Resumo

NUNEZ LOPEZ, Daniel et al. Using regression models for spatially interpolated monthly average rainfall in the Conchos River Basin. Rev. Mex. Cienc. Agríc [online]. 2014, vol.5, n.2, pp.201-213. ISSN 2007-0934.

In the present study, we analyzed monthly precipitation data from 110 weather stations located within and around Rio Conchos Basin (CRC) in order to reliably represent the spatial distribution of mean monthly precipitation (MMP) for each month of the year. With information from 60% of randomly selected stations were adjusted multiple linear regression models (MLRM) by MMP steps to predict based on the elevation of the terrain, the proximity of sea areas and the geographical location of the stations. The MLRM used to spatially interpolate the MMP; yielding monthly maps were calibrated according to the residuals. Statistical validation tests were conducted before and after the spatial calibration, using the remaining 40% of stations not considered in the model fitting process. The proportion of variance attributable to the predictors of MLRM comprising the summer period (June to September) ranged between 71 and 76%, while for models of the winter period (December and January) remained close to 50%. The validation tests showed statistically significant improvements in the reliability after calibrating MMP maps, resulting the months between May and September and November to January period, as the most reliable maps spatially represent the MMP.

Palavras-chave : Conchos River Basin; modeling efficiency; reliability.