Observations for improving high impact weather forecasts from convective to global scales. Roger Saunders and Stefan Klink with input from DAOS-WG members
Overview A good and bad example Satellite Observations In-situ Observations Requirements for high impact weather forecasting
MOGREPS-15 forecast evolution 925hPa wind (gust indicator) > 60kt From 8 days to 2 days ahead Note consistency of signal for highest probability in Biscay throughout the evolution Although ensemble alerted low risk of storm further north, most likely was a useful guide throughout
Outcome 00UTC 16 th Dec Low tracked along English Channel Strong winds in Northern France and Germany but not UK Snow in Wales and English Midlands Ensemble provided excellent guidance and risk assessment throughout St Malo, Brittany
Messages for THORPEX Became TS Joyce TS Isaac
Sri Lankan Storm 25 Nov 11 ASCAT products Many fishermen were lost off the south coast of Sri Lanka Two inconsistent sets of retrieved winds
Sri Lankan Storm 25 Nov 11 ECMWF EFI Many fishermen were lost off the south coast of Sri Lanka
Satellite Observations Crown Copyright 2012. Source: Met Office
Crown Copyright 2012. Source: Met Office Global Observations 00UTC 16/8/2011
Satellite data usage at ECMWF: past, present and near future Millions of observations assimilated per 24h period 10
Observation Impacts in Met Office global NWP Observation Types Metop: AMSU-A, MHS, HIRS, IASI, ASCAT, GRAS Relative Contribution of Observations to NWP forecast NOAA: AMSU-A: N-15, N-18, N-19 MHS: N-18 HIRS: N-17, N-19 AVHRR AMVs: N-15, N-16, N-17, N-18, N-19 Other LEO: EOS-Aqua AIRS, MODIS AMVs EOS-Terra MODIS AMVs DMSP F-16 SSMIS ERS-2 AMI; Coriolis WINDSAT METOP NOAA "SONDE" OTHER LEO AIRCRAFT SFC LAND GEO SFC SEA OTHER RO GEO: Other RO: Aircraft: SONDE : GOES AMVs; MTSAT AMVs; Meteosat AMVs, CLRs CHAMP, GRACE AMDAR, AIREP PILOT, TEMP, Wind profiler, DROPSONDE 0 5 10 15 20 25 30 Relative Observation Impact[%] Forecast sensitivity to observations Surface land: SYNOP, BOGUS Surface sea: BUOY, SHIP, TCBOGUS
ECMWF satellite data denial for Sandy DT 00 UTC 25 October Operational 5 day forecast (left) Rerun with no satellite data (centre) Veryfing analysis (right)
Observation Volumes in 6 hours (20/10/08) Increased resolution=>in creased Obs usage Category Count % used Category Count % used TEMPs 637 99% Satwinds: JMA 26103 4% PILOTs 307 99% Satwinds: NESDIS 142478 3% Wind Profiler 1355 39% Satwinds: EUMETSAT 220957 1% Land Synops 16551 99% Scatwinds: Seawinds 436566 1% Ships 3034 84% Scatwinds: ERS 27075 2% Buoys 8727 63% Scatwinds: ASCAT 241626 4% Amdars 64147 23% SSMI/S 532140 1% Aireps 7144 12% SSMI 698048 1% GPS-RO 776 99% ATOVS 1127224 3% AIRS 75824 6% IASI 80280 3%
14 A. Doerenbecher
Typhoon Sinlaku (2008): Satellite AMVs HOURLY AMVS INCLUDED ALL AMVS REMOVED RAPID-SCAN ADDED Crown Copyright 2012. Source: Met Office 15 Adapted by Rolf Langland from Berger et al. (2011)
T-PARC Summer Phase (2008): Overall impact of dropwindsondes on TC track 16 Weissmann et al. (2011)
2010/11 WSR campaign
Scatterplot of impacts For each case where dropsondes were launched, a downstream target location and verification time are identified. Data are plotted here only for these times/target locations. Verification area here is a +/- 10 degree box centered on target. Verification norm is an approximation to the total-energy norm. Cases above line indicate benefit from targeted data. No obvious beneficial impact.
Summary of targeting Clear benefits of targeted observations for TC forecasts. In study with 4D-Var assimilation & ECMWF model, little impact seen for mid-latitude systems (2011 WSR) but continue to study this. Targeting s future: Global observational network design: use targeting strategies to adaptively select / thin satellite data for assimilation? Potential to adapt targeting concepts to work in concert with rapidly adaptable observational resources, e.g., cloud-drift winds from rapid-scan imagery. But: need quick, efficient algorithms.
