Status: Finalised Material from: Ivan. Linus

 

Discussed in the following Daily reports: http://intra.ecmwf.int/daily/d/dreport/2014/09/30/sc/


Picture

1. Impact

On the 29 September south-western France was hit but heavy rainfall causing flash-floods. Montpellier got around 300 mm and the city was flooded.

2. Description of the event

Warnings on Meteoalarm on the morning of 30 September.


The plot above shows the observed precipitation between 29 September 00UTC and 30 September 00UTC. The worst affected areas were along the coast in the south-west corner of the northern Mediterranean.

Precipitation and MSLP (12-hour forecasts)
1. 28 Sept 00UTC
2. 28 Sept 12UTC
3. 29 Sept 00UTC
4. 29 Sept 12UTC
5. 30 Sept 00UTC

The plots above show 12-hour precipitation and MSLP from 12 hour forecasts valid at the time labelled in the plots.

z500 and t850 analysis
1. 28 Sept 00UTC
2. 28 Sept 12UTC
3. 29 Sept 00UTC
4. 29 Sept 12UTC
5. 30 Sept 00UTC

The plots above show z500 and t850 analyses.

3. Predictability

  

3.1 Data assimilation

 

3.2 HRES


HRES 24-hour precipitation
1. 29 Sept 00UTC
2. 28 Sept 12UTC
3. 28 Sept 00UTC
4. 27 Sept 12UTC
5. 27 Sept 00UTC

The plots above shows the 24-hour precipitation. The hourglass symbol mark Montpellier that got 297 mm.


3.3 ENS

 

The plot above includes the CDF from the ensemble for Montpellier. The last forecast (29 Sept. 00UTC) had a 4-6% probability for more than 100 mm during the day.


Probabilities of >100 mm/24 hours

Prob >100 mm/24 ECMWF
1. 29 Sept 00UTC
2. 28 Sept 12UTC
3. 28 Sept 00UTC
4. 27 Sept 12UTC
5. 27 Sept 00UTC

The plots above shows the probabilities from ECMWF ENS from different initial times. The strongest signal for heavy rain is located north-west of Montpellier.

Prob >100 mm/24h COSMO-LEPS
1. 29 Sept 00UTC
2. 28 Sept 12UTC
3. 28 Sept 00UTC
4. 27 Sept 12UTC
5. 27 Sept 00UTC

The plots above shows the probabilities from COSMO-LEPS  (16 ensemble members, 10 km resolution) from different initial times. The probabilities are not improved by the higher resolution in this case.


EFI precipitation
1. 29 Sept 00UTC
2. 28 Sept 00UTC
3. 27 Sept 00UTC
4. 26 Sept 00UTC
5. 25 Sept 00UTC

The plots above show EFI for precipitation valid the 29 September for different lead times. Also the medium-range forecast had increased EFI for the area.

3.4 Monthly forecasts


3.5 Comparison with other centres


3.6 EFAS

The plots above shows warnings for flash floods from EFAS from the 28 September 00UTC (left) and 29 September (right). The flash flood forecast is based on COSMO-LEPS. None of the simulations forecast flash floods in southern France but in Spain.


4. Experience from general performance/other cases

 

5. Good and bad aspects of the forecasts for the event

  • Signal for heavy precipitation in the medium-range
  • Under-estimation of the amplitude
  • Shift in the maximum from the coast to the mountains in the forecast?
  • No signal in EFAS


6. Additional material