Different drought indices showed different variations and applicability to dendrochronological studies

Abstract

In drought studies based on tree-ring data, there is currently no unified drought index that can capture the drought signals contained in tree rings from different climates, habitats, and species. This makes it difficult to compare the results of numerous studies. This paper compared data charactering wet and dry variations in the Ulan Buh Desert in northwestern China from 1962 to 2017, as indicated by different drought indices. We selected the most commonly used drought indices in tree-ring research: precipitation (Pre), self-calibrating Palmer aridity index (scPDSI), and standardized precipitation evapotranspiration index (SPEI), based on observed meteorological data (-obs) and grid data from the Climatic Research Unit (-cru). The results showed that although the different drought indices were significantly correlated with each other, they showed different linear trends. Correlation analyses with the tree-ring width chronology in the study area showed that the applicability of drought indices based on observed data was better than that of drought indices based on CRU data. The correlation between SPEI-cru and chronology was much lower than that of SPEI-obs, scPDSI-obs, scPDSI-cru, Pre-obs and Pre-cru. The extreme drought years identified by SPEI-cru also differed from other indices. These results suggest that SPEI-cru is not applicable to dendrochronological studies in the Ulan Buh Desert, mainly due to the differences between the CRU data and the measured data, and the high sensitivity of SPEI to potential evapotranspiration. This case study illustrates that tree-ring based drought studies must be based on an assessment of the applicability of different drought indices.