Updates on the Assessment of Environmental Variables Affecting Chlorophyll-α through Multi-media Modeling and Machine Learning: a Lake Erie Case Study

Last year we introduced the concept of multi-media modeling and machine learning to assess environmental variables that affect water quality by using Lake Erie as a case study. Since then, we have improved our model by increasing the study period from 11-years (2002-2012) to 16-years (2002-2017) as well as utilized outputs from updated versions of numerical prediction models. Lake Erie continues to experience seasonal harmful algal blooms (HABs) and hypoxia due to a combination of anthropogenic activities that have contributed to increased loading of nutrient pollution. In-situ chlorophyll-α (chlor-α) observations are provided by the Lake Erie Committee Forage Task Group (LEC FTG) for the 2002-2017 period. Modeling systems involved are the: 1) Weather Research and Forecasting Model (WRF); 2) Variable Infiltration Capacity Model (VIC); 3) Community Multiscale Air Quality Model (CMAQ); and 4) Environmental Policy Integrated Climate Model (EPIC). Select meteorological weather variables from WRF, hydrological variables from VIC, nitrogen deposition from CMAQ, and agricultural management practice variables from EPIC for the 16-year period are used to fit a random forest machine learning model to predict chlor-α concentrations. We discuss the importance of explanatory variables that originate from the multi-media modeling systems, and analyze the contribution of each covariate in the model to better understand the occurrence of high chlor-α concentrations. Lessons learned from developing and testing this machine learning approach can be used to tackle water quality problems in other lakes or coastal areas and inform policy decisions.