Utilization of satellites and products at Met Office since ET-SUP7 Simon J. Keogh WMO ET-SUP8 14-17 April 2014.
Table of Contents Current data usage for NWP including new data assimilated since ET-SUP7 New imagery products evaluated since ET-SUP7 Ocean Forecasting Applications Climate Research Space Weather - MOSWOC Future plans
Current data usage for NWP including new data assimilated since ET-SUP7
Production NWP (end FY2013-14) 2.2km ensemble Up to 36hr f/c 6-hourly update Retired the North Atlantic Model and 18km MOGREPS-R 1.5km model Up to 36hr f/c 3-hourly update 33km ensemble Up to 3day f/c 6-hourly update 4.4km model Up to 120hr f/c 6-hourly update 60km coupled model Up to 6 months Daily lagged ensemble 17km model Up to 144hr f/c 6-hourly update
PS 34 Satellite Package To go live in May 2014 Changes in spatial thinning of: IASI ( 2 increase in data) ATOVS (30% ) Scatterometer data (50% ) Improved assimilation of GPS Radio Occultation data (allowing for tangent point drift) Introduction of Meteosat-7 MVIRI clear sky radiances over the Indian Ocean Changes to snow analysis (use of JULES snow depth and amount) Tested as individual components against ND, and as a package against ND and ENDGAME over 2 seasons (Nov and July 2012)
Satellite data used in NWP (1) April 2014 Observation type Satellites NWP models * AMSU/MHS radiances 4 NOAA + Metop G, R HIRS clear radiances 2 NOAA + Metop G, R IASI and AIRS clear+cloudy radiances Metop + Aqua G, R ATMS & CrIS radiances Suomi NPP G *SSMIS radiances F16 G, R Geo imager clear IR radiances MSG, MFG, GOES, MTSAT2 G, R, UK GPS RO bending angles 5 COSMIC, Metop/GRAS, GRACE-A, TerraSAR-X, CNOFS G, R GPS ZTDs ~350 European stations G, R, UK * G=Global, R=regional=Europe, UK=UK area
Satellite data used in NWP (2) April 2014 Observation type Satellites NWP models * AMVs geo 5 geo satellites G, R, UK AMVs MODIS and AVHRR Aqua, Terra, NOAA, Metop G, R Scatterometers: sea-surface winds Metop/ASCAT, *Oceansat-2/OSCAT G, R, UK MW imager sea-surface winds: Windsat Coriolis G, R SEVIRI cloud height/amount MSG R, UK SSTs: AVHRR, AMSR-E NOAA, Metop, Aqua G, R, UK Soil moisture: ASCAT Metop G, R, UK Sea ice: SSM/I, SSMIS DMSP G, R Snow cover various G, R * G=Global, R=regional=Europe, UK=UK area
Forecast impacts of satellite observations This plot shows the impact per day of various observation types FSO statistics developed in partnership with KMA There is a large impact from IASI on MetOp-A
Impacts of MetOp-B IASI on NWP index Experiment run with various thinning options for IASI It is planned to switch to using 80km thinning in the extratropics and 3 hours temporally. This change is scheduled for May 2014. No preference will be given to MetOp-A or MetOp-B data. The number of IASI observations assimilated will approximately double. The resolution of the analysis will increase from N216 to N320. Further assimilation experiments will be required to test the impact of IASI at this resolution
ASCAT-B NWP impact experiments. The impact of assimilating ASCAT-B winds has been investigated using the two different spatial thinning schemes: operational and individual. Both thinning schemes show a similar, small positive impact on NWP index scores despite the large differences in the number of observations used. Results suggest that ASCAT-B mostly provides additional impact when we add observations into areas not already observed by ASCAT-A. An outage of Oceansat-2 data during the trial period shows that the addition of ASCAT-B leads to a more resilient system and the benefit of a second ASCAT is greatest when we lose another important instrument such as OSCAT.
