Search

• A method to measure the spatial extent of drought at any location was proposed. • The parameters of the method were reasonably specified using sensitivity tests. • The method was tested over a humid region located in southern China. • The difference in spatial lengths of 8 directions showed the anisotropy of drought. • The spatial extents of multi-scale droughts decreased as the time scales increased. The spatial characteristics of droughts are important for understanding drought development and water resource allocation. Most previous studies mainly focus on the spatial display of drought characteristics and the identification of drought spatial extents. Little attention is given to measuring the spatial extent of drought at any location. To this end, a new framework was proposed for measuring the spatial areal extent of drought based on spatial autocorrelation and anisotropy principle. The Global Precipitation Climatology Centre monthly gridded precipitation data with a spatial resolution of 25 km × 25 km were used. The standardized precipitation index on a three-month scale (SPI-3) was applied to characterize meteorological droughts. The base period of 1961–2010 (50 years) was selected for calculating climatology lengths (Cls)/spatial extents (SPEs) of meteorological droughts over the Poyang Lake basin. The parameters of the proposed method, including the drought threshold (SPI ≤ −1), the number of search directions (8), search radius (500 km), distance step (25 km), bandwidth (56.95 km), and angle tolerance (22.5°), were examined using sensitivity tests. The results showed that notable differences in Cls of 8 search directions confirmed the spatial anisotropy of droughts. The statistics of SPEs of the SPI-3 series over the basin were reported. SPEs were mainly between 50 and 100 × 104 km2. The distinctive spatial pattern of the SPEs was revealed and characterized by a high-value belt from the northeast and southwest and two low-value areas. The SPI on other time scales (SPI-1/SPI-6/SPI-12) revealed similar characteristics as SPI-3. Moreover, SPEs are potentially associated with time scales, revealing a decreasing change (or increasing spatial fragmentation) as time scales increased. The proposed method based on anisotropy is reasonable and useful for mining the spatial characteristics of droughts and can be applied in other fields. [ABSTRACT FROM AUTHOR]