Environmental Modeling Team webinars
Our webinars are designed to showcase the latest achievements from our consulting and research work.
Mapping Historical Soil Organic Carbon in the US with Machine Learning
November 20, 2024 (10:00 - 10:50 am EST) | Presented by Dr. Hannah Rubin
Industries trying to decarbonize by replacing fossil fuels with biofuels need to have an accurate accounting of how biofuel feedstock production impacts soil carbon. However, the lack of long-term spatially and temporally continuous estimates of soil organic carbon (SOC) hinders efforts to project how land management decisions will affect future carbon stocks. My work builds on prior approaches to make new maps of SOC stocks in the contiguous U.S. (CONUS) by comparing seven machine learning algorithms and incorporating more sources of soil measurements. I compile a georeferenced dataset of SOC measurements and match observations with climate, topography, soil moisture, and Normalized Difference Vegetation Index (NDVI), a measure of greenness. I then use this dataset to generate an estimate of 60.4 Pg C in the top 30 cm of soil for the contiguous US (CONUS). My research also focuses on evaluating regional land use and climate impacts to emphasize the scale dependence of trends and the historical agricultural practices that influence SOC.
Contact Dr. Hannah Rubin to learn more>>
Machine Learning for Hydrology and Water Quality Assessments: Using Random Forest for Spatial Extrapolation of Streamflow Metrics
July 17, 2024 (10:00 - 10:50 am EST) | Presented by Dr. Jens Kiesel
In this webinar, we introduce the application of the Random Forest machine learning algorithm for hydrological predictions. We will give a quick introduction to machine learning techniques and delve into the pros and cons of the Random Forest algorithm. The core component will cover input data processing, application of the algorithm, a performance assessment of the results, and identifying the importance of the predictor variables targeted to streamflow simulations. We will share insights into lessons learned and pitfalls in using Random Forest and provide a simple code example to get you started implementing your own Random Forest machine learning model.