Example coverage of Scatterometer Data (00Z, 4 Dec 2012)
R&D efforts to utilize the Special Sensor Microwave Imager Sounder (SSMIS) for NWP The instruments are conical scanning radiometers Operate by reflecting upwelling radiation off a large rotating mirror into a group of apertures Unfortunately this design suffers several calibration anomalies, such as: Reflector emissions - the mirror isn t perfect Solar intrusions - into the warm calibration target F16 removed from Met Office assimilation system on 1 st May 2013 due to failure of 56.4GHz oscillator on 25 th April 2013.
SSMIS Calibration anomalies The calibration anomalies manifest as complex, systematic biases Elucidated through comparisons of observed brightness temperatures (O) with those from NWP model backgrounds (B) For example all three instruments exhibit ascending/descending biases Cooler descending pass F-16, Ch24 (60GHz) F-17, Ch6 (unave) (57GHz) C-B (K)..these data have been corrected (C) through pre-processing and using the Harris & Kelly scheme C-B (K) F-18, Ch5 (55GHz) C-B (K) Warmer ascending pass
SSMIS bias correction work Watch this space in future for results
UK Environmental Prediction
Trialing UKAMVs in the UKV, UK4 & Nowcasting Model Domains Model Resolution VAR Time Window Cycling Forecast Length UK4 / UKV 4 km / 1.5km 3D-Var 4/3km 3 hr 3 hr T+36 South UK Fixed 1.5 km 3D/4D-Var 1.5/3km 1 hr 1 hr T+6 or T+12
High Resolution AMVs for assimilation into local models Use of Nowcasting SAF software to produce high resolution AMVs Increased timeliness is important for short range forecasting Challenging to manipulate the BUFR (initially used McIDAS) Assimilated into our UK models with various thinning options Gave neutral impact but impact is hard to measure
Assimilating IASI data into the Met Office convective scale UKV model IASI data will be introduced using same configuration as global except: All 4 fields of view used (1/4 for global) 60km thinning (120km for global) 6 high peaking water vapour channels rejected due to large residual bias Several different configurations were tested varying thinning, observation errors and channel selections Small positive impact indicated by: Improved background fits to MHS and SEVIRI Improved surface temperature forecasts
New imagery products evaluated since ET-SUP7
Satellite Rainfall Rate to Complement Radar Coverage Radar Meteosat-10
Monitoring the Indian Ocean
FY-3B imagery from direct broadcast We are running the direct broadcast processing package from CMA fy3l0pp for processing raw data to level 0 fy3l1pp for processing level 0 to level 1 Example image from MERSI instrument
EARS-VIIRS Extension to VIIRS coverage. Data in Compact VIIRS format Lannion station now added by EUMETSAT and more stations to follow
Sandwich imagery ISS photograph showing overshooting convection,west Africa Use textural information in the high resolution visible imagery to superimpose detail on to IR observations
Ocean Forecasting Applications
SST Assimilation in Shelf-Seas Reanalysis Uses NEMOVAR vn.3.4 (3D-VAR) In shelf seas we only assimilate SST from in situ sources and satellites. Assimilation of altimeter and profile observations will be implemented in future upgrades. Includes new flow-dependent error covariance specification based on the potential vorticity gradient. Satellite SST observations assimilated over reanalysis (1984-2012) SST observations now supplied with individual observation errors This is combined with estimated representativity error. Observations with large measurement error are down weighted in the analysis. In 1 month test, RMS errors against in situ data were 0.27ºC when using a global observation error and 0.23ºC with measurement errors considered separately. 0 5 0 5 Mean measurement (left) and representativity error (right) for September-November (ºC). Assimilation of SST show improvement over free model run. Significant reduction in RMSE and bias. RMSE (solid line) and bias of observation-model SST error for the reanalysis (black lines) and a free model run (blue).
