Spatially continuous soil thickness data at large scales are usually not readily available and are often difficult and expensive to acquire. Various machine learning algorithms have become very popular in digital soil mapping to predict and map the spatial distribution of soil properties. Identifying the controlling environmental variables of soil thickness and selecting suitable machine learning algorithms are vitally important in modeling. In this study, 11 quantitative and four qualitative environmental variables were selected to explore the main variables that affect soil thickness. Four commonly used machine learning algorithms (multiple linear regression (MLR), support vector regression (SVR), random forest (RF), and extreme gradient boosting (XGBoost) were evaluated as individual models to separately predict and obtain a soil thickness distribution map in Henan Province, China. In addition, the two stacking ensemble models using least absolute shrinkage and selection operator (LASSO) and generalized boosted regression model (GBM) were tested and applied to build the most reliable and accurate estimation model. The results showed that variable selection was a very important part of soil thickness modeling. Topographic wetness index (TWI), slope, elevation, land use and enhanced vegetation index (EVI) were the most influential environmental variables in soil thickness modeling. Comparative results showed that the XGBoost model outperformed the MLR, RF and SVR models. Importantly, the two stacking models achieved higher performance than the single model, especially when using GBM. In terms of accuracy, the proposed stacking method explained 64.0% of the variation for soil thickness. The results of our study provide useful alternative approaches for mapping soil thickness, with potential for use with other soil properties.

Vegetation plays an important role in linking water, atmosphere, and soil. The dynamic change in vegetation is an important indicator for the regulation of the terrestrial carbon balance and climate change. This study applied trend analysis, detrended correlation analysis, and the Hierarchical Partitioning Algorithm (HPA) to GIMMS NDVI3g data, meteorological data, and natural vegetation types for the period 1983 to 2015 to analyze the temporal and spatial changes in NDVI during the growing season and its driving factors in the arid region of northwestern China. The results showed that: (1) the growing season length (GSL) was delayed, with a regional trend of 8 d/33 a, due to a significant advancement in the start of the growing season (SOS, −7 d/33 a) and an insignificant delay to the end of growing season (EOS, 2 d/33 a). (2) The regional change in NDVI was mainly driven by temperature and precipitation, contributing to variations in NDVI of forest of 36% and 15%, respectively, and in the NDVI of grassland, of 35% and 21%, respectively. In particular, changes to forested land and medium-coverage grassland (Mgra) were closely related to temperature and precipitation, respectively. (3) The spatial distribution of the mean NDVI of forest was closely related with precipitation, temperature, and solar radiation, with these meteorological variables explaining 20%, 15%, and 10% of the variation in NDVI, respectively. Precipitation and solar radiation explained 29% and 17% of the variation in the NDVI of grassland, respectively. The study reveals the spatial–temporal evolution and driving mechanism of the NDVI of natural vegetation in the arid region of Northwest China, which can provide theoretical and data support for regional vegetation restoration and conservation.

Precipitation plays a dominant role in regulating terrestrial carbon fluxes. In concert with global warming, aridity has been increasing during recent decades in most parts of the world. How global terrestrial carbon fluxes respond to this change, however, is still unclear. Using a remote-sensing-driven, process-based model, the Boreal Ecosystem Productivity Simulator (BEPS), this study investigated the responses of global terrestrial carbon fluxes to meteorological drought, which were characterized by the standardized precipitation evapotranspiration index (SPEI). The results showed that the response of terrestrial carbon fluxes to drought exhibited distinguishable spatial heterogeneity. In most regions, terrestrial carbon fluxes responded strongly to drought. With an increase in annual water balance (annual precipitation minus annual potential evapotranspiration), the response of carbon fluxes to drought became weaker. The lagged time of terrestrial carbon fluxes responding to drought decreased with the increasing strength of carbon fluxes in response to drought. The sensitivity of terrestrial carbon fluxes to drought also showed noticeable spatial heterogeneity. With an increase in annual water balance, the sensitivity first increased and then decreased. Terrestrial carbon fluxes exhibited the highest sensitivity to drought in semi-arid areas.

Retrieving aerosol optical depth (AOD) over land remains challenging because the relatively weak aerosol signal in top-of-atmosphere (TOA) observations must be separated from strong and spatially heterogeneous surface reflectance. Here, we develop a physically informed random forest framework for global 1 km land AOD retrieval from MODIS. The framework integrates multispectral TOA reflectance, surface properties, observation geometry, meteorological conditions, topography, and physically informed aerosol–surface features. Long-term Aerosol Robotic Network (AERONET) observations from 2001 to 2017 were collocated with MODIS and ancillary datasets for model development and evaluation. Two physically informed features were introduced to improve retrieval robustness across diverse aerosol and surface conditions, including minimum AOD derived from long-term AERONET observations and time-series clear-sky reflectance (TSCR) in the blue, red, and shortwave-infrared bands derived using the 6S radiative-transfer model. Independent retrieval evaluation for 2013–2014 showed good agreement with AERONET observations, with R = 0.81, MAE = 0.063, RMSE = 0.096, and 74.93% of matched samples falling within the MODIS land expected-error envelope, although increasing underestimation was observed at high aerosol loading (AOD > 1). The proposed retrievals also showed better agreement with AERONET than the MOD04 Dark Target and Deep Blue products. These results demonstrate the value of incorporating physically interpretable aerosol-background and surface-reflectance information into data-driven retrievals for AOD over land surfaces.