Accurate and high-resolution spatial soil information is crucial for efficient and sustainable land use, management, and conservation. Since the establishment of digital soil mapping (DSM) and the GlobalSoilMap working group, significant advances have been made in terms of the availability and quality of spatial soil information globally. However, accurately predicting soil variation over large and complex areas with limited samples remains a challenge, especially for China, which has diverse soil landscapes. To address this challenge, we utilised 11 209 representative multi-source legacy soil profiles (including the Second National Soil Survey of China, the World Soil Information Service, the First National Soil Survey of China, and regional databases) and high-resolution soil-forming environment characterisation. Using advanced ensemble machine learning and a high-performance parallel-computing strategy, we developed comprehensive maps of 23 soil physical and chemical properties at six standard depth layers from 0 to 2 m in China at a 90 m spatial resolution (China dataset of soil properties for land surface modelling version 2, CSDLv2). Data-splitting and independent-sample validation strategies were employed to evaluate the accuracy of the predicted maps' quality. The results showed that the predicted maps were significantly more accurate and detailed compared to traditional soil type linkage methods (i.e. CSDLv1, the first version of the dataset), SoilGrids 2.0, and HWSD 2.0 products, effectively representing the spatial variation of soil properties across China. The prediction accuracy of soil properties at all depth intervals ranged from good to moderate, with median model efficiency coefficients for most soil properties ranging from 0.29 to 0.70 during data-splitting validation and from 0.25 to 0.84 during independent-sample validation. The wide range between the 5 % lower and 95 % upper prediction limits may indicate substantial room for improvement in current predictions. The relative importance of environmental covariates in predictions varied with soil property and depth, indicating the complexity of interactions among multiple factors in the soil formation processes. As the soil profiles used in this study mainly originate from the Second National Soil Survey of China, conducted during the 1970s and 1980s, they could provide new perspectives on soil changes, together with existing maps based on soil profiles from the 2010s. The findings of this study make important contributions to the GlobalSoilMap project and can also be used for regional Earth system modelling and land surface modelling to better represent the role of soil in hydrological and biogeochemical cycles in China. This dataset is freely available at https://www.scidb.cn/s/ZZJzAz (last access: 17 November 2024​​​​​​​) or https://doi.org/10.11888/Terre.tpdc.301235 (Shi and Shangguan, 2024).

Soil moisture (SM) is a key variable in Earth system science that affects various hydrological and agricultural processes. Convolutional long short-term memory (Conv-LSTM) networks are widely used deep learning models for spatio-temporal SM prediction, but they are often regarded as black boxes that lack interpretability and transparency. This study aims to interpret Conv-LSTM for spatio-temporal SM prediction in China, using the permutation importance and smooth gradient methods for global and local interpretation, respectively. The trained Conv-LSTM model achieved a high R2 of 0.92. The global interpretation revealed that precipitation and soil properties are the most important factors affecting SM prediction. Furthermore, the local interpretation showed that the seasonality of variables was more evident in the high-latitude regions, but their effects were stronger in low-latitude regions. Overall, this study provides a novel approach to enhance the trust-building for Conv-LSTM models and to demonstrate the potential of artificial intelligence-assisted Earth system modeling and understanding element prediction in the future.

Hyper‐resolution land surface modeling provides an unprecedented opportunity to simulate locally relevant water and energy cycles. However, the available meteorological forcing data is often insufficient to fulfill the requirements of hyper‐resolution modeling. Here, we developed a comprehensive downscaling framework based on topography‐adjusted methods and automated machine learning (AutoML). With this framework, a 90 m and hourly atmospheric forcing data set was developed from ERA5 data at a 0.25° resolution, and the Common Land Model (CoLM) was then forced with the developed forcing data over two complex terrain regions (the Heihe River Basin and Upper Colorado River Basin). We systematically evaluated the downscaled forcing and the CoLM outputs against both in situ observations and gridded data. The ground‐based validation results suggested consistent improvements for all downscaled forcing variables with mean RMSE improved by 6.362%–95.86%. The downscaled forcings, which incorporated detailed topographic features, offered improved magnitude estimates, achieving a comparable level of performance to that of regional reanalysis forcing data. The downscaled forcing driving the CoLM model showed comparable or better skills in simulating water and energy fluxes, as verified by in situ validations. The hyper‐resolution simulations provided a detailed and more reasonable description of land surface processes and attained similar spatial patterns and magnitudes with high‐resolution land surface data, especially over highly elevated areas. Additionally, this study highlighted the benefits of using mountain radiation theory‐based shortwave radiation downscaling models and AutoML‐assisted precipitation downscaling models. These findings emphasized the significance of integrating topography‐based downscaling methods for hillslope‐scale simulations.

