A dataset of downscaled monthly precipitation at 1 km resolution in Guangxi, China and ASEAN regions from 2001 to 2020

Abstract

Located in the southeastern Asia, Guangxi, China and the ASEAN regions have a subtropical monsoon climate with abundant rainfall and frequent floods, which has a great impact on social and economic activities. The precipitation data with high precision and high spatial and temporal resolution are of great significance for industrial and agricultural production, water conservancy development, drought and flood monitoring and prevention, and ecological environment protection. In this paper, we used the Global Precipitation Measurement Mission precipitation data (GPM IMERG) from 2001 to 2020 as the dependent variable in combination with the enhanced vegetation index (EVI), surface evapotranspiration (ET), land surface temperature (LST) from MODIS data and elevation (ELV) from ASTER data as explanatory variables. Then we introduced a geographically weighted regression (GWR) model to construct an annual scale model, depicting the spatial variation of 10-km satellite precipitation with the influence of geographical environmental conditions. Five types of kernel functions were adopted in our GWR model, including the gaussian, exponential, bisquare, tricube, and boxcar kernel function. The optimal kernel function was selected based on the correlation coefficient, the root means square error and bias. Thus, we established the dataset of downscaled monthly precipitation in Guangxi, China and ASEAN regions from 2001 to 2020 by extrapolating the GWR model with input variables at 1 km resolution. The dataset of 1-km monthly precipitation from 2001 to 2020 was also generated by the proportional coefficient method. Moreover, ground observation data from 2,679 stations during 2001–2020 were used for verification. The correlation coefficient, root mean square error and deviation were 0.792, 74.610mm and -0.122%, respectively. The results show that this dataset can reflect the precipitation spatial and temporal distribution and its variations at 1-km resolution in detail, and could be potentially promising for ecological environment, hydrological management, flood prediction and other relevant fields.