Evaluation of NARR and CLM3.5 outputs for surface water and energy budgets in the Mississippi River Basin

Abstract

[1] The North American Regional Reanalysis (NARR) and Community Land Model (CLM, version 3.5) outputs are analyzed to characterize the surface water and energy budgets in the Mississippi River Basin (MRB). NARR and CLM performance are evaluated with reference to energy flux observations from 16 AmeriFlux sites in MRB. The issue of point-scale observations versus climate model grid cell outputs is addressed by analyzing the spatial variability in long-term monthly precipitation and temperature observations from 71 United States Historical Climatology Network stations in Indiana and Illinois. The model outputs are also evaluated for their ability to capture spatial and temporal variability in total runoff. Compared to average values at 11 AmeriFlux sites in MRB, NARR show higher biases (compared to CLM) in incoming solar radiation (24%), sensible heat flux (27%), and latent heat flux (59%), whereas CLM show smaller biases (compared to NARR) in incoming solar radiation (0.5%), sensible heat flux (−2%), and latent heat flux (11%). Seasonal cycle of observed sensible heat flux in the crop region shows two peaks (bimodal pattern), which is captured by NARR, but CLM do not show any bimodal pattern. Based on 25 years (1980–2004) monthly climatology in MRB, NARR has 11% energy balance closing error (latent + sensible + ground heat flux = 1.11 net radiation) and 12% water balance closing error (evapotranspiration + runoff = 1.12 precipitation), whereas CLM does not have water and energy balance closing errors, primarily due to model design. In comparison to the observed mean annual runoff of 237 mm/yr based on 1988–1999 data in MRB, NARR and CLM mean annual runoff values are 89 mm/yr and 281 mm/yr, respectively. Overall, CLM provides relatively better characterization of surface water and energy fluxes in the MRB compared to NARR.