In this study, we applied a state-of-the-art U-Net semantic segmentation network to map forested wetland inundation in the Delmarva area by integrating leaf-off WorldView-3 (WV3) multispectral data with fine spatial resolution light detection and ranging (lidar) intensity and topographic data, including a digital elevation model (DEM) and topographic wetness index (TWI). Wetland inundation labels generated from lidar intensity were used for model training and validation. The wetland inundation map results were also validated using field data, and compared to the U.S. Fish and Wildlife Service National Wetlands Inventory (NWI) geospatial dataset and a random forest output from a previous study. Our results demonstrate that our deep learning model can accurately determine inundation status with an overall accuracy of 95% (Kappa = 0.90) compared to field data and high overlap (IoU = 70%) with lidar intensity-derived inundation labels. The integration of topographic metrics in deep learning models can improve the classification accuracy for depressional wetlands. This study highlights the great potential of deep learning models to improve the accuracy of wetland inundation maps through use of high-resolution optical and lidar remote sensing datasets.

Landscape topography is an important driver of landscape distributions of soil properties and processes due to its impacts on gravity-driven overland and intrasoil lateral transport of water and nutrients. Rapid advancements in aerial, space, and geographic technologies have led to large scale availability of digital elevation models (DEMs), which have proven beneficial in a wide range of applications by providing detailed topographic information. In this report, we presented a summary of recent topography-based soil studies and reviewed five main groups of topographic models in geospatial analyses widely used for soil sciences. We then compared performances of two types of topography-based models—topographic principal component regression (TPCR) and TPCR-kriging (TPCR-Kr)—to ordinary kriging (OKr) models in mapping spatial patterns of soil organic carbon (SOC) density and redistribution (SR) rate. The TPCR and OKr models were calibrated at an agricultural field site that has been intensively sampled, and the TPCR and TPCR-Kr models were evaluated at another field of interest with two sampling transects. High-resolution topographic variables generated from light detection and ranging (LiDAR)-derived DEMs were used as inputs for the TPCR model building. Both TPCR and OKr models provided satisfactory results on SOC density and SR rate estimations during model calibration. The TPCR models successfully extrapolated soil parameters outside of the area in which the model was developed but tended to underestimate the range of observations. The TPCR-Kr models increased the accuracies of estimations due to the inclusion of residual kriging calculated from observations of transects for local correction. The results suggest that even with low sample intensives, the TPCR-Kr models can reduce estimation variances and provide higher accuracy than the TPCR models. The case study demonstrated the feasibility of using a combination of linear regression and spatial correlation analysis to localize a topographic model and to improve the accuracy of soil property predictions in different regions.