High-quality gridded soil moisture products are essential for many Earth system science applications, while the recent reanalysis and remote sensing soil moisture data are often available at coarse resolution and remote sensing data are only for the surface soil. Here, we present a 1 km resolution long-term dataset of soil moisture derived through machine learning trained by the in situ measurements of 1789 stations over China, named SMCI1.0 (Soil Moisture of China by in situ data, version 1.0). Random forest is used as a robust machine learning approach to predict soil moisture using ERA5-Land time series, leaf area index, land cover type, topography and soil properties as predictors. SMCI1.0 provides 10-layer soil moisture with 10 cm intervals up to 100 cm deep at daily resolution over the period 2000–2020. Using in situ soil moisture as the benchmark, two independent experiments were conducted to evaluate the estimation accuracy of SMCI1.0: year-to-year (ubRMSE ranges from 0.041 to 0.052 and R ranges from 0.883 to 0.919) and station-to-station experiments (ubRMSE ranges from 0.045 to 0.051 and R ranges from 0.866 to 0.893). SMCI1.0 generally has advantages over other gridded soil moisture products, including ERA5-Land, SMAP-L4, and SoMo.ml. However, the high errors of soil moisture are often located in the North China Monsoon Region. Overall, the highly accurate estimations of both the year-to-year and station-to-station experiments ensure the applicability of SMCI1.0 to study the spatial–temporal patterns. As SMCI1.0 is based on in situ data, it can be a useful complement to existing model-based and satellite-based soil moisture datasets for various hydrological, meteorological, and ecological analyses and models. The DOI link for the dataset is http://dx.doi.org/10.11888/Terre.tpdc.272415 (Shangguan et al., 2022).

Modeling land surface processes requires complete and reliable soil property information to understand soil hydraulic and heat dynamics and related processes, but currently there is no dataset of soil hydraulic and thermal parameters that can meet this demand for global use. In this study, we propose a fitting approach to obtain the optimal soil water retention parameters from ensemble pedotransfer functions (PTFs), which were evaluated using the global coverage NCSS soil water retention curve database and showed better performance for global applications than the individual PTF and ensemble mean PTFs estimations. Soil hydraulic conductivities are estimated as the median values of multiple PTFs as done in Dai et al. [2013], and the results are shown to perform comparably to the estimates from the existing precision‐advanced models. Soil thermal properties are estimated following the schemes identified by Dai et al. [2019a], which evaluated several highly‐recommended schemes based on their land modeling applications. Using these approaches, we developed two global high‐resolution datasets of soil hydraulic and thermal parameters based on GSDE and SoilGrids soil composition databases. The delivered variables include six basic soil properties, four soil hydraulic parameters in the Campbell [1974] model, five soil hydraulic parameters in the van Genuchten [1980] model, and four soil thermal properties. The delivered datasets are available at a 30″× 30″ geographical spatial resolution, and provide four sets of vertical profiles following the resolutions of SoilGrids, Noah‐LSM, JULES, and CoLM/CLM. The datasets can be used in both regional and global applications.

Soil is an important regulator of Earth system processes, but remains one of the least well-described data layers in Earth system models (ESMs). We reviewed global soil property maps from the perspective of ESMs, including soil physical and chemical and biological properties, which can also offer insights to soil data developers and users. These soil datasets provide model inputs, initial variables, and benchmark datasets. For modelling use, the dataset should be geographically continuous and scalable and have uncertainty estimates. The popular soil datasets used in ESMs are often based on limited soil profiles and coarse-resolution soil-type maps with various uncertainty sources. Updated and comprehensive soil information needs to be incorporated into ESMs. New generation soil datasets derived through digital soil mapping with abundant, harmonized, and quality-controlled soil observations and environmental covariates are preferred to those derived through the linkage method (i.e. taxotransfer rule-based method) for ESMs. SoilGrids has the highest accuracy and resolution among the global soil datasets, while other recently developed datasets offer useful compensation. Because there is no universal pedotransfer function, an ensemble of them may be more suitable for providing derived soil properties to ESMs. Aggregation and upscaling of soil data are needed for model use, but can be avoided by using a subgrid method in ESMs at the expense of increases in model complexity. Producing soil property maps in a time series still remains challenging. The uncertainties in soil data need to be estimated and incorporated into ESMs.

