Stochastic modelling of spatial variability of water quality in a lacustrine ecosystem [Modelaje estoc�stico de la variabilidad espacial de la calidad de agua en un ecosistema lacustre]

Abstract

Since the Chapala lake is an important water source for Guadalajara city, its adequate management is imperative. This matter implies the water quality evaluation, including physical, chemical, and biological parameters, that, when evaluated separately, do not give an adequate diagnostic. Thus, in this project a Water Quality Index (WQI) was used to evaluate water quality in the Chapala lake. Fifteen plots were sampled, during five seasons (from the rainy season to the dry season), in order to consider the temporal variability in water quality. However, a single evaluation of WQI is not enough to define management strategies and the spatial variation of WQI must be defined. Therefore, ordinary kriging (OK), a stochastic interpolation technique, was used to convert the punctual data of WQI (for each season) into continuous surfaces. In order to validate the OK advantages, this method was compared with the weighted inverse distance technique (1st and 2nd power), a deterministic process. Based on the results, the spatial distribution of WQI could be modeled. Thus, in general, OK was the best alternative. However, when WQI showed a global trend, the OK spatial continuity was not enough to ensure that this parameter is better than the deterministic alternatives. This suggests that future projects should not only use geostatistic techniques that consider the local variability (such as OK), but also techniques that consider global variability, such as universal kriging. Although better results are not guaranteed, this will allow a deeper understanding of the WQI spatial behavior, taking into account that each case is unique and that comparison among different techniques is interesting.