A comparison of short-range forecasts from two Ensemble Streamflow Prediction Systems G. Thirel (1), F. Rousset-Regimbeau (2), E. Martin (1), J. Noilhan (1) and F. Habets (3) (1) CNRM-GAME, Météo-France, CNRS, GMME/MC2, France, (2) Direction de la climatologie, Météo-France, France, (3) UMR SISYPHE, UPMC, ENSMP, CNRS, Paris, France ([email protected], +33 (0) 5 61 07 97 30)
13
Embed
A comparison of short-range forecasts from two Ensemble Streamflow Prediction Systems
G. Thirel (1), F. Rousset-Regimbeau (2), E. Martin (1), J. Noilhan (1) and F. Habets (3) (1) CNRM-GAME, Météo-France, CNRS, GMME/MC2, France, (2) Direction de la climatologie, Météo-France, France, (3) UMR SISYPHE, UPMC, ENSMP, CNRS, Paris, France - PowerPoint PPT Presentation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
A comparison of short-range forecasts from two Ensemble
Streamflow Prediction Systems
G. Thirel (1), F. Rousset-Regimbeau (2), E. Martin (1), J. Noilhan (1) and F. Habets (3)
(1) CNRM-GAME, Météo-France, CNRS, GMME/MC2, France,
(2) Direction de la climatologie, Météo-France, France,
Every day since 2004 : an Ensemble Streamflow Prediction System (ESPS) based on SIM (hydro-meteorological model) (Rousset, 2007). over all of France forced by ECMWF Ensemble Prediction System (EPS) forecasts medium-range (10 days), validated (more details on a poster, on Thursday, Hall A,
13:30 –15:00)
⇒ Increasing needs for short-range forecasting (Mediterranean region) A short-range ESPS based on the Météo-France EPS (PEARP)
Short-range (2 days)
OBJECTIVES : To compare the impacts of two 2-day EPS forecasts on this ESPS.
• Correct flood intensity and temporality prediction (rise, date of the flood peak, fall)• Correct spreadFlood prediction as early as 11-12 March : pre-alert, alert
Comparison on the first two days of simulation 569 days of simulation (10 March 2005 – 30 September 2006) 881 gauge stations compared
Project outline :
The two EPS forecasts used as input Precipitation disaggregation Streamflow prediction scores Conclusions, perspectives
Project outline
The two EPS forecasts used as input
ECMWF• 51 members• Homogeneous resolution• 10 days (+5)• Singular vector,
– leading time 2 days• Resolution in operational database 1.5°
PEARP• 11 members• Zoomed version • 60 H forecast• Singular vectors
– leading time 12 hours– over Europe
• Resolution in operational database : 0.25°
Precipitation disaggregation
ECMWF : altitudinal gradient (climatological)• 2 mm/m/year where altitude < 800m • 0.7 mm/m/year where altitude > 800m
PEARP : bias removal calibrated over one year
Observations (5000 rain gauges)
ECMWF (Day 1)
PEARP (Day 1)
Results over the test period : 10 March 2005 – 30 September 2006
Statistical scores are always better for PEARP than for ECMWF rainfall
BSS low flows (Q10)
Blue : ECMWF better with 90% of certainty according to the resampling testRed : PEARP better with 90% of certainty
Day 1 Day 2
ECMWF : 98 stations
PEARP : 184 stations
ECMWF : 33 stations
PEARP : 329 stations
BSS High flows (Q90)
Day 1 Day 2
Blue : ECMWF better with 90% of certainty according to the resampling testRed : PEARP better with 90% of certainty
ECMWF : 49 stations
PEARP : 338 stations
ECMWF : 19 stations
PEARP : 486 stations
Distribution by basin size (BSS)
Basins sizes
Q10 Day 1
Q10 Day 2
Q90 Day 2
Q90 Day 1
ECMWF
PEARPBasins sizesBasins sizes
Basins sizes
Conclusions
PEARP disaggregated rainfall was better than ECMWF.– The disaggregation method was different for the two EPS (must be
adapted to the meteorological model).
The PEARP ESPS showed improvements on floods and small scale basins at short-range scale– Results confirmed by various statistical scores (RPSS, reliability
diagrams, False Alarm and Hit Rates, and a seasonal study.)– Interest for flood forecasting in France (SCHAPI)
Details of the study in On the impacts of short-range meteorological forecasts for ensemble streamflow predictions, G. Thirel, F. Rousset-Regimbeau, E. Martin, F. Habets, Journal of Hydrometeorology, 2008, Accepted.
Perspectives
Improvement of rainfall disaggregation
Forecast « real » streamflows :
• To be compared with observations, not a reference run of the model
• Build a realistic initial state for the model : Assimilation of observed streamflows to retrieve a correct soil moisture (description of the method on a poster on Thursday, Hall A, 17:30 - 19:00)