Model performance and parameter behavior for varying time aggregations and evaluation criteria in the WASMOD-M global water balance model

Abstract

Global discharge estimates commonly range between 36.500 km3 a-1 and 44.500 km3 a-1, i.e., around 20%, and continental estimates differ much more. Data uncertainties are assumed to be a main cause of simulated runoff uncertainties, but model performance must also be addressed. The parsimonious WASMOD-M global water balance model, using limited input data, was used to assess data and model uncertainty (contrary to models using much data but being modestly or not at all calibrated). A Monte Carlo technique based on 15,000 parameter value sets was used to evaluate the model against four criteria: observed snow and monthly, annual, and long-term discharge. WASMOD-M was overparameterized when evaluated only against long-term average discharge but not against monthly discharge, and its snow algorithm could be simplified. Sequential calibration is suggested for confining the behavioral parameter space and minimizing model equifinality starting with snow, followed by long-term volume error, and ending with discharge dynamics. Copyright 2009 by the American Geophysical Union.