Evaluation of Terrestrial Water and Energy Budget Components over the St. Lawrence River Basin

Abstract:

This study systematically evaluates the hydrologic potential of two global reanalysis products, namely the European Centre for Medium-Range Weather Forecasts Re-Analysis Generation 5 product (ERA-5) and the Modern-Era Reanalysis for Research and Application version 2 (MERRA-2), and one regional precipitation reanalysis, namely Canadian Precipitation Analysis (CaPA), products over the St. Lawrence river basin through a suite of uncoupled land surface model simulations for a 6-year period from 2014 to 2019. Four combinations of forcing datasets are considered for LSM simulations; they are ERA-5 and MERRA-2 and the other two are formed by replacing precipitation in ERA-5 and MERRA-2 by the CaPA precipitation. Evaluations are focuses on multiple terrestrial water and energy components including soil moisture profiles, evapotranspiration, snow depth, and sensible and latent heat fluxes, separating for a warm (May to October) and a cold (the rest of the year) season. Point-based and pattern-based comparisons are conducted for the variables with ground station observations and other state-of-the-art remote sensing/reanalysis products, respectively. Our results indicate good agreements with the ground-based observations for the sensible and latent heat fluxes and snow depth simulations. The simulated evapotranspiration and snow depth also demonstrate high consistency with other remote sensing and reanalysis products. The warm season simulations outperform the cold season simulation for the variables regardless the forcing used to produce the simulations. The soil moisture simulations indicate a notable bias to both the ground station observations and remote sensing products but capture the timing in both cases. Product-wise comparisons suggest that ERA-5 and MERRA-2 simulations are comparable with ERA-5 slightly outperforms MERRA-2 for more categories. Also, replacing the ERA-5 and MERRA-2 precipitation by CaPA can improve the performance in most of the variables simulations.