AltiKa AltiKa, the altimeter built by CNES and prime payload of the Indian SARAL satellite Launched 25 February 2013 Operates in the Ka band (higher frequency than other altimeters) Observes to quite high latitudes - orbit inclination 98.5 deg (Jason-2 66 degrees, Cryosat-2 92 degrees)
AltiKa in FOAM Altimeter SSH used to correct the mesoscale ocean circulation in FOAM AltiKa went into the FOAM operational suite on 2013/10/21 Took some time due to testing and calibration of the data by the data providers Additionally we are somewhat cautious in introducing data to the operational system Jason-2 AltiKa Cryosat-2
Impact of AltiKa in FOAM AltiKa added to FOAM 0.1 0 15000 every 6 hrs ~8 cm RMSE 1-Sep 1-Oct 1-Nov ~7 cm RMSE A comprehensive assessment of the data (see the SARAL/AltiKa 1st verification workshop) showed that it was as good if not better than Jason-2. Because of this and based on the assessments of other operational centres (e.g. CSIRO) we decided to include the data in FOAM. 2 sats We show here, after AltiKa data is included that the number of observation points is increased 0 3 sats The RMS observation minus model background difference is reduced from 8 cm to 7 cm.
Climate Research
Climate research CALIPSO lidar used to study mid-level cloud CERES sensor used to show differences in model and observed top of the atmosphere radiances
Is mid-level really mid-level? HIGH MID LOW Backscatter from CALIPSO lidar km
Space Weather Artist impression from Wikipedia
Operations Centre MOSWOC Met Office Space Weather Operations Centre Signed MoA with NOAA SWPC. Daily forecast coordination. Operational 24/7 as of 2014 April. Accessing data & imagery from NOAA and NASA. Add UK centric advice and impacts. Scientific partners: Academic partnerships with UK institutes. International (NASA, NOAA, WMO & ESA activities).
Developing our own services Receiving space weather data and alerts from collaborators, e.g., NOAA SWPC, British Geological Survey. Visualisation of data suite for forecasters. WSA Enlil runs operationally. Daily space weather summaries to UK government and other relevant bodies. Enhanced activity during possible major events (e.g. X class flares & CMEs, May 2013).
Visualisation Suite ENLIL Model Runs 12 x / day Interleaves with NOAA runs
What future satellite data would we like to use? L1 data from NOAA DSCOVR Future coronograph data (NASA, ESA, wherever) NASA SunJammer (launch 2015). Obs at closer to the Sun than L1 (this is a research mission, but useful for proof of concept) GNSS RO need data with latency of 15-30 mins. NOAA trying to set up network of ground stations for COSMIC II to achieve this SWARM thermospheric density (currently few/no obs of this). Can exploit this with our recently written DA scheme. But need the SWARM data to be NRT GEO/LEO radiation monitor data (electron/proton flux) only have GOES 13 and 15; other data are potentially available
What concerns do we have about present or future observations gaps? L1 obs (ACE) ACE is old (past planned lifetime) and single point of failure. Loss of ACE means we can t give warning (with 30-45 mins lead time) of geoeffective CME DSCOVR should be launched early in 2015. Need more L1 obs for system robustness CME tracking Uses SOHO coronograph (old, SPOF, like ACE) plus STEREO. But STEREO satellites are heading round the back of the Sun, so less optimal in next few years. Without these data it s hard to diagnose CME direction and speed (and to use this in CME forecast models) Need new coronographs US one planned? ESA phase C/D study upcoming Also L4/L5 ESA mission study planned (long way off from starting a programme)
Summary We have improved the assimilation of IASI and ASCAT and added CSR assimilation over the Indian Ocean We are looking to assimilate SSMIS using a novel bias correction. We are looking forward to using F19 data. We are utilising a variety of Geostationary satellites for imagery. We look forward to using Himawari-8 and GOES-R. AltiKa altimeter has given impressive results Calipso and CERES are providing useful data for climate research MOSWOC has been set up to mirror operations at NOAA SWPC
Questions and answers