Lake Powell's water level is critical infrastructure for the American Southwest. I built an interactive forecasting tool that combines real-time USGS elevation data with NOAA snowpack measurements to project future reservoir levels.
The Data Pipeline
The tool pulls daily elevation readings from USGS Water Services and snowpack water-equivalent data from 10 SNOTEL monitoring stations. A GitHub Actions job runs daily to cache data and generate forecast snapshots for backtesting.
The Forecasting Model
The model uses a mass-balance approach linking snowpack water equivalent to spring net storage gain, achieving an R-squared of 0.899. Users can adjust snowpack scenarios and water release rates via interactive sliders to explore different trajectories. Hindcast validation shows plus or minus 4-foot accuracy for typical operating years.