1. Overview of WG4 Users Survey responses
Anastasia Bundel
RHM
COSMO GM in Saint Petersburg, 04 September 2018

2.
It was decided at the WG4 meeting in Israel, 2017, to carry out the WG4 users survey to better understand perspectives as a group and the user needs

3. Contributing persons Pierre Eckert (MCH) Daniel Cattani (MCH)
Andrzej Mazur (IMGW-PIB)
Dimitra Boucouvala (HNMS)
Anastasia Bundel (RHM)
Comments to the questions from Daniel Rieger and RHM colleagues

4. Survey blocks
General (NWP used, lead times, NWP forms, most important variables)
Verification questions
NWP correction
Most important phenomena (including severe weather)
Probabilistic forecasts, EPS
Nowcasting questions
COSMO/ICON ART
Willingness to share postprocessing methods

5. NWP used
COSMO (1, 2, 4, 7 km) and ECMWF-hres and ENS (except for Russia, which is not an ECMWF member or co-operative state, although some commercial organizations in Russia buy ECMWF products).
The COSMO guidance is estimated as good by majority of answers!
ICON-LAM is not used operationally in any of the services yet, but transition to ICON is previewed.
WG4 will participate in Task 5.6 of the C2I PP, and the survey will be updated for the purpose of obtaining users’ feedback on ICON-LAM forecasts.

6. Critical lead time
Very dependent on the application. Can be stratified by application type
As most of the answering persons represented National Weather Services (NWS) and short-range prediction departments, the typical forecast range of 1-3 days was indicated.
Warnings for the population are most important (wind and precipitation in particular)!

7. NWP representation from
Mostly traditional forms: maps -> meteograms -> other plots

8. Specific products (aeronautical, sea-route, other…)
External products, e.g. MCH receive MOS from DWD, road forecasts (MCH, Poland). HNMS is responsible for sea (automatically produced) and aeronautical forecasts. In Russia, COSMO forecasts are provided for a special semi-commercial institution, which prepares aeronautical forecasts; for the sea models, etc.

9. IMGW-PIB
has long experience of providing the special users with meteorological data:
aviation, road services, and energy production and supply: thunderstorms, frost, wind gusts, strong winter storms, fog, wind shear, power lines- and road icing (probability of ”black ice”).
For renewable energy production: wind, insolation, precipitation for hydroelectric power plants
What else is needed: squall lines, cloud ceiling, road/constructions (bridges) temperatures (for aviation and road maintenance)
However: the number of contracts decreased with the introduction of “open-data law”

10. Verification
Historical verification is taken into account, but could be used wider.
Forecaster experience is essential!
Some forecasters underlined the importance of real-time forecast quality monitoring, that is taking into account the last forecast errors.
Mostly traditional observations are used

11. What other verification products could be useful
Many users would like to have more interactive and real-time verification products
Stratified verification (weather types, …)
Spatial verification using gridded (e.g., radar-gauge) data. Remote sensing data should enter the operational verification chain not only for precipitation, but also for cloudiness, brightness temperature, …

12. Forecast correction Yes, it is necessary: Forecast matrix (MCH)
Automatic techniques : MOS, Kalman filter (MCH, HNMS, Russia) for T2m
Based on forecaster experience (“human calibration”) for temperature, precipitation, cloudiness, snowfall limits, …

13. Most important severe weather forecasts
All of these are important, difficult to classify, but to name among all the types:
Heavy rain and snow, strong winds, thunderstorms, heat waves, fogs
Less traditional: freezing precipitation, 100 m mean wind (for wind energy), downward radiation
Mostly DMO is used (including EPS for some users), sometimes blended with recent observations (kind of nowcasting). IMGW-PIB method for fogs and thunderstorms!
Whence, an idea of a project on forecasting and verification of high impact weather (HIW) events to be discussed in the next joint session!

14. Probabilistic forecasts, EPSs
EPSs are used, but moderately (COSMO-EPS, ICON-EU, ECMWF ENS)
Added value of EPSs by majority of answers. E.g., good experience combining EPS and COSMO-1 in case of convective situations (MCH), good guidance for days 2-7 (HNMS, RHM)
Most useful EPS products: Ensemble median and spread, uncertainty, confidence index, spaghetti plots, clustering, probability maps for precipitation, extreme temperatures, precipitation, wind

15. However:
Not all the forecasters, especially outside NMSs, have enough experience to use EPSs
Thus, further expectations from EPSs:
- More friendly to non-experienced forecasters, easier to interpret
- Statistical adaptation of EPS output
One of the tasks of a PT on Guidelines for forecasters could be devoted to EPS products and how to use them

16. Nowcasting products Not all the members are using nowcasting
INCA combining observations and the model forecasts is used in MCH.
In RHM, an adaptation of STEPS statistical method is issued for very-short range warnings about heavy precipitation and thunderstorms until 2 h. A wish from forecasters providing meteosupport of sport events to extend the time period to 6 h.

