Sensitivity of the RDI and SPEI Drought Indices to Different Models for Estimating Evapotranspiration Potential in Semiarid Regions

摘要

Drought research is of great importance for planning and management of water resources, due to the great impact that droughts have on society and ecosystems. In this study, the effect that using different models for calculating evapotranspiration has on the analysis of droughts in the semiarid region of North Central Mexico is investigated, using climatological information from 14 meteorological stations. Drought was analyzed using the Reconnaissance Drought Index (RDI) and the Standardized Precipitation Evapotranspiration Index (SPEI) at the scales of 3, 6 and 12 months. Eight evapotranspiration models were used: those of Thornthwaite, Hargreaves - Samani, Droogers - Allen, Allen, Dorji, Priestley - Taylor, Makkink and Irmak. According to three of the efficiency indices that were used - the root mean squared error (RMSE), the medium absolute error (MAE) and the concordance index - the Hargreaves - Samani model yields the best evapotranspiration results as compared to the Penman-Monteith model, whereas the models of Thornthwaite and Dorji are the least recommended for this purpose. The non-parametric Wilcoxon test, at a 5 significance level, leads to the conclusion that there are no statistically significant differences between the RDI and SPEI drought indices calculated using the Thornthwaite or the Hargreaves - Samani model. At the three scales of analysis, differences in the RDI index calculated using evapotranspiration estimated with the Thornthwaite or the Hargreaves - Samani model are minimal, but are slightly larger for the SPEI index. Drought events detected with the RDI and SPEI indices are more intense when the Thornthwaite model is used to calculate evapotranspiration instead of the Hargreaves - Samani model. These results may prove valuable in the analysis of droughts, especially in arid and semiarid regions.