- [ONLINE] H21O-VR8935: Enhancing Interannual Water Supply Forecasts of Colorado River Basin Using National Multi-Model Ensembles and Machine Learning Techniques
-
OnlineOnline
Author(s):Generic 'disconnected' Message
Catalina Jerez, University of Colorado Boulder (First Author, Presenting Author)
Balaji Rajagopalan, Univ Colorado
Emerson LaJoie, Climate Prediction Center College Park
Matthew Rosencrans, NOAA
Sarah Baker, University of Colorado at Boulder
Seth Shanahan, Southern Nevada Water Authority
William Miller, NOAA
Edith Zagona, University of Colorado Boulder
Seasonal forecasts tell water managers when to fill or release the Colorado River’s big reservoirs. The official forecast, based on the Bureau of Reclamation’s in-house Ensemble Streamflow Prediction (ESP) traces from the Colorado Basin River Forecasting Center (CBRFC), becomes unreliable after just a few months. We blended those ESP traces with cutting-edge climate-model predictions and machine-learning tools. Testing 40 years of data, our Gradient-Boosting model stayed accurate out to two years, while the current approach lost skill by month six. This longer warning window can help operators prepare for very wet or very dry years, improving drought planning, power generation, and overall river management.
Scientific DisciplineSuggested ItinerariesNeighborhoodTypeWhere to Watch
Enter Note
Go to previous page in this tab
Session