Convective scale data assimilation is still being developed Crown Copyright 2012. Source: Met Office
Impact of additional humidity observations Legend Specific humidity colour scale over land and sea Control Trial 1 Trial 2 Trial 3 Control: standard UK4 dataset, including clear SEVIRI obs Control has large (unphysical) regions of zero humidity (black) UK4 Level 45 ~7km Specific humidity in analysis fields 03Z on 13/2/2012 UK4 Level 30 ~3km Trial description 1. Adding SEVIRI channel 5 over low cloud 2. Adding AMSU-B data 3. Adding SEVIRI channel 5 over low cloud and AMSU-B data Improve ment in UK Index: 0.24% 0.42% 0.77% Crown copyright Met Office
Sting jet observations from Meteosat Crown Copyright 2012. Source: Met Office
In-Situ Observations Crown Copyright 2012. Source: Met Office
EUMETNET E-AMDAR Humidity Trial Commercial aircraft measurements complement but could also partly replace traditional radiosonde soundings; EUMETNET STAC is currently working on a business case for a further roll-out of aircraft humidity sensors; 3 WVSS-II sensors currently tested on commercial aircraft, one on research aircraft; 6 further WVSS-II sensors will be installed in 2013. GIE/EIG EUMETNET, Registered Number 0818.801.249 - RPM Bruxelles
AMDAR q / TAMDAR study (just started) - Motivation Key questions: 1)Does it make sense to further invest money into procurement of more WVSS-II humidity sensors to be installed on commercial aircraft which belong to the E- AMDAR programme? 2)Is TAMDAR an alternative to E-AMDAR observations or is it complementary to E-AMDAR? 3)In case WVSS-II or TAMDAR sensors prove to have a beneficial impact on NWP forecast skill, how many sensors should be installed? GIE/EIG EUMETNET, Registered Number 0818.801.249 - RPM Bruxelles
US AMDAR humidity coverage US Midwest Parameter shown: AMDAR humidity on 2nd April 2012, 00:00-23:59 Picture taken from NOAA AMDAR webpage GIE/EIG EUMETNET, Registered Number 0818.801.249 - RPM Bruxelles
Experiment design Experiment type: OSE Models: Verification: Global: DWD, ECMWF?, Météo-France? LAM over US (Midwest region): DWD, Météo-France? Focus on verification against observations Special emphasis should be given on verification of precipitation forecasts: against hourly SYNOP data or against US radar-derived precipitation? Possible? GIE/EIG EUMETNET, Registered Number 0818.801.249 - RPM Bruxelles
AMDAR q/ TAMDAR OSE Scenarios Scenario no 1: Baseline All operationally used satellite observations + all screen level observations (SYNOP, BUOY, SHIP, METAR, ) + all radiosondes (temperature and humidity) and all AMDAR (temperature and wind only); (no radar-derived precipitation, no GPS, no AMDAR humidity, no TAMDAR, this reflects current European operational setup more or less) GIE/EIG EUMETNET, Registered Number 0818.801.249 - RPM Bruxelles
AMDAR q/ TAMDAR OSE Scenarios Scenario no 2: Baseline + AMDAR humidity Scenario no 3: Baseline + TAMDAR (temperature, wind and humidity) Scenario no 4: Baseline + AMDAR humidity + TAMDAR (temperature, wind and humidity) GIE/EIG EUMETNET, Registered Number 0818.801.249 - RPM Bruxelles
Aircraft (AMDAR) Observations AMDARs ModeS TAMDAR FlyBE Observations: T, wind, RH, icing, turbulence, etc Challenges: getting data into building, licensing agreements, correlated error Crown Copyright 2012. Source: Met Office estimation, bias correction (e.g. VarBC), assimilation of icing, turbulence data.
EUCOS 2nd Space-Terrestrial Study Radnoti, G., P. Bauer, A. McNally, A. Horanyi: ECMWF study to quantify the interaction between terrestrial and space-based observing systems on Numerical Weather Prediction skill Deliverables: assess the impact of a thinned terrestrial observing system on radiance bias correction anchoring and to investigate the impact of a reduced conventional observing system on NWP skill; Published as ECMWF Technical Memorandum 679, 18th July 2012. GIE/EIG EUMETNET, Registered Number 0818.801.249 - RPM Bruxelles
EUCOS 2nd Space-Terrestrial Study HIW, example: impact of buoy data MSLP, 96h forecast, valid at 0 UTC 24 Jan 2009 (storm Klaus) from analysis without (top) and with (bottom) GPRSO data, verifying analysis (courtesy of: Gabor Radnoti, ECMWF) with assimilating full buoy, thinned, and no buoy network. GIE/EIG EUMETNET, Registered Number 0818.801.249 - RPM Bruxelles
ECMWF 2011 IFS cycle 37r2: Better short range precipitation forecasts from assimilation of USA NEXRAD radar rainfall data Change in 24-hour precipitation threat scores due to direct 4D-Var assimilation of NCEP Stage IV rain radar and gauge data [1 April 6 June 2010]. CTRL NEW CTRL NEW Slide 33
EUMETNET OPERA Odyssey ODC Data Hub Composite delivery to 18 EUMETNET or OPERA members*, ECMWF and H-SAF starting 2013; Volume delivery to HIRLAM 2013-2014; Three composites defined: - maximum reflectivity, - rain intensity and - rain accumulation (1 hr); Deliverable for 2013-2014 period: - prepare for dissemination of 3D volume single-site data (reflectivity and radial winds) * Including Romania and Italy GIE/EIG EUMETNET, Registered Number 0818.801.249 - RPM Bruxelles EUCOS PM Meeting, Brussels, 1-2 February 2012
