A new technique has been developed to assimilate satellite radiances or brightness temperatures directly into a numerical forecast model. This new procedure captures some of the best features from both retrieved profile as well as direct radiance assimilation. The advantage of the new scheme is that an adjoint is not required and the procedure is under the control of the modeler, provided satellite observations are available.

So far, it has been tested using the 6.7-micrometer water vapor band on the Geostationary Operational Environmental Satellite (GOES) Imager (channel 3), and the 6.5-micrometer water vapor band on the GOES Sounder (channel 12). The upper tropospheric relative humidity is adjusted in the vertical column using weights determined by the forward radiative model. The adjustment procedure is combined with an optimization scheme so that the moisture field is modified until it reproduces the remotely observed brightness temperature field. The modifications (done in physical, not radiance, space) are retained separately, quality controlled, and adjusted for biases and observational errors.

The impact of the assimilated moisture data is evaluated in forecast experiments from an arbitrary 10 to 15 day period from September 2000. Forecasts made from the initial conditions, following the modifications described above, are contrasted to forecasts from the original conditions, all compared to later verifying conditions. Results from these experiments are discussed.