3rd thorpex international science symposium, monterey, usa, 14 sept. 2009 overview of t-parc falcon...
TRANSCRIPT
3rd THORPEX International Science Symposium, Monterey, USA, 14 Sept. 2009
Overview of T-PARC Falcon operations and ECMWF data denial experiments
Martin Weissmann, Florian Harnisch, Stephan Rahm, Martin WirthInstitut für Physik der Atmosphäre, DLR Oberpfaffenhofen, Germany
Pat Harr, Naval Postgraduate School, USATetsuo Nakazawa and Kotaro Bessho, MRI-JMA, Japan
Sarah Jones and Doris Anwender, Universität Karlsruhe/Forschungszentrum Karlsruhe, GermanyDavid Parsons, World Meteorological Organization
Hee-Sang Lee, National Institute for Meteorologial Research, KoreaRon McTaggert-Cowan, Environment Canada
Carla Cardinali, ECMWF, United Kingdom
3rd THORPEX International Science Symposium, Monterey, USA, 14 Sept. 2009
Falcon flights
Falcon objectives:- typhoon targeting- extratropical transition of tropical cylcones- lidar observations for NWP
Falcon funding institutionsDLR (Germany), NSF (USA), JMA (Japan) Forschungszentrum Karlsruhe (Germany), NIMR (Korea)Environment Canada, EUCOS
Falcon operations23 August - 3 October 2008 in Japan25 research flights 93 flight hours in total
hosted by US Navy @ NAF Atsugi (near Tokyo)
Overview of Falcon operations
3rd THORPEX International Science Symposium, Monterey, USA, 14 Sept. 2009
source of satellite images: https://www.fnmoc.navy.mil/tcweb/cgi-bin/tc_home.cgi/
Sinlaku
Japan
IR Satellite Image09/11/08 2030 LT
Highlight 1: Water vapour lidar observations in typhoon environment (11 Sept. 2008)
16W
S-Korea Japan
OkinawaIR Satellite Image09/11/08 1830 LT
contentrations of 0.01-15 g/kg measured by new worldwide unique 4-wavelength DIAL
3rd THORPEX International Science Symposium, Monterey, USA, 14 Sept. 2009
Highlight 2: Wind lidar observations
30 Sept 2008: flight into jet for ET objectives
A-TReC 2003
A-TReC 2003: ~ 3% reduction of mean forecast error over Europe by lidar (Weissmann and Cardinali, QJ 2007)GOAL: investigate the general validity of these results with larger data set (60 flight hours during T-PARC)T-PARC: twice the amount of A-TReC lidar data
A-TReC 2003
3rd THORPEX International Science Symposium, Monterey, USA, 14 Sept. 2009
Highlight 3: Joint mission for typhoon targeting (TY center, surrounding and sensitive area)
Concept for ideal mission: joint mission on 11 September WC-130 observations in typhoon center (green) DOTSTAR observations in typhoon surrounding (blue) Falcon obs. in sensitive area highlighted by e.g. SV, ETKF (red)
sensitivity experiments by Florian Harnisch --> poster in the afternoon ECMWF experiments for center, surrounding, sensitive area
Japan
China
ETKF
SV
3rd THORPEX International Science Symposium, Monterey, USA, 14 Sept. 2009
Tracks of typhoons Sinlaku and Jangmi with flight times
3rd THORPEX International Science Symposium, Monterey, USA, 14 Sept. 2009
Influence of T-PARC observations an ECMWF typhoon track prediction
gray: no dropsondesgreen: operational dropsonde data (two different model versions)magenta: with time-corrected dropsondes
on average, forecast error reduction with corrected dropsondeson average, operational setup has similar errors as without dropsondes
improvements at a high level of forecast skill, no real forecast busts
Sinlaku Jangmi
smallsample
significant
3rd THORPEX International Science Symposium, Monterey, USA, 14 Sept. 2009
Influence of T-PARC observations an ECMWF typhoon track prediction
gray: no dropsondesgreen: operational dropsonde data (two different model versions)magenta: with time-corrected dropsondes
on average, forecast error reduction with corrected dropsondeson average, operational setup has similar errors as without dropsondes
improvements at a high level of forecast skill, no real forecast busts
Sinlaku Jangmi
smallsample
from Jim Doyle, Navy Research Lab., USAsignificant
3rd THORPEX International Science Symposium, Monterey, USA, 14 Sept. 2009
Examples of improving cases
uncertainties of recurvature point --> improvement with dropsondes
largest improvement with DOTSTAR dropsondes, degradation with WC-130 observations at 2008091000
(for details see poster by F. Harnisch in the afternoon)
3rd THORPEX International Science Symposium, Monterey, USA, 14 Sept. 2009
Examples of neutral and slightly deteriorating cases
- very good forecast without observations
- remaining errors are supposedly rather connected to problems with land interaction than initial condition errors
(for details see poster by F. Harnisch in the afternoon)
3rd THORPEX International Science Symposium, Monterey, USA, 14 Sept. 2009
Dropsonde observation statistics for WC-130 flight through typhoon center
Dropsondes within the typhoon are successfully assimilated in the ECMW model, but only a relatively small part of the information can be used with current DA systems due to large differences to the model first guess field
Potential for further improvements through flow dependant covariance matrix, e.g. ensemble 4D-Var
Center observations led to significant analysis differences, mainly typhoon intensification, but relatively little forecast improvements
3rd THORPEX International Science Symposium, Monterey, USA, 14 Sept. 2009
Period: 2008090900 - 2008100112
driftsondes
ET
recurvature
T-PARC data coverage @ ECMWF
formation
3rd THORPEX International Science Symposium, Monterey, USA, 14 Sept. 2009
Influence on mid-latitude forecast error
Z 500 hPa Pacific: 30-65N, 155E - 130W
Z 500 hPa northern hemisphere 20-90N
reduction of mid-range forecast error over the Pacific and long range northern hemisphere forecast error
3rd THORPEX International Science Symposium, Monterey, USA, 14 Sept. 2009
Influence on mid-latitude forecast error (uncycled)
Z 500 hPa Pacific: 30-65N, 155E - 130W
no improvement without "cycling", i.e. when the analysis uses the first-guess from the control run
3rd THORPEX International Science Symposium, Monterey, USA, 14 Sept. 2009
on average, improved typhoon track prediction with additional dropsondes
improvements in particular at early typhoon stages before recurvature
largest improvements through observations in the surrounding of the storm (DOTSTAR strategy) rather than in the typhoon center or sensitive areas indicated by singular vectors or ETKF (see poster)
timing errors in operational dropsonde data set at ECMWF, corrected for the future and to my knowledge no other NWP center is affected
in mid-latitudes, improved mid-range forecast over the Pacific and long range forecast on the northern hemisphere
nearly no improvement without "cycling", i.e. when the analysis uses the first-guess from the control run
improvements are moderate, ECMWF forecasts without dropsondes are already relatively accurate
studies to investigate the influence of airborne lidar observations will follow
Thanks to:- all people involved in T-PARC- Milan Dragosavac, Ioannis Mallas and all others at ECMWF involved
Conclusions (preliminary results)