Combined use of meteorological drought indices at multi-time scales for improving hydrological drought detection

Prediction of hydrological drought in the absence of hydrological records is of great significance for water resources management and risk assessment. In this study, two meteorological drought indices, including standardized precipitation index (SPI) and standardized precipitation evapotranspiration index (SPEI) calculated at different time scales (1 to 12months), were analyzed for their capabilities in detecting hydrological droughts. The predictive skills of meteorological drought indices were assessed through correlation analysis, and two skill scores, i.e. probability of detection (POD) and false alarm rate (FAR). When used independently, indices of short time scales generally performed better than did those of long time scales. However, at least 31% of hydrological droughts were still missed in view of the peak POD score (0.69) of a single meteorological drought index. Considering the distinguished roles of different time scales in explaining hydrological droughts with disparate features, an optimization approach of blending SPI/SPEI at multiple time scales was proposed. To examine the robustness of the proposed method, data of 1964-1990 was used to establish the multiscalar index, then validate during 2000-2010. Results showed that POD exhibited a significant increase when more than two time scales were used, and the best performances were found when blending 8 time scales of SPI and 9 for SPEI, with the corresponding values of 0.82 and 0.85 for POD, 0.205 and 0.21 for FAR, in the calibration period, and even better performance in the validation period. These results far exceeded the performance of any single meteorological drought index. This suggests that when there is lack of streamflow measurements, blending climatic information of multiple time scales to jointly monitor hydrological droughts could be an alternative solution.