| Lakes are characterized by low albedo,large heat capacity,reduced roughness,and unlimited water supply when compared to the surrounding land surfaces.These differences affect local surface-atmospheric exchanges of energy,moisture,and momentum fluxes.Consequently,the accurate representation of lakes in regional models is crucial to describe the lake-land-atmosphere interactions and improve the performances of numerical simulation and forecast.There are three main key scientific and technical issues to be solved for the lake modeling.First,in regional models,lake surface temperature has been typically prescribed with observations or predicted by interactive 1-D lake models.However,the former cannot represent the interactive lake-air processes,and the latter shows limited performances in reproducing the thermal patterns of the shallow lakes(like Lake Taihu)and is not suitable for deep lakes(like Great Lakes)due to the absence of horizontal processes.Second,the 3-D hydrodynamic models have been shown with obvious advantages in reproducing the thermal patterns of the deep lakes,but are difficult to be coupled with climate models mainly due to the technical difficulties involving different computational grids.Third,the previous regional coupled systems were unable to predict the water level fluctuations because it did not account for the lateral land surface runoff and connecting channel flows.This research develops the regional CWRF-Lake coupled system to effectively address the above issues.The main contributions of this study are demonstrated as follows:1)The 1-D upper ocean turbulence model UOM(built in CWRF)was customized for shallow-lake application.Its performance was evaluated in Lake Taihu with comprehensive measurements against three popular 1-D lake models.These models were based on different concepts,including the self-similarity(FLake),the wind-driven eddy diffusion(LISSS),the-turbulence closure(SIMSTRAT),and a simplified turbulence closure(UOM).All models in their default formulations presented obvious cold water temperature biases and largely underestimated the lake surface temperature(LST)diurnal range.The performances of all models were enhanced through the improvements on surface roughness,radiation extinction,and vertical mixing.The primary modifications for UOM included(1)a new scheme of decreased surface roughness lengths to better characterize the shallow lake,(2)a solar radiation penetration scheme with increased light extinction coefficient and surface absorption fraction to account for the high water turbidity,and(3)turbulent Prandtl number increased by a factor of 20 to reduce the turbulent vertical mixing.Given these improvements,UOM showed superior performance to other models in capturing LST diurnal cycle and daily to seasonal variations,as well as seasonal vertical stratification changes.2)The 3-D hydrodynamic model FVCOM was coupled with the regional climate model CWRF via the use of CPL7 coupler for deep-lake application.The coupled system was first elaborated in its main design and implementation,and then applied over the Great Lakes Region.Its performance was evaluated in representing lake-climate conditions against observations during 1999-2015 in comparison with the CWRF baseline using the 1-D LISSS model.Note that the LISSS model has been calibrated over the Great Lakes by previous study,and UOM was not implemented for the lack of snow and ice processes.It was shown that LISSS tended to predict earlier spring stratification,leading to the significant warm biases,and failed to simulate the ice cover.On the other hand,as coupled with CWRF,FVCOM outperformed LISSS in simulating water surface temperature,ice cover,and vertical thermal structure at seasonal to interannual scales for all the five lakes by fully resolving the lake hydrodynamics and also realistically reproduced the circulation patterns.3)The improved conditions of Great Lakes had substantial impacts on regional climate.In warm seasons,the removal of the warm biases significantly reduced the turbulent heat fluxes(especially for the latent heat flux),leading to the correction of the overestimates of surface air temperature together with notable geostrophic responses of low-level wind circulation.Meanwhile,the general reduction of precipitation over the lake basin was mainly related to the decreased moisture and heat fluxes as well as the enhanced atmospheric stability over the lakes.On the other hand.The extent of the lake water level variation was mainly controlled by the net basin supply(NBS = Precipitation-Evaporation + Runoff)and connecting channel flow.Given the key advances in representing NBS components(especially for runoff)and connecting channel flows,CWRF-FVCOM succeeded in the dynamic prediction of long-term water level fluctuation.It is shown that CWRF-FVCOM reasonably captured the NBS variations,leading to the realistic representation of the long-term water level fluctuations.In summary,the regional CWRF-Lake coupled system developed in this study consists of the interactive 1-D UOM model(tightly coupled at the same grid)and 3-D FVCOM model(through flux coupler across varying grids)to represent the shallow and deep lakes,respectively.Consequently,the system not only shows improved performances on lake modeling,but also is able to dynamically predict the water level of lakes.However,the snow and ice processes,which were not presently accounted,must be incorporated into UOM to facilitate its utility for high-latitude lakes.Moreover,the coupled system can also be applied for coastal oceans,as both UOM and FVCOM have the built-in ability to predict salinity variations.Ongoing efforts are to refine China CWRF grid resolution and improve the physics representation(especially for hydrology)as well as build unstructured meshes for the coastal oceans.Ultimately,the subseasonal-to-seasonal regional climate prediction infrastructure over China is expected to be developed and can be applied to predicting water levels in the coastal oceans as well as their interactions with regional climate and watersheds. |
PDF Full Text Request