Abstract
Abstract. Large lakes and reservoirs play important roles in modulating regional hydrological cycles and climate; however, their representations in coupled models remain uncertain. The existing lake module in the Weather Research and Forecasting (WRF) system (hereafter WRF-Lake), although widely used, did not accurately predict temperature profiles in deep lakes mainly due to its poor lake surface property parameterizations and underestimation of heat transfer between lake layers. We therefore revised WRF-Lake by improving its (1) numerical discretization scheme; (2) surface property parameterization; (3) diffusivity parameterization for deep lakes; and (4) convection scheme, the outcome of which became WRF-rLake (i.e., revised lake model). We evaluated the off-line WRF-rLake by comparing simulated and measured water temperature at the Nuozhadu Reservoir, a deep reservoir in southwestern China. WRF-rLake performs better than its predecessor by reducing the root-mean-square error (RMSE) against observed lake surface temperatures (LSTs) from 1.4 to 1.1 ∘C and consistently improving simulated vertical temperature profiles. We also evaluated the sensitivity of simulated water temperature and surface energy fluxes to various modeled lake processes. We found (1) large changes in surface energy balance fluxes (up to 60 W m−2) associated with the improved surface property parameterization and (2) that the simulated lake thermal structure depends strongly on the light extinction coefficient and vertical diffusivity. Although currently only evaluated at the Nuozhadu Reservoir, we expect that these model parameterization and structural improvements could be general and therefore recommend further testing at other deep lakes and reservoirs.
Highlights
Inland waters such as lakes and reservoirs differ from their surrounding land with altered albedo, larger thermal conductance and heat capacity, and lower surface roughness, and different radiative and thermal properties
We evaluated the improved lake model, WRFrLake, at the Nuozhadu Reservoir, a deep reservoir in southwestern China, and conducted sensitivity experiments to evaluate the effect of dominant processes and parameters on simulated lake water temperature and surface heat fluxes
In comparing Weather Research and Forecasting (WRF)-Lake simulations (CTL) to the observations (Fig. 5), the lake surface temperatures (LSTs) are underestimated most of the time with a mean bias error (MBE) of −0.69 ◦C
Summary
Inland waters such as lakes and reservoirs differ from their surrounding land with altered albedo, larger thermal conductance and heat capacity, and lower surface roughness, and different radiative and thermal properties These lake properties can exert significant influence on local and regional climate and are important in understanding lake– atmosphere interactions (Hostetler et al, 1994; Bonan, 1995; Lofgren, 1997; Krinner, 2003; Long et al, 2007; Samuelsson et al, 2010; Dutra et al, 2010; Subin et al, 2012b; Deng et al, 2013; Xiao et al, 2016). We evaluated the improved lake model, WRFrLake (i.e., revised lake model), at the Nuozhadu Reservoir, a deep reservoir (up to 200 m deep) in southwestern China, and conducted sensitivity experiments to evaluate the effect of dominant processes and parameters on simulated lake water temperature and surface heat fluxes
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call