Status: Ongoing analysis Material from: Linus, Fatima, Tim H., Esti, Andrea
Some extraordinary flash floods hit several parts of Sardinia on 28 November, which for a widespread event was probably only eclipsed in the last decade, on that island, by the events of 18 November 2013. In the 2013 event the peak event rainfall was about 450mm, in the recent event it was closer to 350mm. They were in fact remarkably similar, in synoptic terms, with an 'upside down' cold front moving south to north across the island, over about a day, on the eastern flank of a deep Mediterranean cyclone, and with warm moist air on the NE flank of the cold front, feeding up from a marine source that starts near the N coast of Libya. This airflow delivered periods of heavy rain to the eastern side of the island in particular, where sharply rising mountains can also lead to significant orographic enhancement. Widespread lightning strikes over land (and sea), and IFS model output (profiles), together attest to the fact that the 2020 event was highly convective over Sardinia, notably in the first half of 28th.
2. Description of the event
Animation of synoptic charts.
Rainfall observation breakdown:
12-hourly totals (third plot, with smaller spots, shows data from the second plot augmented by data from another Sardinian network, for 23-11UTC):
The plots below show analyses of MSLP and 6-hour forecasts of precipitation from 27 November 00UTC to 29 November 00UTC, every 12th hour.
3.1 Data assimilation
The plots below show 24-hour precipitation on 28 November in observations (first plot) and short-range HRES forecasts.
The plots below show EFI for 24-hour precipitation valid 28 November in forecasts with different initial times.
The plot below shows the evolution of the forecasts for 24-hour precipitation on 28 November in the box outlined in the plots above. The plot includes HRES(red), ENS control (purple), ENS distribution (blue), model climate distribution (red) and mean of observations (green).
3.4 Monthly forecasts
3.5 Comparison with other centres
4. Experience from general performance/other cases
5. Good and bad aspects of the forecasts for the event
6. Additional material
This case was also relevant for the MISTRAL initiative (an EU project in which Italian partners and ECMWF have collaborated). In that project we have the "Italy flash flood use case" (see here and here), spearheaded by ECMWF, in which post-processing, of different types, is applied to the ECMWF ENSemble, and also a 2.2km Italy-centred limited area COSMO ensemble. The post-processed outputs are blended together, with lead-time-dependant weighting, to make the final product, which aims to give better probabilistic rainfall forecasts, in particular with a view to providing improved early warnings of flash flood risk (via the association with extreme short period rainfall). Users are encouraged to focus on the higher percentiles (or probabilities of exceeding high thresholds) in the MISTRAL products, to gauge the potential for localised extremes. However some of the plots below illustrate also how the ensemble mean is handled in the raw model and post-processed output, with differences between the two (for a given system) indicating the nature of any bias-correction being applied on the model grid-scale. This plot focusses on a 6h accumulation period, 06-12UTC 28th, when most of the rainfall was diagnosed in the ECMWF model runs to be convective. In contrast Sections 3.2 and 3.3 above deal with a 24h accumulation period.
|Model||0. Observations (all the same)||1. Raw ECMWF ENSemble||2. PP-ENS (ecPoint)||3. Raw COSMO Ens||4. PP-COSMO Ens||5. Difference: 1 minus 3||6. 98th %ile ecPoint||7. 98th %ile PP-COSMO||8. 98th %ile blended|
|DT 12UTC 26th|
|DT 00UTC 27th|
|DT 12UTC 27th|
|DT 00UTC 28th|
Whole table corresponds to 6h rainfall totals for 06-12UTC 28th November 2020
Column 0: Verifying observations
Columns 1-5: Ensemble Mean fields for ECMWF IFS (runs from 00UTC and 12UTC) and COSMO 2.2km (runs from 21UTC only, nominally shown as 00UTC here). PP means post-processed (different approaches for the two systems)
Columns 6, 7: 98th percentiles from ecPoint and post-processed COSMO
Column 8: MISTRAL-style output: 98th Percentile, a blended version of post-processed ENS (12UTC run) and post-processed COSMO (21UTC run same day)
Some points to note from the above:
- Some very localised extremes occurred in this time window (related to flash flood events) - e.g. see pink spots in column 0.
- RAW ECMWF ensemble mean forecasts, at short lead times (as used for the MISTRAL products), were very consistent.
- RAW COSMO ensemble mean forecasts (admittedly only two sets available) were not very consistent. The shorter lead time forecasts showed much more rain in the south of Sardinia, and a bit less in the north.
- Near the east coast the RAW ECMWF ensemble (mean) over-predicted rainfall, whilst COSMO under-predicted; respectively about 30mm and 2mm versus 10-20mm consistently observed.
- ecPoint post-processing of the RAW ECMWF ensemble removed most of the positive bias (based on calibration which tells the post-processing that over-prediction in such situations is commonplace). COSMO post-processing did not remove the negative bias (because bias correction is not built in).
- ecPoint post-processing can in principle shift the spurious coastal maximum inland, to where it is needed, if different bias levels are incorporated via calibration: there are some hints of a shift in this direction. With predictor variable adjustments more could probably be done.
- COSMO mean fields seem to perform better in the NW and SW of Sardinia, where the RAW ECMWF ensemble seems to have insufficient rain.
- 98th percentiles of the post-processed fields show much larger values than the ensemble mean, which is consistent with the observations that show a lot of local variability, with values broadly of the right order (about 2 of the available observations should exceed the values shown).
- Blending can help mitigate against the systematic biases of the two systems particularly when they are "in opposition", as here, although the impact of giving increasing weight to COSMO at shorter leads is also clear to see (column 8).