Reconstitution stochastique des données hydrologiques de la région andine Bolivienne

Abstract

When stations are dispersed in a wide hydrometeorological network, the use of regression methods is complex and hazardous. This is the case for the Desaguadero river basin, located in the Bolivian highland plain where the completion of missing data in hydrological series requires the use of stochastic models. These models were reviewed in order to evaluate the possibility of eliminating the periodicity and sequential relationship of the hydrological series. This analysis allowed the detection of the effect of groundwater rising from the water table and, to a lesser extent, that of hot springs. These conditions rendered the modelisation of the system complex. The use of stochastic models to reconstitute a new series statistically alike to the historic series was successful. The statistical characteristics of the new series are similar to those of the observed series.