Spatiotemporal characteristic and forecast of drought in northern Xinjiang, China

Understanding monitor accurately and predicting drought conditions are necessary to prevent economic losses and natural disasters. Now, there is still a lack of comprehensive assessment of drought prediction in Xinjiang. In order to better predict the future drought trend, firstly we analyzed the trends of SPEI and the magnitude and frequency of drought in selected surface meteorological stations from 1960 to 2017. Secondly, we compared drought trends and spatial distribution characteristics of SPEI and PDSI. Based on historical data from 1960 to 2017, the paper predicted the trends of SPEI and PDSI in the next five years. The results show that the winter SPEI has risen significantly, and there was a SPEI value mutation in 1994. The Spring, Summer and Autumn SPEI show an insignificant downward trend. It is relatively stable on other time scales , and there is no significant sudden change. It is inferred thatthe increasing trend of wetting that may occur under different time scales at NRX in 2021, winter and annual scales will continue to be humidified, and spring, summer, and autumn will continue to be arid. Droughts in NRX occur at least once every three years, which is a high frequency event. The prediction results of the artificial neutral network (ANN) method based on multilayer feed forward network show that probability of humidification at NRX is more likely in the future, and the prediction results of wavelet and Hurst index are basically consistent with that of ANN method.

Automated summary

The study provides a comprehensive analysis of drought prediction in Xinjiang, including trends of SPEI and PDSI in selected surface meteorological stations from 1960 to 2017, as well as the prediction for the next five years. The results suggest a potential increase in wetting trend at NRX in various time scales, with winter and annual scales continuing to be humidified and spring, summer, and autumn remaining arid. Droughts in NRX occur at least once every three years, indicating a high frequency event. Prediction methods such as artificial neural network (ANN), wavelet, and Hurst index show consistent results in forecasting future humidification probabilities at NRX.