A quick regional OSSE impact study on Geostationary Hyperspectral Infrared Sounder for Hurricane Forecasts

Future geostationary (Geo) hyperspectral InfraRed (IR) sounders have finer spatial, spectral, and temporal resolutions compared to the existing Geostationary Operational Environmental Satellite (GOES) sounders, providing much improved resolving power of atmospheric thermodynamic information. When quantitatively assessing the value added impact from such instruments over the current sounding systems onboard the Low Earth Orbit (Leo) satellites, the real question is what is the optimal impact using the current assimilation/forecast system. More specifically, will assimilating more observations from future Geo IR sounders with the current assimilation/forecast system yield improved forecast as expected? And if so, how to assimilate the high temporal resolution Geo sounding information and what is the impact on forecasts? Taking tropical cyclone (TC) forecasting for an example, this study tries to answer these questions through a quick regional Observing System Simulation Experiments (OSSE) study. Synthetic observations are simulated from the sample data obtained from the European Center for Medium-Range Weather Forecast (ECMWF) T1279 nature run (NR) for Hurricane Sandy (2012), including the radiosonde observations (representing the conventional observations), the Leo Atmospheric InfraRed Sounder (AIRS), and the Geo AIRS. Assimilation experiments were carried out using WRF 3.6.1 and GSI 3.3 to study the impact on Sandy track forecast. Results from the study show that a) it is critical to assign an appropriate observational covariance matrix (R matrix) in order to show larger positive impacts from Geo AIRS over Leo AIRS; b) cycling experiments show improved positive impacts over no cycling, but hourly cycling does not show further improvement on forecast over 6 or 3 hour cycling, and c) with thinning between 120 ~ 240 km, the impacts of different cycling experiments have the following order: hourly > 3-hourly > 6-hourly > none cycling. These experiments indicate that while more observations is expected to improve forecast, with the current assimilation system configurations too many observations may degrade the forecast performance. There may exist a tradeoff between the number of observations and influence of single observation in order to maintain the positive impact. The objective of this study is to find the optimal point where the advantage of observations is maximized within existing assimilation/forecast system. This will be done by adjusting the relative weight between the background and the observations. While keeping the R matrix and background covariance matrix (B matrix) unchanged, a single regulation parameter will be multiplied to the background term in the cost function, with a large value indicating more weight on the background. For each cycling experiment, an optimal value of the regulation parameter will be determined to yield an optimal forecast. These forecasts will then be used to determine the value-added impact of future high temporal resolution Geo IR sounders over the current existing sounding systems on hurricane forecasts.