Construction of 3D drought structures of meteorological drought events and their spatio-temporal evolution characteristics

Drought events often exhibit distinct spatio-temporal heterogeneity and coupling properties at different spatio-temporal scales. In previous studies, the spatial and temporal evolution of drought events were considered in a separate manner, and some mild drought events were likely to be missed in the identification process. This paper proposed a new drought structure extraction method for extracting all drought events from long time series data and constructing the 3D drought structure, and analyzed their spatio-temporal distribution patterns and characteristics under the interweaving of spatial and temporal dimensions. The 3D threshold isosurface was plotted to cluster temporally and spatially interconnected grid points, so that a 3D drought structure was formed to represent a drought event and describe the spatio-temporal distribution pattern in long time series and the spatial–temporal variability in the development process. On this basis, the geometric and physical characteristics inside and outside the drought structure were analyzed using seven characteristic indicators, such as drought duration, drought area, drought mass, drought volume, drought density, drought aggregation index, and longitude and latitude coordinates of drought centroid. The meteorological drought events in the Huai River basin for the period 1961–2015 were identified and their spatio-temporal distribution pattern and variation were characterized. Finally, Pearson correlation analysis, M−K trend analysis and wavelet analysis were performed to quantify the evolution of drought events. The results show that (1) the drought structure extraction method can reflect the spatio-temporal evolution characteristics and the development trend of multiple drought events according to the outer contour, the internal color and the migration of the location of the drought structure; (2) the drought area in the Huai River basin shows a decreasing trend, but the severity shows an increasing trend due to the increase in drought duration and intensity, and thus the shape of the drought structure is slender and the spatial distribution of drought events becomes more discrete; and (3) there is a good agreement in the identification between the 3D clustering method and the drought structure extraction method. However, the drought structure extraction method can accurately identify mild drought events that are easily overlooked in the 3D clustering method. The proposed drought structure extraction method has the potential to review the development and evolution trend of drought events and thus it is of practical significance for drought mitigation.