青藏高原500m日尺度表层土壤湿度数据集（2015-2023）

表层土壤水分（SSM）是陆气水文过程的重要纽带，但现有微波卫星反演在青藏高原受限于空间分辨率低、轨道覆盖不足及冻融影响，导致数据时空不连续、区域性能不稳定，难以支撑精细化的生态水文研究。为弥补这一缺口，本研究构建了超参数优化的随机森林降尺度框架，利用SMAP粗分辨率SSM反演与融合地形、土壤、植被、TRIMS 全天候地表温度及 TPMFD 气象在内的区域化高分辨率因子之间的非线性关系，生成2015–2023年500 m日尺度表层土壤水分数据集。与近期公开发表的多种SM网格产品及站点观测对比表明，该数据集在相关性（平均R = 0.78）与无偏均方根误差（ubRMSE = 0.0369 m³/m³）方面均表现最佳，能够稳定刻画季节变化及降水响应，在湿润、半湿润及干寒区均具一致优势。因子重要性结果显示，植被指数（EVI）与土壤有机碳（SOC）贡献最高，土壤质地与地表昼夜温差亦为关键因子；非生长季和不同子区域中，近地面湿度等气象因子的影响更为突出。本研究建立的降尺度体系可显著提升青藏高原土壤水分的时空描述能力，填补了青藏高原地区高分辨率土壤水分产品的关键空白，为区域水文预报、冻融过程研究及气候–生态系统相互作用分析提供了重要的数据基础。本次发布的数据集的属性为：空间分辨率为500m，时间步长为每日，单位为m3/m3，数据类型为整数，比例因子为0.001。

Surface Soil Moisture (SSM) serves as a critical link in land-atmosphere hydrological processes. However, current microwave satellite retrievals over the Tibetan Plateau are constrained by low spatial resolution, insufficient orbital coverage, and freeze-thaw effects, resulting in discontinuous spatiotemporal data and unstable regional performance, which hinders support for fine-scale ecohydrological research. To address this gap, this study developed a hyperparameter-optimized random forest downscaling framework that leverages the non-linear relationship between coarse-resolution SSM retrievals from SMAP and regionalized high-resolution factors including topography, soil properties, vegetation, TRIMS all-sky surface temperature, and TPMFD meteorological data, to generate a 500 m daily-scale surface soil moisture dataset spanning 2015–2023. Comparisons with multiple recently published SM grid products and in-situ station observations demonstrate that this dataset achieves the best performance in terms of correlation (average R = 0.78) and unbiased root mean square error (ubRMSE = 0.0369 m³/m³). It can reliably capture seasonal variations and precipitation responses, and exhibits consistent advantages across humid, semi-humid, and arid-cold regions. Factor importance analysis reveals that Enhanced Vegetation Index (EVI) and Soil Organic Carbon (SOC) contribute the most, while soil texture and diurnal surface temperature difference are also key factors. In non-growing seasons and different sub-regions, meteorological factors such as near-surface humidity exert more prominent impacts. The downscaling system established in this study can significantly improve the spatiotemporal characterization capability of soil moisture over the Tibetan Plateau, fill a critical gap in high-resolution soil moisture products for the region, and provide an important data foundation for regional hydrological forecasting, freeze-thaw process research, and climate-ecosystem interaction analysis. The attributes of the released dataset are as follows: spatial resolution of 500 m, daily time step, unit of m³/m³, integer data type, and scale factor of 0.001.