AbstractDrought is a common natural disaster that greatly affects the crop yield and water supply in China. However, the spatiotemporal characteristics of drought in China are not well understood. This paper explores the spatial and temporal distributions of droughts in China over the past 40 years using multiscale standardized precipitation evapotranspiration index (SPEI) values calculated by monthly precipitation and temperature data from 612 meteorological stations in China from 1980 to 2019 and combines the space-time cube (STC), Mann-Kendall (M-K) test, emerging spatiotemporal hotspot analysis, spatiotemporal clustering and local outliers for the analysis. The results were as follows: 1) the drought frequency and STC show that there is a significant difference in the spatiotemporal distribution of drought in China, with the most severe drought in Northwest China, followed by the western part of Southwest China and the northern part of North China. 2) The emerging spatiotemporal hotspot analysis of SPEI6 over the past 40 years reveals two cold spots in subregion 4, indicating that future droughts in the region will be more severe. 3) A local outlier analysis of the multiscale SPEI yields a low-low outlier in western North China, indicating relatively more severe year-round drought in this area than in other areas. The low-high outlier in central China indicates that this region was not dry in the past and that drought will become more severe in this region in the future.
Nonnegative Coupled Matrix Tensor Factorization for Smart City Spatiotemporal Pattern Mining