The generation of the dataset requires the saturated vapor pressure and actual vapor pressure values from various meteorological stations along the Yellow River basin, as well as the Digital Elevation Model (DEM) of the Yellow River basin.

 1) The saturated vapor pressure deficit at each meteorological station is calculated by subtracting the actual vapor pressure from the saturated vapor pressure of the air at a given temperature. The data for saturated vapor pressure and actual vapor pressure are sourced from the National Meteorological Information Center (http://www.nmic.cn/);

 2) During interpolation calculations, the DEM data used as covariates are sourced from the 1km, 500m, and 250m national DEM data (SRTM 90m) provided by the Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences (https://www.resdc.cn/data.aspx?DATAID=123).

 The processing workflow for this dataset is as follows: 1) Based on the data of saturated vapor pressure and actual vapor pressure from 102 meteorological stations along the Yellow River basin, the saturated vapor pressure deficit at each station is calculated by subtracting the actual vapor pressure from the saturated vapor pressure; 2) Using the meteorological interpolation software ANUSPLIN, with geographic elevation as a covariate, batch interpolation processing is performed on the saturated vapor pressure deficit data from each meteorological station to obtain monthly saturated vapor pressure deficit data with a spatial resolution of 1km; 3) Using Python code and the Arcpy library, processes such as projection transformation, clipping, and data format conversion are performed on the data to obtain the final dataset.

 To improve the accuracy of the data, the following measures were taken during the data processing: 1) The DEM data used as covariates have a range larger than the boundaries of the Yellow River basin, in order to reduce interpolation boundary errors; 2) During the interpolation process, the vapor pressure deficit (VPD) data from meteorological stations were subjected to a square root operation to reduce their kurtosis, and then fitted using a thin plate spline function. This improved the interpolation accuracy while ensuring that the processing results were non-negative. After interpolation processing based on the ANUSPLIN software, the Generalized Cross-Validation (GCV) value was very small, and the Model Residual Ratio (MRR) and Signal-to-Noise Ratio (SNR, the ratio of signal degrees of freedom to residual degrees of freedom) were also favorable. The signal degrees of freedom were less than half of the number of stations, and there were no asterisks in the model success rate judgment, indicating that the processed data had reasonable accuracy and reliable quality.