ΔFSS 0.2 mm acc scale 55km Doppler Radial Wind Assimilation 6 radars currently providing radial winds Plans to upgrade whole network by 2013 1 st assimilation in UKV: July 2011 (4 radars) Crown Copyright 2012. Source: Met Office David Simonin
Example Additional Data Types For Convective-Scale DA High resolution, rapid-scan imagery: SEVERI Cloud Top Height: Sang-Won Joo, KMA Radar reflectivities: Crown Copyright 2012. Source: Met Office High resolution water vapour from lidar:
Slides from Siebren de Haan, KNMI Ground-based GNSS data and RADAR radial winds assimilation in HIRLAM Cycle dedicated for ATC Input Aircraft Surface pressure Additional GPS radar radial winds Operational 2011/05 Windprofilers: 2010/08 passive FC+09: output every 10min 11km resolution GIE/EIG EUMETNET, Registered Number 0818.801.249 - RPM Bruxelles
Rainfall forecast verifcation GIE/EIG EUMETNET, Registered Number 0818.801.249 - RPM Bruxelles
00:00 01:30 03:00 04:30 06:00 07:30 09:00 10:30 12:00 13:30 15:00 16:30 18:00 19:30 21:00 22:30 00:00 01:20 02:40 04:00 05:20 06:40 08:00 09:20 10:40 12:00 13:20 14:40 16:00 17:20 18:40 20:00 21:20 22:40 Assimilate cloud optical depth? Downward SW direct and diffuse radiation 3.5 3 2.5 2 25+28 May 2012 Clearer Cloudy Cloud optical depth 1.5 1 2.5 0.5 2 Cloud optical depth 0 1.5 1 0.5 A measure of cloud optical depth 0-0.5-1 Crown Copyright 2012. Source: Met Office Courtesy Roger Saunders http://wow.metoffice.gov.uk
Roadside sensor network OpenRoad full network SYNOPS Crown Copyright 2012. Source: Met Office
Priorities for the EUMETNET Observations Programme 2013-2017 1) To foster the OPERA developments in order to be able to produce quantitatively usable 2D radar products and to exchange single site 3D volume data (reflectivity, Doppler winds) by the end of the programme phase; 2) To further expand the E-AMDAR Operational Service by trying to extend the horizontal coverage over the EUCOS area and by considering a further roll-out of humidity sensors on board E-AMDAR aircraft; 3) GIE/EIG EUMETNET, Registered Number 0818.801.249 - RPM Bruxelles
Motivation for impact studies (External) drivers and developments: Different observation networks evolve differently (e.g. regarding availability, accuracy, cost,...) Data assimilation algorithms improve and can make use of more data New requirements from km-scale NWP EUMETNET Observations Programme objective: Design and coordinate the evolution of the ground based EUMETNET composite observing system (EUCOS) to be optimized at European scale with a view to improve short range forecast... A modification of the meteorological observing network might become necessary EUMETNET Observations Programme needs approval for network changes or modifications from STAC/PFAC and EUMETNET Assembly respectively In order to get the 29 Members convinced of such changes it was decided to base them on scientific analyses (e.g. impact studies) GIE./EIG EUMETNET, Registered Number 0818.801.249 - RPM Bruxelles
Requirements Crown Copyright 2012. Source: Met Office
Requirements (Global) Global NWP has a wealth of observations but there are still some significant data voids: Global rainrate observations at hourly sampling or less Satellite data over land at low levels (T, q) Satellite and in-situ data over polar regions Stratospheric constituent concentration Measurements of solid precipitation (snow, hail) Data thinned (spatial, temporal and spectral)
Requirements (Mesoscale) Convective scale models need new observations at high spatial resolution from both surface and satellites Radar reflectivity and line of sight winds Rapid scan AMVs GPS TCVW sites Geo and polar radiances (IR/MW sounders, imagers) Groundbased lidars Regional aircraft profiles
Key Messages How to bridge the gap between data assimilation/other users (e.g. nowcasting) and observation network management in NHMSs. Satellites dominate the forecast impact for global NWP. To date targeted observations only show impact for TCs. More work on using adaptive observing systems (e.g. radiosonde launch on demand, changing sat data thinning) Convective scale DA is in its infancy and will benefit from new mesonet observing systems in addition to radar and satellite observations. We will always need more impact studies for HIW!
6 th WMO DA Symposium At NCEP, Maryland 7-11 th October 2013 Abstract submission by 30 April 2013 Global and regional atmospheric DA Convective scale DA Atmospheric constituent DA Coupled DA Global and regional ocean DA Assimilation of observations for the land surface Assimilation of satellite, in-situ, and radar observations Methodology Diagnostic tools http://www.emc.ncep.noaa.gov/wmo6da/.php