You are here:
Using a Freshwater Lake Model Coupled with WRF for Dynamical Downscaling Applications
Citation:
Mallard, M., Chris Nolte, R. Bullock, T. Otte, AND J. Gula. Using a Freshwater Lake Model Coupled with WRF for Dynamical Downscaling Applications. Presented at AMS Annual Meeting, Atlanta, GA, February 04 - 05, 2014.
Impact/Purpose:
The National Exposure Research Laboratory (NERL) Atmospheric Modeling and Analysis Division (AMAD) conducts research in support of EPA mission to protect human health and the environment. AMAD research program is engaged in developing and evaluating predictive atmospheric models on all spatial and temporal scales for forecasting the air quality and for assessing changes in air quality and air pollutant exposures, as affected by changes in ecosystem management and regulatory decisions. AMAD is responsible for providing a sound scientific and technical basis for regulatory policies based on air quality models to improve ambient air quality. The models developed by AMAD are being used by EPA, NOAA, and the air pollution community in understanding and forecasting not only the magnitude of the air pollution problem, but also in developing emission control policies and regulations for air quality improvements.
Description:
The ability to represent extremes in temperature and precipitation in regional climates (including those affected by inland lakes) has become an area of focus as regional climate models (RCMs) simulate smaller temporal and spatial scales. When using the Weather Research and Forecasting (WRF) model to downscale future global climate model (GCM) projections, model users typically must rely on the GCM to represent temperatures at all water points. However, GCMs typically have insufficient resolution to adequately represent even large inland lakes, such as the Great Lakes. In some cases, a single GCM point is tasked with representing the lake surface temperature (LST) and ice concentration over multiple large, heterogeneous lakes. This treatment can result in lakes as large as Lake Superior freezing completely in the space of a single timestep. When no water points are close enough to interpolate from, the representation of lakes can be further complicated by the setting of the LSTs from the nearest water point, even if the only available water temperatures are from ocean points hundreds of km away. The current study examines three different ways in which LSTs and lake ice can be set in the WRF model, where it is applied as an RCM to produce 12-km simulations over the eastern U.S. In order to assess the model’s performance, the 1.875⁰ NCEP–DOE Atmospheric Model Intercomparison Project Reanalysis-2 (R2) data is used as a proxy for a typically-coarse GCM, and the downscaled WRF output is compared with other observational or analyzed resources. In the control run referred to as “CTLR2”, LSTs and ice are set from the R2 dataset, where the Great Lakes are collectively represented by only three points. A second control run, CTLOb, is driven with high-resolution observations of ice from the National Ice Center and LSTs from the Advanced Very High Resolution Radiometer (AVHRR) dataset. CTLOb is a benchmark “best case scenario” run that demonstrates WRF’s performance when analyzed products that are of an appropriate scale for use within a 12-km simulation are utilized. However, it does not provide guidance for dynamical downscaling, since those observational resources will not be available for future GCM projections. Finally, a version of WRF which is dynamically coupled to the Freshwater Lake (FLake) model is compared with the previously-described control runs. FLake is a 1D column model, consisting of a two-layer parametric representation of a time-varying temperature profile that includes a mixed layer and a thermocline extending down to a layer of thermally-active sediment. WRF-FLake’s simulated LSTs and ice concentrations are compared to the NIC and AVHRR observations. Analysis of all three runs will focus on 2-m temperatures and rainfall, assessing what impact the choice of lake representation has on WRF’s performance in an RCM setup over a two-year period.