Spatial distribution of precipitation and evapotranspiration estimates from WorldClim and Chelsa datasets: Improving long-term water balance at the watershed-scale in the Urabá region of Colombia

Abstract

In this paper, we have evaluated high-resolution spatial gridded climate data from two long-term global datasets, WorldClim V.2.0 and Chelsa V.1.2, in representing variables like precipitation and temperature for the Urabá region of Colombia. Additionally, climate variables from these datasets have been used to estimate evapotranspiration using traditional methods such as the Turc, Hargreaves and Thornthwaite equations. finally, the results of long-term spatial climate characterization are used to apply the water balance equation in the surface at the watershed scale, to obtain the long-term average streamflow of the main streams of the Urabá region; these streamflows are compared with the observations of hydrological stations. We find that the WorldClim and Chelsa rainfall estimates show average differences between 20% and 23% compared to the average annual rainfall in the area from in situ measurements. Both datasets are able to reproduce the rainfall average annual cycle, although Chelsa shows a slightly better performance. Regarding near surface air temperature we find that WorldClim shows a good performance, while Chelsa significantly underestimates the average temperature. finally, we found that the Hargreaves and Thornthwaite methods lead to the best performance in estimating streamflow from the water balance, probably because details of the seasonal behavior of variables like temperature and radiation are explicitly included in these methods. On the other hand, the Turc method yields larger estimates of evapotranspiration and therefore the corresponding derived streamflows are lower than those observed. The good performance of the WorldClim and Chelsa datasets in representing variables like precipitation, temperature, and the derived watershed-scale streamflow, suggest that these long-term global climate datasets can be used to study the spatial distribution of important hydrological variables in the Urabá region of Colombia, and consequently the estimation of average streamflows through the method of the long-term water balance.