Flash drought is one kind of extreme hydro-meteorological events that frequently occur worldwide in recent years. Different from traditional slowly-evolving droughts, flash droughts unusually have rapid onset, threatening food and agricultural security, and the health of the eco-environment. To compare the spatio-temporal characteristics of flash drought and slowly evolving drought, the soil moisture data were extracted from the ERA-Interim reanalysis product (at a depth of 0-100 cm) with a spatial resolution of 0.25°and a temporal coverage from 1979 to 2018. Records of the soil moisture data were converted into weekly values. Seven candidate probability distribution functions (including Gamma, Beta, logarithmic, logical logarithmic, Weibull, extreme value, and generalized extreme value distributions) were employed to fit the weekly data series. The optimal probability distribution function for calculating soil moisture percentile was selected according to the Kolmogorov-Smirnov test (KS test) and two evaluation coefficients, i.e., the root mean square error and bias. For drought event identification, the 40% and 20% of soil moisture percentile were used as the upper limit and the lower limit, respectively, to ensure the identified drought events really falling into drought associated with vegetation stress. The onset duration was determined as the period when soil moisture percentile declines persistently and maintains a rather low value. The differences between flash droughts and slowly-evolving droughts in terms of drought frequency and drought area were analyzed. The relationships between flash droughts and slowly-evolving droughts were investigated based on the spatio-temporal patterns of two typical drought events. The results showed that droughts with long duration (more than 26 weeks) were more frequent in the north of the Yangtze River. In contrast, the south of the Yangtze River suffered few droughts and short-term drought. Different from drought events with short duration, the time series of affected area percentage for drought events with long duration (26 weeks or longer) presented significant increments from 2005 to 2015, especially in 2010 when more than 40% of the area suffered drought, suggesting a general trend of droughts with long duration and enlarged area extent. During the drought onset stage, the duration of drought in the south region was shorter than that of the north region. Typical flash drought events occurred in the middle and eastern part of Inner Mongolia, southwest China, the upper reaches of the Pearl River, and lower reaches of the Yangtze River, while slowly-evolving droughts happened mainly in the northwest China, and some areas in Xinjiang. The area percentage of the flash drought was comparable to that of the slowly-evolving drought before 2000, after that, the affected area of slowly-evolving drought expanded significantly up to twice that of flash drought. The spatio-temporal migration paths of the 2010 autumn drought and the 2017 summer drought showed that the flash drought and slowly-evolving drought may develop independently, and may also present a co-existing relationship. Finally, the in-situ soil moisture measurements from the International Soil Moisture Network (ISMN) were employed for validation of the results. The spatial pattern of moisture was similar to that of the ERA-Interim reanalysis product. For example, soil moisture in humid and semi-humid zones decreased much quickly than in arid and semi-arid zones during a drought episode. This may be related to the climate condition, soils, and vegetation characteristics of the underlying surface. The soil moisture data of the ERA-Interim reanalysis product is capable of capturing the moisture dynamics during the onset stage of drought, and could be an alternative for flash drought monitoring and assessments. The results can provide some important information for understanding the mechanisms of drought and the development of drought monitoring techniques. © 2021, Editorial Department of the Transactions of the Chinese Society of Agricultural Engineering. All right reserved.
