Hydrologic response modelling: comparing the VIC and PCR-GLOBWB models

Abstract

Accurate information about water reserves is crucial in a world where the water demand will increase over the coming century. Global Hydrological Models (GHMs) can play a large role supporting good decision making by predicting the available water resources. GHMs have different features because of their background, each with their own strengths and weaknesses. VIC and PCR-GLOBWB are both capable of simulating the global water balance, but VIC has the potential to simulate the hydrological balance on a local scale more accurately because it solves the energy balance. This research will look at the effect that an energy balance in a GHM has on the accuracy of the output. Using the WFDEI climate forcing, PCR-GLOBWB and two versions of VIC (with and without energy balance) are used to simulate the global water balance. The model output is compared to global evapotranspiration (ET), snow water equivalent (SWE) and soil moisture datasets, as well as to discharge measurements of the Amazon, Brahmaputra, Mackenzie, Magdalena, Mississippi and the Nile. The results show that PCR-GLOBWB has higher Kling-Gupta, Nash-Sutcliffe and correlation scores for ET and equal scores to VIC for SWE. VIC, on the other hand, has higher accuracy scores for discharge in five of the six rivers and for soil moisture. The effect of an energy balance is small, as VIC-EB performs similar to VIC-WB and the results indicate that the calibration of VIC plays a larger role in the higher accuracy of VIC for the discharge and soil moisture than the energy balance. The sensitivity of PCR-GLOBWB to different climate forcing, potential ET and resolution changes was also tested. Using CRU precipitation instead of GPCC precipitation leads to significantly lower discharges, but comparable local accuracy scores. Use of the Penman-Monteith potential ET equation results in more accurate results for the entire water balance. Running PCR-GLOBWB on a higher resolution (5 arcminutes) leads to lower accuracy.