Estimación de precipitación espacial mediante correlación con variables secundarias y la ayuda de tecnologías de información geográficas

Abstract

This study focuses on implementation and validation of four spatial interpolation techniques, Thiessen (TH), inverse distance weighted (IDW), kriging (KO) and cokriging (CK) at values of monthly and annual rainfall registered in 23 rain gauges of Paute River Basin in the southern Ecuadorian Andes, and seek topographic variable that best correlates with the precipitation to be used as auxiliary cokriging interpolation technique, using Geographic Information Systems (GIS) and geostatistics.

Seventy six topographic variables were calculated, of which 50 were obtained from Digital Elevation Model and 26 are scalar products between some of the first variables. The variables showed values very poor correlation with rainfall (r<0.3) or at most a moderate correlation (r<0.6) only for 1-3 months a year. These low correlations led to seek other variables called orographic indices IO1 and IO2 defined in this study as expressions that combine two orographic elements the barrier effect of the range and the location of studied gauge on area of mountain or valley. IO1 has very good correlation with the annual and monthly precipitation in the period May to September (r>0.91), while IO2 gives acceptable correlations (r>0.56) for the months not explained by IO1.

Cross-validation of the interpolation methods shows that the best method for the months January, February, April, October, November and December is CK, while for the months March, May, June, July, August and September and the annual average is KO.