Combining high-resolution numerical weather predictions ... fileEidgenössisches Departement des Innern EDI Bundesamt für Meteorologie und Klimatologie MeteoSchweiz Combining high-resolution

Eidgenössisches Departement des Innern EDI

Bundesamt für Meteorologie und Klimatologie MeteoSchweiz

Combining high-resolution

numerical weather predictions

with human expertise for

localized weather forecasts

Daniel Cattani, Lionel Moret and Dominique Stussi

Federal Office of Meteorology and Climatology MeteoSwiss

2 EMS 4–8 September 2017 | Dublin, Ireland MeteoSwiss

To be the best one must have

- Reactivity

- Consistency

- High resolution

- Quality

- Many wx parameters

- Top scientific level

3 EMS 4–8 September 2017 | Dublin, Ireland MeteoSwiss

Data4WEB’s goal and underlying

principle

Goal: feed MeteoSwiss website with high resolution temporal

and spatial data.

Underlying Principle : man-machine mix to get the best of

the two worlds:

• Model’s high resolution

• Forecaster’s reactivity and local expertise.

4 EMS 4–8 September 2017 | Dublin, Ireland MeteoSwiss

Modified or DMO

gridded forecast

Gridded analyses

6 EMS 4–8 September 2017 | Dublin, Ireland MeteoSwiss

Reactivity

• Rapid update cycle 30 min. Most recent model runs but also last

forecaster’s modification.

• Automatic measurement integration through INCA (nowcasting

system).

7 EMS 4–8 September 2017 | Dublin, Ireland MeteoSwiss

Consistency

Production of coherent forecasts

• Forecasts supervised by a single person

• Single point forecast source

• Consistency with present weather and active warnings

• Coherence with mobilephone app animations : based one the

same source

8 EMS 4–8 September 2017 | Dublin, Ireland MeteoSwiss

High resolution

Spatial and temporal pattern from high resolution models

reproduced

Downscaling of forecaster supervision with IFS (9 km), COSMO-E

(2km) and COMSO-1 (1km).

Adjusting high-resolution NWP ensemble prediction (COSMO-E) with

regionally and temporally aggregated estimate by forecaster (prototype

Adjusting high-resolution NWP (COSMO-E median) with regionally aggregated estimate by

forecaster

9 EMS 4–8 September 2017 | Dublin, Ireland MeteoSwiss

Quality

Full verification (models, forecasters, final products)

- Allows process improvement

10 EMS 4–8 September 2017 | Dublin, Ireland MeteoSwiss

Many parameters

Mosted desired parameter is the pictogram. But a high demand

exists for addional parameters.

data4WEB ensure consistency between parameters;

• Pictograms, temperature and dispersion, precipitation and dispersion

• Wind, precipitation probability, wind and temperature in free atmosphere,

sunshine, confidence index, stratus’ top, and soon road surface forecast

COSMO-1 COSMO-E (median) IFS-ENS (median)

11 EMS 4–8 September 2017 | Dublin, Ireland MeteoSwiss

High scientific level

data4WEB system built on high level science

• Measurement integration through INCA (2D-analyse and

shortterm forecasts), CombiPrecip (2D-analyse combining

raingauge and radar measurement)

• High resolution models; COSMO-NExT, ECMWF Hres

• Ensemble forecast COSMO-e, and ECMWF ENS

12 EMS 4–8 September 2017 | Dublin, Ireland MeteoSwiss

13 EMS 4–8 September 2017 | Dublin, Ireland MeteoSwiss

Perspectives

• Build a complete framework for the verification

• Improve the TS localisation

• New probabilistic approach for precipitations, and winds

14 EMS 4–8 September 2017 | Dublin, Ireland MeteoSwiss

• That’s all

15 EMS 4–8 September 2017 | Dublin, Ireland MeteoSwiss

- Champs de modèles utilisés

- Puis pour chaque point d’intérêt, interpolation spatiale

Principe ‘data4web’ Précipitations

16 EMS 4–8 September 2017 | Dublin, Ireland MeteoSwiss

‘data4web’ in progress

17 EMS 4–8 September 2017 | Dublin, Ireland MeteoSwiss