Extended Abstract (224K)P3.10
Assimilation of retrieved water vapor profiles into the Navy Operational Global Atmospheric Prediction System (NOGAPS)
Clay B. Blankenship, NRL, Monterey, CA; and N. L. Baker
Water vapor profiles retrieved from Advanced Microwave Sounding Unit-B (AMSU-B) observations are assimilated using the NRL Atmospheric Variational Data Assimilation System (NAVDAS) with the Navy Operational Global Atmospheric Prediction System (NOGAPS) forecast model. This paper will explain the methods used to retrieve water vapor profiles and assimilate those profiles and will assess the impact of using these data on the analysis and forecast accuracy.
The water vapor retrievals use observations at frequencies near the 183.31 GHz water vapor absorption line from the AMSU-B microwave radiometer on the NOAA-16 polar orbiter. The retrieval algorithm is a physical optimal estimation inversion of the observed brightness temperatures constrained by the NOGAPS background (6-hour forecast); this is equivalent to a one-dimensional variational assimilation (1DVAR) of radiances at each observation point.
Assimilation of retrievals rather than radiances is used because NAVDAS cannot currently handle the nonlinear water vapor Jacobian (the matrix of partial derivatives of brightness temperatures with respect to layer specific humidities). This approach lets the 1DVAR retrieval handle the nonlinearities inherent in a water vapor profile retrieval so all observations input to the 3DVAR system (NAVDAS) are linearly related to model quantities.
Water vapor retrievals are compared with radiosondes for validation. Innovations (retrieval minus background) generally have the correct sign as indicated by the radiosonde. Anomaly correlation and rms forecast error, both verified against self-analysis, will be used to assess the impact of assimilating retrieved water vapor on forecasts. In addition, forward modeled TB's will be compared with observations from instruments not used by NAVDAS in order to validate the model moisture fields.
Poster Session 3, Data Assimilation
Tuesday, 11 February 2003, 3:30 PM-5:30 PM
Previous paper Next paper