Getting water right: A case study in water yield modelling based on precipitation data

摘要

Water yield is a key ecosystem service in river basins and especially in dry regions around the World. In this study we carry out a modelling analysis of water yields in the Chubut River basin, located in one of the driest districts of Patagonia, Argentina. We focus on the uncertainty around precipitation data, a driver of paramount importance for water yield. The objectives of this study are to: ⅰ) explore the spatial and numeric differences among six widely used global precipitation datasets for this region, ⅱ) test them against data from independent ground stations, and ⅲ) explore the effects of precipitation data uncertainty on simulations of water yield. The simulations were performed using the ecosystem services model InVEST (Integrated Valuation of Ecosystem Services and Tradeoffs) with each of the six different precipitation datasets as input Our results show marked differences among datasets for the Chubut watershed region, both in the magnitude of precipitations and their spatial arrangement Five of the precipitation databases overestimate the precipitation over the basin by 50% or more, particularly over the more humid western range. Meanwhile, the remaining dataset (Tropical Rainfall Measuring Mission - TRMM), based on satellite measurements, adjusts well to the observed rainfall in different stations throughout the watershed and provides a better representation of the precipitation gradient characteristic of the rain shadow of the Andes. The observed differences among datasets in the representation of the rainfall gradient translate into large differences in water yield simulations. Errors in precipitation of + 30% (-30%) amplify to water yield errors ranging from 50 to 150% (-45 to -60%) in some sub-basins. These results highlight the importance of assessing uncertainties in main input data when quantifying and mapping ecosystem services with biophysical models and cautions about the undisputed use of global environmental datasets.