Federal Department of Home Affairs FDHA Federal Office

  • Slides: 16
Download presentation
Federal Department of Home Affairs FDHA Federal Office of Meteorology and Climatology Meteo. Swiss

Federal Department of Home Affairs FDHA Federal Office of Meteorology and Climatology Meteo. Swiss Postprocessing of temperature and wind for COSMO-7 and COSMO-2 COSMO General Meeting Vanessa Stauch Offenbach, September 2009

calibration with Kalman Filter >> recursive estimation of forecast error (prediction – correction) >>

calibration with Kalman Filter >> recursive estimation of forecast error (prediction – correction) >> requires online observations >> can be used quasi-instantaneously (no large historical database) >> cannot predict fast changes (assumption of persistent error) >> suitable for a subset of parameters (normally distributed errors) Statistical PP | COSMO-GM 2009 Vanessa Stauch 2

calibration with Kalman Filter error : error model: ^ with states evolution: prediction: Statistical

calibration with Kalman Filter error : error model: ^ with states evolution: prediction: Statistical PP | COSMO-GM 2009 Vanessa Stauch with ^ 3

COSMO models COSMO-LEPS 10 km, +132 hours COSMO-2 COSMO-7 6. 6 km, +72 hours

COSMO models COSMO-LEPS 10 km, +132 hours COSMO-2 COSMO-7 6. 6 km, +72 hours COSMO-2 COSMO-7 Statistical PP | COSMO-GM 2009 Vanessa Stauch 2. 2 km, +24 hours 4

Kalman Filter @ Meteo. Swiss operational: in preparation: T 2 m, TD 2 m

Kalman Filter @ Meteo. Swiss operational: in preparation: T 2 m, TD 2 m FF 10 m, TW 2 m, RH 2 m for COSMO-LEPS mean COSMO-7 COSMO-2 IFS Statistical PP | COSMO-GM 2009 Vanessa Stauch 5

Swiss met. measurement network 62 stations Statistical PP | COSMO-GM 2009 Vanessa Stauch 6

Swiss met. measurement network 62 stations Statistical PP | COSMO-GM 2009 Vanessa Stauch 6

T 2 m predictions COSMO-7 COSMO-2 KFC 7 KFC 2 performance? Statistical PP |

T 2 m predictions COSMO-7 COSMO-2 KFC 7 KFC 2 performance? Statistical PP | COSMO-GM 2009 Vanessa Stauch 7

benefit COSMO-2 vs COSMO-7? 04. 04. 08 – 31. 10. 08 All ANETZ stations

benefit COSMO-2 vs COSMO-7? 04. 04. 08 – 31. 10. 08 All ANETZ stations Low ANETZ stations High ANETZ stations Statistical PP | COSMO-GM 2009 Vanessa Stauch C 2 vs C 7 C 2 -KF vs C 7 -KF RMSE 11. 0 % 4. 0 % 8. 8 % 5. 2 % 13. 0 % 3. 2 % STD 12. 3 % 4. 0 % 12. 7 % 5. 2 % 12. 6 % 3. 2 % = 8

benefit COSMO-2 vs COSMO-7 -KF? ? 04. 08 – 31. 10. 08 RMSE C

benefit COSMO-2 vs COSMO-7 -KF? ? 04. 08 – 31. 10. 08 RMSE C 2 -DMO vs C 7 -KF STD C 2 -DMO vs C 7 -KF All ANETZ stations -25 % -20 % Low ANETZ stations -22 % -12 % High ANETZ stations -27 % -20 % Statistical PP | COSMO-GM 2009 Vanessa Stauch 9

stations for wind speed calibration Schaffhausen (SHA) Chasseral (CHA) Üetliberg (UEB) Oron (ORO) Piz

stations for wind speed calibration Schaffhausen (SHA) Chasseral (CHA) Üetliberg (UEB) Oron (ORO) Piz Martegnas (PMA) Gütsch (GUE) SMN station WKA Evionnaz (EVI) Statistical PP | COSMO-GM 2009 Vanessa Stauch 10

height differences CHA/Wi. Cro SHA UEB ORO Wi. Fel Wi. Gue PMA EVI/Wi. Col

height differences CHA/Wi. Cro SHA UEB ORO Wi. Fel Wi. Gue PMA EVI/Wi. Col Station CHA EVI GUE ORO PMA SHA UEB Wi. Col Wi. Fel Wi. Gue Wi. Cro Höhe (obs) 1599 480 2287 830 2670 437 1043 450 1020 2331 1230 Wi. Cro Wi. Col Wi. Gue ORO PMA SHA UEB Wi. Col Wi. Fel Wi. Gue Wi. Cro Höhe (mod) 1088 1024 2322 811 2334 432 541 1024 1013 2322 1088 deltaheight_7 (mod-obs) -511 544 35 -19 -336 -4. 8 -502 574 -6. 8 -8. 8 -142 cosmo 2 CHA EVI Wi. Gue ORO PMA SHA UEB Wi. Col Wi. Fel Wi. Gue Wi. Cro Höhe (mod) 1293 746 2298 805 2520 477 604 796 941 2298 1114 deltaheight_2 (mod-obs) -306 266 11 -25 -150 40 -439 346 -79 -33 -116 cosmo 7 Statistical PP | COSMO-GM 2009 Vanessa Stauch 11

represenativeness of met. station wind turbine Gütsch model prediction representative for (mean) grid box

represenativeness of met. station wind turbine Gütsch model prediction representative for (mean) grid box Statistical PP | COSMO-GM 2009 Vanessa Stauch local point observation (specific conditions) 12

COSMO-7 vs COSMO-2 CHA ORO SHA UEB PMA GUE r. RMSE (%) für 1

COSMO-7 vs COSMO-2 CHA ORO SHA UEB PMA GUE r. RMSE (%) für 1 -24 h, period 01. 09. 08 – 31. 03. 09 COSM O-7 Chasseral (CHA) Evionnaz (EVI) Gütsch (GUE) Oron (ORO) Statistical PP | COSMO-GM 2009 Vanessa Stauch EVI COSMO 2 (03) 56 > 44 90 < 99 55 53 > < 45 59 13

effect on MOS-postprocessing CHA ORO SHA UEB PMA GUE r. RMSE (%) für 1

effect on MOS-postprocessing CHA ORO SHA UEB PMA GUE r. RMSE (%) für 1 -24 h, period 01. 09. 08 – 31. 03. 09 COSM O-7 MOS Chasseral (CHA) Evionnaz (EVI) Gütsch (GUE) Statistical PP | COSMO-GM 2009 Vanessa Stauch EVI COSMO 2 (03) MOS 34 > 32 78 > 77 45 > 39 14

effect on KF-postprocessing CHA ORO SHA UEB PMA GUE r. RMSE (%) für 1

effect on KF-postprocessing CHA ORO SHA UEB PMA GUE r. RMSE (%) für 1 -24 h, Zeitraum 01. 09. 08 – 31. 03. 09 EVI COSM O-7 KF Chasseral (CHA) Evionnaz (EVI) Gütsch (GUE) Oron (ORO) Statistical PP | COSMO-GM 2009 Vanessa Stauch COSMO 2 (03) KF 33 = 33 77 < 83 47 52 > > 43 50 15

summary >> statistical postprocessing profits from a better NWP input model >> „dynamical downscaling“

summary >> statistical postprocessing profits from a better NWP input model >> „dynamical downscaling“ does not replace statistical adaptation to local observations (in particular if results being verified against those) Statistical PP | COSMO-GM 2009 Vanessa Stauch 16