Report on the First SPARC Data Assimilation Workshop
University of Maryland Baltimore County, USA, 10-12 June, 2002

The workshop began by a welcome given by T. Low from GEST. A. O'Neill, SPARC co-chair, presented his view on the rationale and priorities of the SPARC data assimilation project. He started by describing existing SPARC projects. He advocated a formation of a SPARC DA working group with a purpose of coordinating and promoting data assimilation relevant to SPARC. He suggested a list of priorities for the working group, and encouraged workshop participants to attend the Third SPARC General Assembly in July 2004.

An overview of ESA's missions collecting constituent data was given by C. Zehner. He described data assimilation as a key tool for exploitation of ESA's atmospheric measurements. Data assimilation is one of the techniques that will be used in calibration and validation of Envisat products.

G. Manney presented an example of inter-comparisons of assimilated fields that this SPARC initiative is hoping to extend to different regions and geophysical fields. She presented inter-comparisons of temperature fields in polar regions. She focused on a quantitative study of 6 data sets for cold Arctic winters in 1995-1996 and 1999-2000.

R. Swinbank gave an overview of the Stratospheric Data Assimilation at the UK Met Office. He described the Analysis Correction system, which produces a realistic quasi-biennial oscillation (QBO) in the analysed winds. He continued with a description of a newer 3D-Var system and a comparison of assimilated fields from the two systems. Differences of temperature sonde observations and model forecasts in the troposphere are smaller for the 3D-Var system. Planned development includes changes in the dynamical core, background error statistics, and a gravity wave drag scheme. He concluded that the stratospheric DA products are invaluable, and that stratospheric DA is becoming a part of the numerical weather prediction mainstream.

An overview of the meteorological analyses at the NASA/Goddard Data Assimilation Office (DAO) was given by S. Pawson. He showed how the scrutiny of the system products during mission support such as SAGE III Ozone Loss and Validation Experiment (SOLVE) leads to a better understanding of the system. A new DAO system GEOS-4 has reduced noise in meteorological fields that has a massive impact on trace gas transport, including an older age of air. The new system will be used for a reanalysis over the Upper Atmosphere Research Satellite (UARS) period for years 1991-1995. Shuhua Li showed that including travelling gravity waves has a large impact on DAO's GEOS-4 forecast and analysis fields, in the upper stratosphere.

S. Edouard presented the impact of the recent changes in the Global Environmental Multiscale (GEM) model: assimilation of radiances, a new quality control scheme, a new gravity wave drag scheme, and assimilation of cloud motion wind vectors. She presented the plans to change to a new hybrid vertical coordinate, assimilate new data types, and for development of a 4D-Var assimilation system.

S. Polavarapu described the development of a coupled meteorological and chemical assimilation system in collaboration of the Meteorological Service of Canada and Canadian Universities. They adapted the Canadian Meteorological Centre's operational 3D-Var scheme and coupled it to the Canadian Middle Atmosphere Model (CMAM). The system currently shows a reasonable comparison against UK Met Office analyses, and many improvements are planned.

A method for construction of proxy ozone fields based on the correlation of ozone measurements from occultation instruments and of smoothed potential vorticity (PV) fields on isentropic surfaces was presented by C. Randall. The proxy ozone based on Polar Ozone and Aerosol Measurement (POAM) data alone compares well with ozone sondes in mid to high latitudes. The addition of Stratospheric Aerosol and Gas Experiment (SAGE) and Halogen Occultation Experiment (HALOE) produces a better quality of the proxy ozone at low latitudes. This method can be used for construction of water vapour or aerosol surface area proxy fields. The method is limited by the noise in the PV field and during the period of the polar vortex break-up.

D. Allen presented a way of constructing another ozone proxy using a tracer equivalent latitude (TrEL). A synthetic tracer is advected with diffusion by non-divergent UKMO winds on isentropic surfaces. The tracer contours are used to define an equivalent latitude. Equivalent latitude/potential temperature bins are used to define an ozone proxy. The proxy is further constrained using Total Ozone Mapping Spectrometer (TOMS) level 2 data. This proxy was used to investigate an extreme ozone mini-hole event in November 1999.

S. Kondragunta gave an overview of the operational and reprocessed calibration of the Solar Backscatter Ultra-Violet/2 (SBUV/2) instrument ozone data. She pointed out the changes in the operational calibration and the retrieval algorithm and their effects on the ozone product. She also presented a new method for producing global total column ozone maps by combining the SBUV/2 product in the middle and upper stratosphere with a TIROS Operational Vertical Sounder (TOVS) ozone product between 400 and 30 hPa.

J. Hornstein talked about using 3-D global ozone fields for testing of data assimilation and satellite retrieval algorithms. Global proxy ozone fields can be used to provide simulated data with a representative variability. They can also be used to construct a climatology and a priori error covariances matrices for satellite retrievals.

