Spatiotemporal Patterns of Precipitation in Qaidam Basin from 2000 to 2019 Base on the ANUSPLIN Model

Abstract

The Qaidam Basin is sensitive to the warming and humidification trend of climate change on the Qinghai-Xizang Plateau， and an ecologically vulnerable region.Evaluating its spatiotemporal patterns of precipitation is crucial for the rational utilization of water resources as well as ecological management.However， the scarcity and uneven distribution of meteorological stations within the basin poses a challenge to such analysis.This study was to explore an optimal method for precipitation interpolation in the basin in this context.Specifically， the Australian National University Spline （ANUSPLIN） model was used for the interpolation with different numbers of meteorological stations （9， 20， 40， 60， 80， 100， 120 140， 160 and 180 stations） within and around the basin and 9 kinds of thin disk smooth spline functions （various combinations of independent variables， covariates and spline times）.We screened the optimal number of interpolation stations and the optimal function by accuracy assessments using the data in 2019.The spatiotemporal patterns of precipitation in this region from 2000 to 2019 were then analyzed.The results showed that： （1） The ANUSPLIN model achieved the highest accuracy with 120 meteorological stations and the function of trivariate local thin disk smooth spline （TVPTPS4）.The root mean square error （RMSE）， expected true mean square error （RTMSE） and signal-to-noise ratio （SNR） were less than 0.6 mm， 0.3 mm and 0.25， respectively， which was the lowest among all combinations.（2） Precipitation in the Qaidam Basin had substantial differences in regional distribution and seasonality.Both annual precipitation and seasonal precipitation were abundant in the east and scarce in the west， while the precipitation in summer was the highest， accounting for 62.13% of the annual total value.（3） From 2000 to 2019， both the annual precipitation and seasonal precipitation in the basin showed an increasing trend.The precipitation in summer showed a significant increasing trend， with a maximum growth rate of 5.85 mm·a-1 （P<0.05）.The regions with significant trend accounted for approximately 42.36% of the total area of the basin.The results of this study demonstrate that the AUNSPLIN model can accurately reflect the distribution of precipitation in the Qaidam Basin， comparing with the ordinary Kriging and the inverse distance weighted methods.Obtaining an accurate precipitation patterns is of great theoretical and practical significance for the optimal management of water resources in the region.

Keywords

• qaidam basin
• anusplin model
• precipitation
• spatial pattern
• time changes