Applicability of five drought indices to drought assessment of Ordos City

In combination with historical drought disaster records from the Meteorological Disaster Management System of the China Meteorological Administration, this paper analyzes the applicability of five drought indices—Precipitation Anomaly in Percentage (PA), Standardized Precipitation Index (SPI), Relative Moisture Index (MI), Meteorological Drought Composite Index (MCI), and Relative Soil Moisture (Rsm) over the 2002 to 2022 period—to the typical drought years and drought-prone areas of Ordos City, Inner Mongolia, China. In addition, the spatiotemporal evolution characteristics of drought in Ordos are revealed through an empirical orthogonal function (EOF) analysis. It was found that (1) the assessment results of the five drought indices differed among the typical drought years. For instance, the MI was suitable for assessing the drought conditions in May and June during the spring/summer drought of 2017, whereas the Rsm was more applicable to the drought conditions in June and July of that year. Meanwhile, the PA and SPI best assessed the actual drought situation during the autumn drought in 2022. (2) In the spatial analysis, the MI and MCI were relatively strongly correlated (moderately high Pearson correlation coefficients) with thenormalized difference vegetation index (NDVI) in the drought-prone areas. In regions with a large proportion of grassland, the proportions of grassland and cultivated land were correlated with PA, SPI, MI, and MCI at the 0.05 significance level. However, as the Rsm calculation accounts for deep soil moisture, Ram is not significantly correlated with the proportion of shallow-rooted plants. From a temporal perspective, the drought conditions assessed by MI and Rsm are more aligned with the actual drought situation during the plant-growth season from May to August. (3) In the EOF1 spatial analysis, the Rsm measure optimized the drought assessment performance; the other indices exhibited consistent variationsoverall, with negative values in parts of the southwest and east of Ordos and positive values in all other regions. In EOF2, the indiceswere negatively valued in the central part of Ordos andpositively valued in the surrounding areas, presenting strong spatiotemporal heterogeneity. Analyzing the variation in time coefficients, it was observed that the southwest and partial eastern areas covered by EOF1 were drought-stricken in 2002-2009, humid in 2010-2013, and again drought-stricken in 2014-2022, consistent with the vegetation distribution in Ordos. Meanwhile, EOF2 was characterized by alternating dry and wet conditions between the central part and the surrounding areas of Ordosat 2-3 a periods. The results provide a scientific basis for monitoring and early warning of droughts, optimal allocation of water resources, and prevention and control of drought disaster risk in Ordos City.