David M Wright, Ph.D.

Postdoctoral Research Fellow

University of Michigan

College of Engineering Department of Climate and Space Sciences and Engineering

Research

Dyn-Ctr_SWE
Difference in Snow Water Equivalent (SWE) accumulation between hourly updating FVCOM and static RTG-SST lake conditions in a December 2017 case study.

UFS, FV3GFS, FV3LAM, and FVCOM Coupling

NOAA's latest atmospheric dynamical core (FV3) is the basis for weather modeling in the Unified Forecasting System (UFS). This dynamical core has been coupled to both the GFS physics suite for global, medium-range weather forecasting (FV3GFS) and adapted to perform regional, high-resolution, short-term weather forecasts (FV3LAM). As an early adopter of these two atmospheric models, my research focuses on running these models at high spatial resolutions (3km) over the Great Lakes while coupling them with a lake hydrodynamic model (FVCOM) to represent lake characteristics. This coupling is expected to better resolve atmospheric and lake interactions over the region and improve weather forecasts. One key component that is being explored is the potential improvements that can be made to a forecast by allowing lake conditions to vary over time as compared to having them remain static across the forecast horizon which is how current operational weather models treat the lakes.

Using Machine Learning to Predict Lake Fluxes

The Great Lakes Evaportaion Network (GLEN) uses eddy covariance flux towers strategically located around the Great Lakes to record fluxes off the lake surfaces for use in water budget management and monitoring. These records have been used to evalutate the performance of atmospheric models in capturing the magnitude of these fluxes, in which significant limitations have been shown. To attempt to improve the representation of these fluxes in modeled atmospheric surface layer schemes, I am testing the use of the Random Forest Machine Learning algorithm with the GLEN observations with the ultimate goal of coupling this algorithm into future simulations over the Great Lakes.

PVLAKE
3D visualization of WRF simulation for a May 2003 case study

WRF Modeling

My research involves using high-resolution (< 5km horizontal resolution) Weather Research and Forecasting Model (WRF) simulations over the Great Lakes region in order to better understand precipitation processes which are unique to this region and few other places on Earth. These simulations look at the role the lake surface temperature plays on atmospheric dynamics during both summer and winter precipitation. The overall goal is to try to connect changes in seasonal temperature with the generation of precipitation, which is a step towards being able to make connnections between climate and weather.

Great Lakes Atmospheric Dynamics

The Great Lakes offer unique challenges in terms of understanding the role of the lake surface in modifying or creating atmospheric circulations. These challenges are due to lack of fundamental understanding in the energy exchange between lake and atmosphere due to limited observations. These gaps in the communities knowledge can be addressed through a combination of high-resolution, coupled simulations and observational field campaigns to make measurements over the lake surface.

Back to top