Soil thermal conductivity (STC) is an important physical parameter in modeling land surface processes. Previous studies on evaluations of STC schemes are mostly based on direct measurements of local conditions, and their recommendations cannot be used as a reference in selecting STC schemes for land modeling use. In this work, seven typical STC schemes are incorporated into the Common Land Model (CoLM) to evaluate their applications in land surface modeling. Statistical analyses show that the Johansen [1975] scheme and its three derivatives, Côté and Konrad [2005], Balland and Arp [2005] and Lu et al. [2007], are significantly superior to the other schemes with respect to both STC estimations and their applications in modeling soil temperature and the partitioning of surface available energy in the CoLM. The Balland and Arp [2005] scheme ranks at the top among the selected schemes. Uncertainty analyses based on single‐point and global simulations both show that the differences in STC estimations can induce significant differences in simulated soil temperature in arid/semiarid and seasonally frozen regions, especially at deep layers. The hydrology‐related variables are slightly affected by STC variations, but slight changes in these variables can induce notable changes in soil temperature by altering soil thermal properties. These results emphasize the important role of STC in modeling soil thermodynamics, and suggest the necessity of further developing STC schemes based on land modeling applications. According to the evaluation analyses, we recommend the Balland and Arp [2005] scheme as a more suitable selection for use in land surface models.

Satellite-based leaf area index (LAI) products, such as the MODIS LAI, play an essential role in land surface and climate modeling research, from regional to global scales. However, data gaps and high-level noise can exist, thus limiting their applications to a broader scope. Our previous work has reprocessed the MODIS LAI Collection 5 (C5) product, and the reprocessed data have been widely used these years. In this study, the MODIS C6.1 LAI data were reprocessed to broaden its application as a successor. We updated the integrated two-step method that is used for MODIS C5 LAI and implemented it into the MODIS C6.1 LAI product. Comprehensive evaluations for the original and reprocessed products were conducted. The results showed that the reprocessed LAI data had better performance in validation against reference maps. In addition, the site scale time series of reprocessed data was much smoother and more consistent with adjacent values. The global scale comparison showed that, though the MODIS C6.1 LAI does have improvements in ground validation with LAI reference maps, its spatial continuity, temporal continuity, and consistency showed little improvement when compared to C5. In contrast, the reprocessed data were more spatiotemporally continuous and consistent. Based on this evaluation, some suggestions for using various MODIS LAI products were given. This study assessed the quality of these different versions of MODIS LAI products and demonstrated the improvement of the reprocessed C6.1 data, which we recommended for use as a substitute for the reprocessed C5 data in land surface and climate modeling.

Terrestrial ecosystems have absorbed roughly 30 % of anthropogenic CO₂ emissions over the past decades, but it is unclear whether this carbon (C) sink will endure into the future. Despite extensive modeling and experimental and observational studies, what fundamentally determines transient dynamics of terrestrial C storage under global change is still not very clear. Here we develop a new framework for understanding transient dynamics of terrestrial C storage through mathematical analysis and numerical experiments. Our analysis indicates that the ultimate force driving ecosystem C storage change is the C storage capacity, which is jointly determined by ecosystem C input (e.g., net primary production, NPP) and residence time. Since both C input and residence time vary with time, the C storage capacity is time-dependent and acts as a moving attractor that actual C storage chases. The rate of change in C storage is proportional to the C storage potential, which is the difference between the current storage and the storage capacity. The C storage capacity represents instantaneous responses of the land C cycle to external forcing, whereas the C storage potential represents the internal capability of the land C cycle to influence the C change trajectory in the next time step. The influence happens through redistribution of net C pool changes in a network of pools with different residence times. Moreover, this and our other studies have demonstrated that one matrix equation can replicate simulations of most land C cycle models (i.e., physical emulators). As a result, simulation outputs of those models can be placed into a three-dimensional (3-D) parameter space to measure their differences. The latter can be decomposed into traceable components to track the origins of model uncertainty. In addition, the physical emulators make data assimilation computationally feasible so that both C flux- and pool-related datasets can be used to better constrain model predictions of land C sequestration. Overall, this new mathematical framework offers new approaches to understanding, evaluating, diagnosing, and improving land C cycle models.