Uncertainties in the estimation of Thornthwaite's Aridity and Moisture Indices in Greece over the last century using ground and gridded datasets

Abstract

During the last decades several gridded climatic datasets have been developed offering thus long-term, continuous, and spatially uniform records of key hydrometeorological parameters. Few studies, however, have rigorously evaluated the accuracy and reliability of these datasets, especially in regions characterized by complex topography and sparse observational data, such as the Mediterranean. Moreover, even fewer have explored how uncertainties inherent in these datasets affect derived climatic indices. To address this gap, the present study deals with the application of ground station measurements and a gridded dataset (CRU_TS 4.04) to estimate the aridity index (AI) and Thornthwaite's Moisture index (Im) over the Greek peninsula for three different consecutive climatic periods. Ground station data estimations are compared against the outcomes of the respective gridded datasets. Results show that the application of gridded data led to AI values increase in many areas, altering the aridity classification from semi-arid (SA) to sub-humid (SH) or humid (H) categories. Further, the application of gridded data led to both underestimations and overestimations of the Im, depending on the site: more dry conditions along the western coast (in most cases) and the eastern Aegean islands (with significant differences in some of the islands) and more humid conditions over the northwestern mainland and the southern part along with the Aegean islands. These results suggest that while gridded datasets always offer an alternative, careful consideration is required when they are used in regions characterized by complex topography (heterogenous terrain, site continentality and extensive land-sea contrast) and also temporal climatic variability.