Effects of Uncertainty in Climate Inputs on Simulated Evapotranspiration and Runoff in the Western Arctic

Abstract

Hydrological models require accurate precipitation and air temperature inputs in order to adequately depict water fluxes and storages across Arctic regions. Biases such as gauge undercatch, as well as uncertainties in numerical weather prediction reanalysis data that propagate through water budget models, limit the ability to accurately model the terrestrial arctic water cycle. A hydrological model forced with three climate datasets and three methods of estimating potential evapotranspiration (PET) was used to better understand the impact of these processes on simulated water fluxes across the Western Arctic Linkage Experiment (WALE) domain. Climate data were drawn from the NCEP?NCAR reanalysis (NNR) (NCEP1), a modified version of the NNR (NCEP2), and the Willmott?Matsuura (WM) dataset. PET methods applied in the model were Hamon, Penman?Monteith, and Penman?Monteith using adjusted vapor pressure data. High vapor pressures in the NNR lead to low simulated evapotranspiration (ET) in model runs using the Penman?Monteith PET method, resulting in increased runoff. Annual ET derived from simulations using Penman?Monteith PET was half the magnitude of ET simulated when the Hamon method was used. Adjustments made to the reanalysis vapor pressure data increased the simulated ET flux, reducing simulated runoff. Using the NCEP2 or WM climate data, along with the Penman?Monteith PET function, results in agreement to within 7% between the simulated and observed runoff across the Yukon River basin. The results reveal the high degree of uncertainty present in climate data and the range of water fluxes generated from common model drivers. This suggests the need for thorough evaluations of model requirements and potential biases in forcing data, as well as corroborations with observed data, in all efforts to simulate arctic water balances.