Project Cryosphere: Computational Forecasting of Snowpack-Driven Water Availability

Water scarcity is an increasingly important issue in New Mexico, where communities and ecosystems depend on the Rio Grande. Because river flow is influenced by snowpack, temperature, and precipitation, understanding these relationships could help improve predictions of future water availability. My project will investigate how accurately historical environmental data can predict spring streamflow in the Rio Grande near Albuquerque.

I hypothesize that winter snowpack and temperature can be used to predict spring streamflow, but that combining these variables will produce more accurate predictions than using snowpack alone. To test this hypothesis, I will collect historical streamflow measurements from the United States Geological Survey (USGS) and snowpack data from the Natural Resources Conservation Service (NRCS). I will also obtain historical temperature data to examine how warmer conditions may affect the relationship between snowpack and river flow.

Using Python, I will organize the datasets and analyze relationships between environmental variables and average spring streamflow. I will first develop a linear regression model, then compare its performance with a more advanced machine-learning model, such as random forest regression. I will evaluate both models using historical years excluded from training and compare their prediction errors.

The goal is to determine which environmental variables best predict spring streamflow and whether machine learning offers a meaningful improvement over simpler statistical methods. This project could help illustrate the potential and limitations of computational modeling for understanding water availability in New Mexico. It will also provide insight into how environmental conditions relate to regional water resources.