17. So, almost unanimously:
“Blending of obs and model should give the possibility to extend the useful range to a few hours”
“Seamless forecast from actual measurements to model forecast”
“A nowcasting product would be helpful for short term predictions (0-9h). It should be available in almost real time to the forecaster, and provide information for decision making in case of the evolution of a phenomenon so mainly important in severe weather”
Another task of a PT on Guidelines for forecasters on using nowcasting products ?

18. COSMO/ICON-ART At present: Pollen in MCH
In RHM, concentrations of pollutants: CO, NO, NO2, O3, etc. in Moscow are sent to Mosecomonitoring (an organization controlling the air quality)
In future:
«Processes like fog formation could benefit from a prognostic (hygroscopic) aerosol forecast. Radiation as well» (Pierre E.)
«We are interested in COSMO-ART in the near future. The most important species are O3, SO2, NOx, aerosols” (Dimitra B.)

19. What type of postprocessing method are you ready to share with the other COSMO members?
All those who answered wrote that they could share all the available methods, possibly, after official approval of their administration

20. Many thanks to those who answered the questions as it already helped a lot in preparing the project plans and will serve in the future!

21. DMO vs. Official human forecasts (Sochi-2014 experience)
The skill of official and model forecasts as a function of the lead time. MAE of T2m aggregated over the mountain cluster, period from 1 November 2013 to 23 February Official forecasts issued at 1100 UTC, the models started at 1200 UTC
From Kiktev, D., P. Joe, G. Isaac, A. Montani, I. Frogner, P. Nurmi, B. Bica, J. Milbrandt, M. Tsyrulnikov, E. Astakhova, A. Bundel, S. Belair, M. Pyle, A. Muravyev, G. Rivin, I. Rozinkina, T. Paccagnella, Y. Wang, J. Reid, T. Nipen, and K. Ahn, 2017: AMERICAN METEOROLOGICAL SOCIETY FROST-2014: The Sochi Winter Olympics International Project. Bull. Amer. Meteor. Soc. doi: /BAMS-D

22. DMO vs. Official human forecasts (Sochi-2014 experience), cont.2
The skill of official and model forecasts as a function of the lead time. EDI of precipitation occurence aggregated over the mountain cluster, period from 1 November 2013 to 23 February 2014
From Kiktev, D., P. Joe, G. Isaac, A. Montani, I. Frogner, P. Nurmi, B. Bica, J. Milbrandt, M. Tsyrulnikov, E. Astakhova, A. Bundel, S. Belair, M. Pyle, A. Muravyev, G. Rivin, I. Rozinkina, T. Paccagnella, Y. Wang, J. Reid, T. Nipen, and K. Ahn, 2017: AMERICAN METEOROLOGICAL SOCIETY FROST-2014: The Sochi Winter Olympics International Project. Bull. Amer. Meteor. Soc. doi: /BAMS-D

23. DMO vs. Official human forecasts (Sochi-2014 experience), cont.3
Automated temperature forecasts, especially blended multi-model forecasts, were competitive to manual forecasts;
for wind speed and visibility, the human forecasts demonstrated the psychological biases towards higher speed and lower visibility (the phenomenon of overforecasting hazardous events by human forecasters is discussed, e.g., by Doswell (2004);
for precipitation, the manual forecasts did add value to model forecasts.
From Kiktev, D., P. Joe, G. Isaac, A. Montani, I. Frogner, P. Nurmi, B. Bica, J. Milbrandt, M. Tsyrulnikov, E. Astakhova, A. Bundel, S. Belair, M. Pyle, A. Muravyev, G. Rivin, I. Rozinkina, T. Paccagnella, Y. Wang, J. Reid, T. Nipen, and K. Ahn, 2017: AMERICAN METEOROLOGICAL SOCIETY FROST-2014: The Sochi Winter Olympics International Project. Bull. Amer. Meteor. Soc. doi: /BAMS-D