Accurate spatiotemporal predictions of surface soil moisture (SM) are important for many critical applications. Machine learning models provide a powerful method for building an accurate and reliable predictive model of SM. However, the models used in recent studies have some limitations, including lack of spatial autocorrelation (SAC), vague representation of important features, and primarily focused on the one-step forecast. Thus, we proposed an attention based convolutional long-short term memory model (AttConvLSTM) for multistep forecasting. The model includes three layers; spatial compression, axial attention, and encoder-decoder prediction, which are used for compressing spatial information, feature extraction, and multistep prediction, respectively. The model was trained using surface SM from Soil Moisture Active Passive L4 product at 18km spatial resolution over the United States. The results show that AttConvLSTM predicts 24 hours ahead SM with mean R² and RMSE is equal to 0.82 and 0.02, respectively. Compared with LSTM, AttConvLSTM improves the model performance over 73.6% of regions, with an improvement of 8.4% and 17.4% in R² and RMSE, respectively. The performance of the model is mainly influenced by temporal autocorrelation (TAC). Moreover, we also highlight the importance of SAC on model performance, especially over regions with high SAC and low TAC. Moreover, our model is also competent for SM predictions from several hours to several days, which could be a useful tool for predicting all meteorological variables and forecasting extremes.

Soil moisture (SM) has significant impacts on the Earth’s energy and water cycle system. Remote sensing, such as the Soil Moisture Active Passive (SMAP) mission, has delivered valuable estimations of global surface soil moisture. However, it has a 2~3 days revisit time leading to gaps between SMAP areas. To achieve accurate and comprehensive real-time forecast of SM, we propose a spatial–temporal deep learning model based on the Convolutional Gated Recursive Units with Data Integration (DI_ConvGRU) to capture the spatial and temporal variation in SM simultaneously by modeling the influence of adjacent SM values in space and time. Experiments show that the DI_ConvGRU outperforms the ConvGRU with Linear Interpolation (interp_ConvGRU) and the Long Short-Term Memory with Data Integration (DI_LSTM). The best performance (Bias = 0.0132 m³/m³, ubRMSE = 0.022 m³/m³, R = 0.977) has been achieved through the use of spatial–temporal deep learning model and Data Integration term. In comparison with interp_ConvGRU and DI_LSTM, DI_ConvGRU has improved the model performance in 74.88% and 68.99% of the regions according to RMSE, respectively. The predictability of SM depends highly on SM memory characteristics. DI_ConvGRU can provide accurate spatial–temporal forecast for SM with missing data, making them potentially useful for applications such as filling observational gaps in satellite data.

Recent land surface models (LSMs) have evolved significantly in complexity and resolution, requiring comprehensive evaluation systems to assess their performance. This paper introduces the Open Source Land Surface Model Benchmarking System (OpenBench), an open-source, cross-platform benchmarking system designed to evaluate the state-of-the-art LSMs. OpenBench addresses significant limitations in the current evaluation frameworks by integrating processes that encompass human activities, facilitating arbitrary spatiotemporal resolutions, and offering comprehensive visualization capabilities. The system utilizes various metrics and normalized scoring indices, enabling a comprehensive evaluation of different aspects of model performance. Key features include automation for managing multiple reference datasets, advanced data processing capabilities, and support for station-based and gridded data evaluations. By examining case studies on river discharge, urban heat flux, and agricultural modeling, we illustrate OpenBench's ability to identify the strengths and limitations of models across different spatiotemporal scales and processes. The system's modular architecture enables seamless integration of new models, variables, and evaluation metrics, ensuring adaptability to emerging research needs. OpenBench provides the research community with a standardized, extensible framework for model assessment and improvement. Its comprehensive evaluation capabilities and efficient computational architecture make it a valuable tool for both model development and operational applications in various fields.