Impact Analysis of LEO Hyperspectral Sensor IFOV size on the next generation NWP model forecast performance
Satellite observations are currently thinned during data assimilation to reduce computational time, achieve better convergence and to reduce spatial correlations between observations. The analysis is usually done at a coarser resolution than the forecast grid spacing. Under clear sky assimilation, cloud detection will be an important quality control that rejects observations from assimilation for a hypersepctral infrared sounder such as Cross-track Infrared Sounder (CrIS). A smaller CrIS field-of-view (FOV )will have a higher probability of being free from cloud contamination. Given a constant analysis grid resolution, the number of CrIS observations available for selection is larger when the sensor FOV is smaller. This will increase the number of CrIS observations entering the data assimilation process and making a larger contribution to the analysis.
Impact of FOV size for CrIS on NWP will be assessed through satellite data assimilation using the NCEP Global Forecast System (GFS). Even though CrIS is currently in orbit, impact assessment will be performed in a simulated environment also known as an Observing System Simulation Experiment (OSSE). This is done because the candidate observation, i.e. CrIS with a smaller FOV is not available in any of the present satellite. To make a fair comparison, CrIS observations at both the current and increased resolution need to be simulated from a known state of the atmosphere or the Nature Run. The control run assimilates CrIS observations at the current resolution, which is about 14km at nadir; and the experiment run assimilates CrIS observations that have a smaller FOV. Forecasts between the two runs will be evaluated. These experiments will be conducted in the presence of all current major observing systems. Prior to carrying out the CrIS experiments, the OSSE system will be calibrated against the real system to verify that the simulated data impact by comparing it to the real data impact.