Improvement of comprehensive meteorological drought index and its applicability analysis

Drought is one of the frequent and destructive natural disasters around the world. Drought index can be used to represent the dry and wet condition of the land surface, further to serve as a sound basis for drought monitoring, assessment, and early warning in the subsequent drought research. Therefore, it is of great significance to select the appropriate indicators, in order to understand the real drought scenario, and thereby to predict the trend of drought. Generally, a comparative analysis of several indicators was also made to accurately monitor the regional drought. In this study, a comprehensive meteorological drought index (CI) was modified to make it more suitable for the accurate monitoring the drought. This field study was taken Hubei Province, China, as a typical research area. The reason is that most droughts often occur in various regions and seasons for Hubei Province, located in the central part of China, particularly belonging to a subtropical monsoon climate zone, with the uneven distribution of precipitation during the year, while the large inter-annual variability. The data captured from the China Meteorological Forcing Dataset from 1982-2015, as well as the drought data of Hubei in recent years. Various methods, including the runs theory, correlation analysis and Mann-Kendall (M-K) trend test, were used to compare the accuracy of several common drought indicators, and the evolvement of drought. In the research on the applicability of drought indexes, the results demonstrated that: (1) The improved comprehensive drought index (CIn) can identify the major droughts in this area as a useful indicator. Among 15 major droughts, 13 of them can be identified when taking that indicator into consideration. (2) Compared with CI, the CIn index was calculated using the monthly data of precipitation and evaporation, indicating the index changed more steadily with a better correlation with long-term SPEI. The determinant coefficient of CIn and SPEI6 reached 0.45, while that of CI and SPEI6 was only 0.19. (3) Compared with CI index and two SPEI indexes with different time scales, two degrees of drought area that obtained by CIn have the best correlation with the drought area in the statistical data, where the correlation coefficients with the drought-affected area and drought-disaster area were 0.69 and 0.67, respectively. However, none of the other drought indices exceeded 0.65. (4) The spatial distribution of drought duration (DD) and drought severity (DS) obtained by CIn and the runs theory were also consistent with the previous studies. However, the drought duration and severity distribution based on SPEI1 and SPEI6 were not consistent with the actual situation. According to the analysis of temporal and spatial evolution trend of drought, the drought area in Hubei province was slightly increasing, where the drought trend was obvious in northern, northwest, and southwest Hubei, together with the Jianghan Plain, showing a significance level of 0.05. In the M-K test, the changed area proportions of significant drought trend were 40.19% and 53.99% in the CIn and SPEI6, much higher than 26.40% and 18.93% of CI and SPEI1, indicating that the CIn and SPEI6 were more sensitive to the long-term change of climate. Therefore, the appropriate drought monitoring indicators can be selected, and thereby to be considered in trend analysis. This finding can offer a significant reference for the selection of meteorological drought index, and further for the effective implementation of regional meteorological drought monitoring. © 2020, Editorial Department of the Transactions of the Chinese Society of Agricultural Engineering. All right reserved.