1 / 27

Effect of Model Calibration on Streamflow Forecast Results

Effect of Model Calibration on Streamflow Forecast Results. Ali Akanda, Andrew Wood, and Dennis Lettenmaier Civil and Environmental Engineering University of Washington Seattle, WA. Problem

Download Presentation

Effect of Model Calibration on Streamflow Forecast Results

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Effect of Model Calibration on Streamflow Forecast Results Ali Akanda, Andrew Wood, and Dennis LettenmaierCivil and Environmental EngineeringUniversity of WashingtonSeattle, WA

  2. Problem Calibration is always a time-consuming and labor intensive part of modeling process. Automatic calibration routines are available but still not used widely. Has hampered implementation of models in different operational settings. • 1970s: ESP method developed in NWS • 2000s: ESP implemented in NWRFC s Question if operational water supply forecasting is mostly concerned with seasonal volumes, do we need to calibrate? Objective see whether bias-correction can achieve same goals as calibration in forecasting streamflow values

  3. Contents • Domain • Calibration • Forecasts • Results • Summary

  4. Domain • Western U.S. • Small-Medium • 1/8th degree • 9 being used • Mostly unimpaired

  5. Calibration Manual • Visual comparison of averaged streamflow hydrographs Base State: NLDAS Parameters • Ds • Ds max • Ws • binf • Soil Depth [each layer]

  6. Green - ColoradoDrainage: 468 sq. miles

  7. North Fork FlatheadColumbia Falls, MTDrainage: 1548 sq. miles

  8. White - ColoradoDrainage: 755 sq. miles

  9. Weber – Great BasinDrainage: 162 sq. miles

  10. West Walker – Great BasinDrainage: 181 sq. miles

  11. Verde – ColoradoDrainage: 5858 sq. miles

  12. Forecasts • ESP (Ensemble Streamflow Prediction) • 30 Ensembles for each run (1970-1999) • Forecasts for 25 different years: 1975-99 • Yearly forecasts run from January 1 / April 1 • Dry season streamflow average values (April-July and April-September)

  13. Streamflow forecasts – Weber River Basin, UTApril 1 forecasts All error values are in cfs

  14. Streamflow forecasts – White River Basin, COApril 1 forecasts All error values are in cfs

  15. Streamflow forecasts - North Fork Flathead (NOFOR @ PNW)

  16. Streamflow forecasts - North Fork Flathead (NOFOR @ PNW)Error Values Jan 1 Forecasts Apr 1 Forecasts

  17. Results • Bias Correction performed based on respective 25-year climatology (75-99) • Percent Anomaly • Rank Percentiles • Streamflow Error Values (averaged over ensembles / years) • MAE (Mean Average Error) • RMSE (Root Mean Squared Error)

  18. Bias Corrected Streamflow forecasts Weber River Basin, UTApril 1 forecasts All error values are in cfs

  19. Bias Corrected Streamflow forecasts Weber River Basin, UTApril 1 forecasts All error values are in cfs

  20. Bias Corrected Streamflow forecasts White River Basin, COApril 1 forecasts All error values are in cfs

  21. Bias Corrected Streamflow forecasts White River Basin, COApril 1 forecasts All error values are in cfs

  22. Bias Corrected Streamflow forecasts N Flathead River , MTApril 1 forecasts All error values are in cfs

  23. Bias Corrected Streamflow forecasts N Flathead River , MTApr 1 forecasts All error values are in cfs

  24. Bias Corrected Streamflow forecasts White River Basin, COJan 1 forecasts All error values are in cfs

  25. Bias Corrected Streamflow forecasts White River Basin, COJan 1 forecasts All error values are in cfs

  26. Summary • Calibration helps to reduce the error of streamflow forecast results (expected) • Difference of Uncalibrated vs Calibrated forecast results greatly reduced if bias is removed by either method • Percentile-based bias correction performs better than anomaly-based bias correction • Error reduction from bias-correction similar to that achieved by calibration • Similar trends observed with both January 1 and April 1 forecasts

  27. Work to be done • Comparison of forecast results with different initiation dates (Jan/ Apr 1) • Similar results for calibrated basins • Study even larger basins (Salmon?)

More Related