CSTPR has closed May 31, 2020: Therefore, this webpage will no longer be updated. Individual projects are or may still be ongoing however. Please contact CIRES should you have any questions.

Archived Projects:
Hydro-Climate Research and Decision Making

Hydrologic Model Inputs

Our research goal is to provide skillful local-scale forecasts of precipitation and temperature ranging from lead times of days to seasons.

Lead-time up to 2 weeks

We have developed statistical downscaling models (SDS) to derive 14-day forecasts of 6-hourly mean areal precipitation (MAP) and mean areal temperature (MAT) from the 20+ years archive of the NCEP-1998 (National Centers for Environmental Prediction) medium range forecast (MRF) model for the Upper Colorado River Basin (CBRFC). The MRF data archive is available from the NOAA Climate Diagnostic Center (CDC) ftp-server.

CDC is currently running the NCEP-1998 MRF model in real-time. Using outputs from this model, since January 2003 we have provided CBRFC with 14?day forecasts of 6-hourly MAP and MAT on a daily basis.

Lead-time, 2 weeks to seasons

To extend the forecasts leading out to seasons, we have developed a conditional weather generator. Weather generators resample historical weather sequences to generate events that preserve statistical moments. We have developed a method to condition this generation of weather sequences upon large-scale climate indices such as the ENSO or conditioning on probabilistic forecasts such as exceedance probabilities of climate variables in seasonal outlooks from the CPC (Climate Prediction Center).

Summary of Results

  1. In the Upper Colorado, for precipitation, the forecast skills are minimum even for a lead-time of 1?day. But for temperature we obtained significant skills even for forecast lead-times of 5-days.
  2. Skills in streamflow forecasts (April-July volumes) in the Upper Colorado exceed the traditional ESP by over 35 percent.
  3. Due to skills in temperature forecasts, we are obtaining improved results for snowmelt dominated basins, but for rainfall dominated basins there are no significant improvements.


  1. The SDS codes that we developed in the research was transferred to the CBRFC in December 2003, and is being tested as part of the AHPS (Advanced Hydrologic Prediction Services) as an experimental tool for ensemble streamflow forecasting.
  2. Publications. Only titles are listed below, to obtain the full papers click here.
    • Use of medium-range numerical weather prediction model output to produce forecasts of streamflow.
    • The Schaake Shuffle: A method for reconstructing space-time variability in forecasted precipitation and temperature fields.
    • A resampling procedure to generate conditioned daily weather sequences.

Ongoing Research

  1. Comparing different statistical downscaling techniques.
  2. Improving forecast skills for precipitation.