Resumen
This study proposes a method for averaging future climate scenarios from multiple global models through a multivariate statistical analysis. The method gives more weight to the most relevant variables in a specific system, which is associated with the availability of water resources. The proposal considered a climate baseline (precipitation, temperature, relative humidity, wind speed, sunlight) registered in stations near the municipality of Nilo, Colombia (illustrated as a case study), and the future climate scenarios derived from 11 models and 4 scenarios of greenhouse gas emissions (representative concentration pathways, RCPs). Additionally, other variables derived from the current and future water balances (WB) were taken into consideration, such as: deficit (Def), potential evapotranspiration (ETo), real evapotranspiration (ETR), excess (Exc), and other indexes which determine the aridity of a region such as the Lang?s index (IL) and the hydric availability index (HAI). Results suggest that by 2070, precipitation may decrease by 14.0% in the arithmetic average and 25.3% in the weighted average with respect to the current condition (1292 mm yr?1), decreasing to 1111.5 and 964.8 mm, respectively. Additionally, the arithmetic average projects an increase of 2.3 ºC in the average temperature, while the weighted average projects an increase of 2.7 ºC. The proposed methodology is a useful tool to analyze multiple climate change scenarios.