Contacts:


Cooperators:

Project Plan:

Start Date: October 1, 2002

Status: Completed - December 2003

End Date: December , 2003

Abstract:

The Colorado Basin River Forecast Center (CBRFC) currently uses historical climate information in its ensemble stream prediction (ESP) procedure. In the ESP approach, a hydrologic model is driven with observed precipitation and temperature data up to the beginning of the forecast date, and then run using precipitation and temperature data for the same date from every year in the historical record. This project enables the CBRFC to replace the ensemble of historical data with an ensemble of 0-14 day forecasts from a medium range forecast model (MRF) that is being run in real time at the NOAA Climate Diagnostics Center (CDC).

The first step in the project was to correct the model bias in the 1998 MRF product and downscale the MRF product to appropriate spatial scales for river forecasting. The CBRFC provided CIRES with twenty years of 6-hourly mean areal precipitation (MAP) and mean areal temperature (MAT) data for all basins in the study area. Using this archive and techniques of Model Output Statistics (MOS), empirical relationships were developed by CIRES between grid point values of global-scale numerical weather prediction model output and actual observations use in hydrologic models (mean areal data). Ensembles of MAPs and MATs in 6 hour time steps are produced for all basin segments in the study area. These ensembles are used as input to the ESP procedure. Probabilistic streamflow forecasts on instantaneous flow and flow volumes are prepared and issued on the WEB valid for the next three weeks.

The procedure was evaluated by the CBRFC by comparing forecasts from ESP Using historical climatology with the experimental forecasts resulting from the MRF Ensembles. A rank probability skill score (RPSS) was used for the evaulation.

The study area is located in the Upper Colorado basin, above the station at Cameo, Colorado. It consists of 26 forecast sub-basins. This area was chosen for several reasons: (1) a focus on an area where temperature Forecasts are important and associated with springtime peak runoff, (2) its an area where the forecast of the peak is needed in fish enhancement projects, (3) the sub-basins in this area have recently been recalibrated.

Comments:

The project was conducted in association and support of the Advanced Hydrologic Prediction Service (AHPS). It has been a cooperative effort between a NOAA Research Lab, Climate Diagnostic Center in Boulder, Coloraod, an operational NOAA field office, CBRFC and a University Collaborator, CIRES-Cooperative Institute for Research in Environmental Sciences, University of Colorado.

Project Area:


Goals:



Method:



Downscaling:

CIRES researchers were responsible for determining the downscale parameters. Details can be found in some of the related publications but the initial step was to use MLR. A complimentary study determined whether it was more accurate to downscale to station sites ( and then to compute mean areal values ) or to directly downscale to the mean areal value. It was determined that for the grid scale used it was better to downscale directly to the mean areal values.






Hydrologic Model Execution:

The ESP ( Ensemble Streamflow Prediction ) component of the NWSRFS was used as the hydrologic forecast system. Ensembles of precipitation and temperature were 'fed' as input and flow values on a six hourly time step out to 3 weeks were forecast.





Reforecasting/Verification:

The ESP model was run in historical/reforecast mode. Ensembles of precipitation and temperature derived from the MRF were developed. They were then used as input into ESP for each historical year. Forecast flows were calculate and anaylize. As a comparison ensembles calculated from the nominal case ( climatology ) were used as input to ESP. The RPSS ( Rank Probability Skill Score ) was used to examine the skill of including the MRF ensembles over climatology.




Additional Information:


Presentation by: Kevin Werner

Presentation by: Kevin Werner, WRCW - Reno, NV, September, 2003

Presentation by: Dave Brandon, CBRFC, April 23, 2003

Presentation by: Martyn P. Clark and Lauren E. Hay

Presentation by: Lauren E. Hay and Martyn P. Clark

Presentation by: Martyn P. Clark, Lauren E. Hay and Subhrendu Ganpopadhyay

Paper by: David Yates, Subhrendu Ganpopadhyay, Balaji Rajagopalan, and Kenneth Strzepek