THE GREAT LAKES CLIMATE ANALYSIS USING A TWO-WAY COUPLED 3-D GREAT LAKES -ATMOSPHERE REGIONAL MODEL WITH DATA ASSIMILATION METHODOLOGY

The objective of this work is to provide the best estimation of physical state of the Great Lakes using the two-way coupled Great Lakes–Atmosphere Regional Model (GLARM) integrated with Data Assimilation (DA) methodology. The aim of the first part is to understand the lake internal process that determines the relationship between lake surface temperature (LST) and lake thermal variations. A 3-D hydrodynamic model was used to examine the nonlinear processes of water mixing and ice formation that caused changes in lake heat content and further variation of LST. The results show that heat content trends do not necessarily follow (and can even be opposed to) trends in LST. In addition, the lake total lake heat content, thermal properties, length of stratification periods, and lake stability intensity were analyzed using validated GLARM 3-D results from 1983-2016. Furthermore, the lake thermal variations were analyzed using physical stability indices. The results reveal that climate change would not only affect the air-lake energy exchange but can also alter lake internal dynamics. In the second part, a Great Lakes forecast system with both long-term and short-term predictions is presented. A downscaled, robust, and sophisticated two-way coupled GLARM model was used to project the climate change over the Great Lakes region over the periods of 2030 – 2050 and 2080 – 2100. Two Representative Concentration Pathway (RCP) emissions scenarios (RCP4.5, RCP8.5) were included. As a result, the stress in air temperature and precipitation during the period of 2080 – 2100 under the high emission scenario (RCP8.5) will be exacerbated with larger spatial variability compared with the medium emission scenario (RCP4.5). For lake conditions, annual mean LST of Lake Erie shows the largest changes among the five lakes. The decrease in the mean lake ice coverage is projected over all the five lakes, while the largest decrease occurs along the coast. Furthermore, the application of DA using Lake Erie as a case study was evaluated. The results show that DA can effectively improve the model performance with limited observational data. The data assimilative model also improves forecasting accuracy and restrains the forecasting uncertainty to an acceptable level on a timescale of 1-7 days after unleashed from DA