Evaluating the impact on NWP of sea level atmospheric pressure data over the ocean from drifting boys By L.R. Centurioni (1) and R.F. Lumpkin (2) (1) PI of the Global Drifter Program CIMEC/Scripps Institution of Oceanography; (2) PI f the Global Drifter Program, AOML NOAA
Goals of the GDP The overall objectives of the GDP are to: 1) maintain a network of at least 1250 Lagrangian drifters (5 X5 ) that, through the Argos and Iridium satellite systems, returns data of meteo marine variables including near surface ocean currents, sea surface temperature (SST), sea surface salinity (SSS), sea level atmospheric pressure (SLP), sealevel winds (SLW) and subsurface temperature (Tz). 2) to provide a data processing system for the scientific use of the data.
Why the GDP buys barometers? SVPB drifters array provide global SLP measurements for: a) Correction of inverse barometer effect (1hPa=1cm) of interest for oceanographers; b) NWP of interest for NWS. Co operation between Oc Met; Hurricane drifter array: targeted deployments of drifting temperature chains (0 150m), and drifters with sea level wind and air pressure sensors of interest for oceanographers and meteorologists.
Implementation of the barometer array GDP SIO buys 290 barometer upgrades/year; An additional 190 barometers are purchased every 2 nd year by GDP SIO; Another 100 GDP AOML drifters are upgraded to barometer every year by WS (Australia, New Zealand, South Africa, etc.;) ~80 SVP/year are purchased/deployed by E SURFMAR; Total: 565 barometers/year (=> $565K/year, $400K/year from NOAA s GDP funds); While the DBCP has recommended outfitting the whole GDP array with barometers by 2012, the current funding level suggests that this target will be delayed or not met even in years to come. Drifters are deployed by VOS or by Research or Operational Agencies; $500K additional would be required to fit each drifter with a barometer.
Workshop held on May 21, 2012 in Sedona, AZ, between GDP and NWP users Three possible ways to assess the impact of SLP on NWP Impact of observations (fast and cheap, uses adjoint. Addressed with this talk)* OSE (long and expensive)** OSSE (longest and most expensive)*** *based on a definition of total energy norm. **also called data denial. Compares denial with control runs. Can use to assess the resilience of the system (self compensating effects) ***requires generation of synthetic observations.
Definition of Observation Impact following Langland and Baker (2004); extended for nonlinear analysis schemes by Trémolet (2008) Forecast Verifying Analysis Analysis Background Observation Impact: δe < 0 the observation(s) improve the forecast Credit Ron Gelaro, NASA
BUOYS-SHIP DFS and FEC Monthly Average DFS 150 W 120 W 90 W 60 W 30 W 0 E 30 E 60 E 90 E 120 E 150 E 60 N 30 N 0 N 30 S 60 S 1 4.47 4.25 4.04 3.83 3.61 3.40 3.18 2.97 2.75 2.54 2.33 2.11 1.90 1.68 1.47 1.25 1.04 0.82 0.61 0.14 0 0.7 0.5 0.3 150 W 120 W 90 W 60 W 30 W 0E 30 E 60 E 90 E 120 E 150 E 60 N 30 N FEC 0 N 30 S 60 S 150 W 120 W 90 W 60 W 30 W 0 E 30 E 60 E 90 E 120 E 150 E Credit Claudia Cardinali, ECMWF WMO DBCP Workshop Sedona 2012 Slide 7 41.93 27.02 24.02 21.02 18.01 15.01 12.01 9.01 6.00 3.00 0.00-0.00-3.00-6.00-9.01-12.01-15.01-18.01-21.02-24.02-138.46 Min=-138 Max=41 Mean=-2.5 ECMWF
Summary of All Data Counts (Used) Global Domain January 2012 July 2011 Buoys are among the least numerous data types assimilated Credit Ron Gelaro, NASA 8
Summary of Observation Total Impact Global Domain January 2012 July 2011 Shading indicates observation count (buoys are among the least numerous data types assimilated) Credit Ron Gelaro, NASA 9
Credit Ron Gelaro, NASA Summary of Impact Per Observation Global Domain January 2012 July 2011 On a per ob basis, buoys have among the largest beneficial impacts of all observation types in terms of the 24h global error metric Only dropsondes in January and PIBALS in July have larger impact per ob 10
Fraction of Beneficial Observations S. Hemisphere January 2012 July 2011 Buoys have the largest or nearly largest fraction of beneficial observations in most locations (globe, NH, SH) in both seasons Credit Ron Gelaro, NASA 11
Forecast impact experiment from Dec. 2010 to Jan. 2011 Impact Impact / Obs. number BUOY BUOY 12 Credit Jean Francois Mahfouf, MeteoFrance
Timeliness of observations BUOY H+2 =>70 % H+1 =>55 % H+2 =>82 % H+1 =>67 % SHIP 13 Credit Jean Francois Mahfouf, MeteoFrance
Credit John Eyre, Met Office Satellite surface wind impact Forecast Sensitivity to Observations (FSO) Increasing fractional increase SHIP BUOY HIRS Satwind AMSUA AIRS MODIS/AVHRR PILOT IASI Dropsonde SEVIRI SYNOP SSMIS AMSUB TEMP Aircraft WINPRO GPSRO TCBOGUS Scatwind BOGUS Control Scat_denial -5% 0% 5% 10% 15% 20% 25% 30% 35% When ASCAT, ERS-2 and WindSat winds are denied, other surface-marine observations partially compensate Crown copyright 2007
Credit Claudia Cardinali, ECMWF 6 th November: Case of a rapidly developing cyclogenesis Analysis 12h Forecast 24h Forecast 36h Forecast Minimum pressure from 990 to 950 hpa between 00Z 6/11 and 18Z 7/11 WMO DBCP Workshop Sedona 2012 Slide 15 ECMWF
Credit Claudia Cardinali, ECMWF 00Z 6 th November FEC in the 30 x30 DRIBU MSL pressure WMO DBCP Workshop Sedona 2012 Slide 16 ECMWF
Results: SP-Denial versus Control Credit Claudia Cardinali, ECMWF Control better Relative error diff Denial better WMO DBCP Workshop Sedona 2012 Slide 17 ECMWF
CONCLUSIONS 1. Impact of SLP from drifters on NWP is extremely positive; 2. Adopt alternate metrics of high relevance (i.e. surface Kinetic Energy => wind) 3. At least one OSE specific to drifter data should be run to have extra proof and to understand the effect of the reduction ($60 $80K, ~12% of array cost for one year);