The ozone assimilation session started with the presentation given by A. Dethof about the ozone assimilation at the European Centre for Medium-Range Weather Forecasts (ECMWF). They have included TOMS and SBUV/2 ozone assimilation in the ERA 40 reanalysis and Global Ozone Monitoring Experiment (GOME) and SBUV/2 ozone assimilation in the operational system. They are planning to include a radiative feedback of ozone fields and a multivariate ozone-PV assimilation.

I. Stajner gave an overview of the ozone assimilation at the DAO. She demonstrated a positive impact of including SBUV/2 ozone data from the 63-126 hPa layer, and of modelling TOMS errors as spatially correlated on the quality of the lower stratospheric assimilated ozone. Both the inclusion of a parameterised chemistry model and the better quality of the GEOS-4 assimilated winds were shown to improve the assimilated ozone. She presented examples of successful monitoring of satellite instrument errors, and discussed plans for monitoring and assimilation of ozone data from new instruments.

D. Jackson described ozone data assimilation research at the UK Met Office. The goal is to include ozone as an analysis variable in the operational 3D-Var system by assimilating SBUV/2 ozone retrievals and TOVS channel-9 radiances. The ultimate goal is to show a positive impact on forecast scores of the operational system. Observing system simulation experiments for several future instruments (SWIFT, IASI, AMSU, GOME2) are planned.

H. Eskes showed results from GOME ozone data assimilation and ozone forecasting at the KNMI. In the system, that is driven by ECMWF winds, forecast error variances are evolved, and the correlations are fixed in time. Ozone forecasts are produced and have anomaly correlations above 60% for 6 days in the extratropics, and for 2 days in the tropics. Their plans include SCIAMACHY validation and assimilation, assimilation of nadir radiances, and estimation of tropospheric ozone from GOME or Ozone Monitoring Instrument (OMI) assimilation.

T. Sasaki presented the progress and plans for ozone assimilation within a coupled general circulation model (GCM) and chemistry and transport model (CTM) at the Meteorological Research Institute (MRI) and the Japan Meteorological Agency (JMA). This system uses TOMS total column ozone data through a nudging technique.

An approach to SBUV/2 radiance assimilation was presented by P.-K. Bhartia. He advocated the use of a linearised forward model around the retrieved ozone profile in the radiance assimilation. He also asked the ozone data assimilation egy for Envisat level 2 chemical data validation and assimilation at the Data Assimilation Research Centre (DARC). He showed results from GOME and/or UARS Microwave Limb Sounder (MLS) ozone assimilation. The assimilation results are better when data from both instruments are used than when either instrument is used alone.

H. Elbern presented observability tests with a variational data assimilation system. He showed that in an identical twin experiment evolution of most species in a box model can be recovered from the six species that will be provided operationally from MIPAS observations. This feasibility study indicates that a 4D-Var approach will provide estimates of unobserved species by assimilation of Envisat routine data.

D. Fonteyn showed the progress made towards operational chemical data assimilation BASCOE. It is based on a 4D-Var scheme for a 3-D Eulerian stratospheric CTM with heterogeneous chemistry and polar stratospheric cloud microphysics. Systematic differences were found in the assimilation of CRISTA chemical data. Validated data from three Envisat instruments will be assimilated.

J.-F. Lamarque presented the recent work at NCAR on the tropospheric chemical data assimilation. He showed results on ozone and carbon monoxide assimilation in the MOZART-2 model with a tropospheric chemistry. Satellite data from TOMS, MLS and Measurements of Pollution in the Troposphere (MOPPIT) instruments were used in a sub-optimal Kalman filter. The assimilated fields compare well with independent observations.

D. Lary presented his work on the chemical data assimilation in flow-tracking coordinates. The UARS chemical observations were assimilated using the Kalman filter into a box chemical model. Inconsistencies between measurements of different chemical species were found. They need to be addressed through a quality control of observations and by devising ways of conserving the budgets of chemical species in the assimilation.

The workshop concluded by a discussion of goals for an emerging SPARC DA working group. I. Stajner was appointed as convener. Yearly workshops and data assimilation sessions at scientific meetings are planned. Three near-term tasks were agreed on. G. Manney will coordinate collection of information about availability and quality of stratospheric meteorological and chemical data sets. After the data are available, process focused assessments of their quality will follow. For example, they can include a representation of polar processes, equatorial dynamics such as QBO, tape recorder in moisture, the subtropical barrier, or ozone mini-holes. J.-F. Lamarque will coordinate collection, validation and intercomparison of chemical assimilated data sets. The winter of 1999/2000 was suggested as one of the periods for an intercomparison and validation against independent measurements. D. Fonteyn will collect information on CTM and GCM assimilation systems that are available for intercomparisons and coupling.

More information and electronic presentations are available from:
http://polar.gsfc.nasa.gov/sci_research/cooperative_ventures/sparc/report.php