Diseño óptimo de la red pluviométrica utilizando Cokriging y Entropía en la cuenca alta del río Guayllabamba, Distrito Metropolitano de Quito

Abstract

The objective of this study is to propone an optimal rain gauge network for the Guayllabamba river watershed and its sub-basins. The optimization has been made by way of geostatistics models such as Cokriging and Entropy. The current network has 133 rain gauges with a precipitation records between 2001 – 2016 with 50.03% of empty data. The analysis for the conformation of the database started from the classification of the stations by the amount of information presented (in for classes; A, B, C and D). Of the current network, 91 stations with records of more than 5 years were chosen (A, B and C classes). Subsequently, the filling of missing data was done, irregularities were found in 25 rain gauges in the 2001 and 2002 years, therefore, the study period was reduced to 2003 – 2016. Cokriging adjusted the precipitation spatial variability as a regionalized variable, also used elevation as an auxiliary variable to enhance the prediction. As a result, Cokriging prediction and variance maps are generated. Entropy measured the uncertainty level of the network through multivariate distributions, associated with different dependencies, using the principle of maximum entropy. The non-transferred information coefficient was calculated to obtain isoinformation contours. The results reflect a proposal for an optimal rainfall network consisting of 47 existing stations and 30 additional stations proposed. The optimal rainfall network was designed in such a way that the spatial and temporal variability of precipitation is recorded optimally, decreasing high uncertainty areas to obtain reliable information within the study area.