Identification and three-dimensional evolution characterization of long duration and high severity drought in Xinjiang based on precipitable water vapor

Abstract

Due to the scarcity of measured precipitation data in arid zones and the numerous factors affecting the uncertainty of precipitation data from multiple sources, understanding precipitation patterns is limited, and drought monitoring is constrained. Whereas water vapor content is closely related to precipitation, changes in water vapor can be an important indicator of drought development. Therefore, this paper takes Xinjiang as the study area, introduces precipitable water vapor, and constructs two new drought indices, the standardized and nonparametric standardized precipitable water indices (SPWI and NSPWI), using parametric and nonparametric kernel density estimation methods. Aiming at the drought characteristics of long duration and intensity in the study area, an identification method for long-duration and high-severity drought (LHD) events was proposed using the three-dimensional clustering method and the Copula function. The three-dimensional evolution characteristics and centroid migration patterns of LHD events were then analyzed.The results indicate that: (1) The newly constructed drought index can be used for drought monitoring in Xinjiang, and the performance of SPWI is better than that of NSPWI. (2) Based on the three-dimensional clustering method, which can better identify the drought events in the study region, it was found that the droughts were mostly concentrated in 1960–1987, and there was a trend from dry to wet. (3) The LHD events were identified based on the joint distribution probability of drought duration and severity greater than 75 %, and it was found that droughts originated in the southern and died out in the western part of the study region, that droughts were likely to be more frequent the deeper inland and farther away from the ocean, and that droughts lasted for longer periods and had a wider impact area; mountainous areas are more profoundly affected by the LHD events than plains and deserts areas. (4) LHD events have a wider centroid migration distance, and their drought development process is more complex, with spatial and temporal distributions. The drought index that introduces precipitable water vapor and the LHD event identification method proposed in this study can provide some guidance for the enhancement of drought prevention and mitigation capabilities in arid regions.