NCEP data assimilation systems status and plans John

NCEP data assimilation systems status and plans John C. Derber Environmental Modeling Center NCEP/NWS/NOAA With input from: Many others 1

Data Assimilation • At EMC we do not have resources to develop and maintain Global, Regional, Mesoscale, RTMA, AQ, Analysis of Record systems of different structures and characteristics • Unified system – GSI/ENKF hybrid • Enhancements meant for one application of the system should be applicable for other systems and should benefit other applications. • Must be certain uniformity across systems. 2

Data assimilation group collaborators • GMAO – Global – 4 d. Var – Constituent and Aerosol assimilation • DTC – Regional/Mesoscale – Support Universities and other users • ESRL – Regional – Rapid Refresh – Hybrid system • NCAR/AFWA – Regional – Operational – ARW • NESDIS • JCSDA • Others welcome 3

Collaborative development • Established Code management system • Numerous meetings with development groups • Code management and oversight group beginning through DTC 4

Future Directions • Hybrid Assimilation - En. KF/variational • 4 d-Var • Improved cloud/precipitation analysis – Inclusion of cloudy radiances • • Additional observations Improved balance Improved use of current data Storm relocation 5

Hybrid En. KF/GSI assimilation • Part of background error covariance defined by ensembles – Situation dependent background error – Improved balance between variables (includes all variables including ozone and moisture) • • • Testing beginning for global system (Whitaker based), but computational resources limited Mingjing working on hurricane system For regional/hurricane system use global ensembles? Being tested – substantial cost savings Inter-variable interactions may not always be good (e. g. , influence of ozone observations on wind field) Need to run ensembles and high resolution system whenever changes to either system or the hybrid analysis Further development work may be necessary to properly handle model bias 6

Improved Cloud/precipitation analysis • CRTM cloud and precipitation capable • Including cloud/precipitation physics in analysis system • Beginning work to include cloudy microwave radiances in analysis • IR radiances being included through partners • Inter-variable covariances will be handled by ensemble covariances • Implications for bias correction and quality control 8

Additional observations • Preparing for use of NPP/JPSS and GOES-R instruments • Working on including more IASI and AIRS moisture channels • Attempting to include SSM/IS F 16 -18. • Testing AMSRE data • Improving quality control for AMSU-A, AMSU-B and MHS resulting in use of more data • Adding additional GPS-RO platforms • Working on inclusion of SEVIRI radiances • Inclusion of aircraft based doppler wind estimates 9

Improved Balance • Pursuing multiple paths to improve balance in initial state – Post-processing (after relocation and/or after analysis) – Inclusion of improved balance within analysis – Inclusion of TLNMI in regional system – Balance through Hybrid assimilation system 10

Improved use of current observations • Upgrading CRTM – surface emissivity, non-LTE, etc. • Including additional effects in Radiative transfer calculations, FOV size and shape, Clouds, trace gases, aerosols, etc. • Pursuing more accurate observational data base dictionary • Creating revised observational data base to allow improved quality control based on recent history. • Use of GPS-RO bending angles rather than refractivities 11

Hurricane Data Assimilation • Inclusion of aircraft Doppler winds • Develop storm relocation code (w/ MIT) to improve relocation of hurricane. External to analysis code and requiring separate balancing code. • Include diabatic effects into GSI Balance and further enhancements to balance equation • Develop capability to incorporate microwave cloudy radiances • Develop Hurricane hybrid ensemble/GSI capability • Incorporate storm relocation code within GSI (to allow direct use of balance code and account for background error properly) 12

Challenges • Collaborations – – Collaborations allow more straightforward inclusion of new developments Must ensure that collaborations don’t slow us down Code management becomes much more difficult External groups must conform to Collaboration requirements. • Multiple funding agencies, multiple bosses, different goals • Data – Processing of new data still slow and difficult because many new formats – Data handling system and quality control • • Assimilation of clouds and precipitation Use of satellite radiances in regional systems (bias correction) Computational resources Slow implementation process 13

Global Implementation Q 1 FY 10 • Major components – – – – Inclusion of NOAA-19 AMSU-A/B, HIRS Inclusion of RARS 1 b data Inclusion of NOAA-18 SBUV/2 and OMI Assimilate EUMETSAT-9 wind vectors Use of Hurricane central pressure estimate Improved use of GPS RO observations Include Dry mass constraint Inclusion of 2 additional GPS RO satellites in Q 2 FY 10 14

Prepare for FY 11 implementations • All implementations should be snapshots on trunk – short branches created when problem found • Many enhancements completed (extracted from trac pages) – – – Improved OMI QC Removal of redundant SBUV/2 total ozone Retune SBUV/2 ozone ob errors Relax AMSU-A Channel 5 QC New version of CRTM 2. 0. 2 Inclusion of Field of View Size/Shape/Power for Radiative transfer Remove down weighting of collocated radiances Limit moisture to be >= 1. e-10 in each outer iteration and at end of analysis Include ASCAT winds Improve location of Buoys in vertical (move from 20 to 10 m) Improved GSI code with optimization and additional options 15

Prepare for FY 11 implementations • Many enhancements completed (cont. ) – Add Radar and Lightning data IO and cloud analysis for RR – GSI Bundle to generalize code/analysis variables – Hybrid use of ensembles for global (regional soon) system (Dual resolution) – Add analysis tools to repository – Upgrade background error covariance files – Upgrade quality control of MHS/AMSU-B water vapor channels – Add reading of Aerosol data – Regression tests for code 16

Prepare for FY 11 implementations • Additional Enhancements underway (extracted from trac) – – – – – 4 D-Var system using perturbation model Common Makefile for multiple platforms Add more tools into repository Update assimilation of TMI rainrate using improved convection parameterization Inclusion of moist physics in TLNMI Include noise reduction in processing of SSM/IS data Include F 16 -F 18 SSM/IS data in GSI analysis Updated radiance bias correction (preconditioning, updating channels not used and incorporate angle dependent part inside GSI) Include external estimates of trace gases in CRTM 17

Prepare for FY 11 implementations • Additional Enhancements underway (Cont. ) – Inclusion of addition (higher peaking) IASI/AIRS moisture channels – Inclusion of AMSRE data – New localization options for hybrid En. KF/GSI – Use of sensitivity tool – SST analysis within GSI – GPS RO bending angle and use of compressibility factors – Regional strong constraint – Inclusion of aerosols in CRTM – General code clean-up and optimization 18