Hyper-spectral infrared satellite observations are made up of many thousands of channels but contain only a few tens of pieces of independent information. It is therefore desirable to present this information to a data assimilation system in a more efficient form. The type of compression chosen will depend on a number of factors including: - the ability to efficiently forward model the new data type - the ability to define an observation error covariance matrix that is well-conditioned and reflects the true error properties of the observation - the ability to perform accurate quality control - ease of monitoring - robustness of the method against change in instrument characteristics In addition, with the advent of hyper-spectral geostationary imagers, spatial and spectral compression may be necessary.
C1 - Events DA - 2013 LA - eng N2 -Hyper-spectral infrared satellite observations are made up of many thousands of channels but contain only a few tens of pieces of independent information. It is therefore desirable to present this information to a data assimilation system in a more efficient form. The type of compression chosen will depend on a number of factors including: - the ability to efficiently forward model the new data type - the ability to define an observation error covariance matrix that is well-conditioned and reflects the true error properties of the observation - the ability to perform accurate quality control - ease of monitoring - robustness of the method against change in instrument characteristics In addition, with the advent of hyper-spectral geostationary imagers, spatial and spectral compression may be necessary.
PY - 2013 T3 - ECMWF/EUMETSAT NWP-SAF TI - Data compression - Data user perspective ER -