中国逐日0.25度雪水当量融合产品（1980–2020）

雪水当量（SWE）是表征积雪水资源属性的重要参数和全球气候变化的敏感要素之一。基于梯度提升模型（XGBoost）联合5种SWE产品（GlobSnow，ERA5-Land，ERA-Interim，GLDAS，1980–2020年中国逐日雪水当量产品数据集）和时空协变量（土地覆盖、气温、积雪相关特征、时间、空间）构建时空XGBoost模型，通过融合现有SWE产品优势，辅以地形、时间等协变量信息，综合考虑SWE的时空异质性及其与多重影响因素之间的非线性关系，实现高精度雪水当量估算。利用647个实测站点数据对融合雪水当量产品进行精度验证，基于随机交叉和年分交叉的验证精度分别为R=0.77,0.70, MAE=7.54, 8.62mm, RMSE=12.29,13.73mm。基于该模型生产了1980–2020年0.25度分辨率的中国逐日雪水当量产品。该数据集可为积雪水资源管理提供更准确的雪水当量数据。

Snow Water Equivalent (SWE) is a critical parameter characterizing the water resource properties of snow cover and one of the sensitive indicators of global climate change. A spatiotemporal XGBoost model was developed based on the eXtreme Gradient Boosting (XGBoost) framework, combined with five SWE products (GlobSnow, ERA5-Land, ERA-Interim, GLDAS, and the daily SWE product dataset of China spanning 1980–2020) and spatiotemporal covariates including land cover, air temperature, snow-related features, time and space. By integrating the strengths of existing SWE products and supplementing with covariate information such as topography and time, the model comprehensively considers the spatiotemporal heterogeneity of SWE and the nonlinear relationships between SWE and multiple influencing factors, enabling high-precision SWE estimation. The fused SWE product was validated using data from 647 in-situ measurement stations. The validation accuracies via random cross-validation and year-wise cross-validation were R=0.77, 0.70, MAE=7.54, 8.62 mm, RMSE=12.29, 13.73 mm, respectively. Based on this model, a daily SWE product over China with a spatial resolution of 0.25° spanning 1980–2020 was generated. This dataset can provide more accurate SWE data for snow cover water resource management.

搜集汇总

背景与挑战

背景概述

该数据集是一个中国范围内1980年至2020年的逐日雪水当量融合产品，空间分辨率为0.25度。它基于梯度提升模型融合了多种SWE产品和时空协变量，实现了高精度估算，并通过647个站点验证，精度较高（如R=0.77）。数据以tif格式提供，但存在部分日期缺失，适用于积雪水资源管理和气候变化研究。