Variational mesoscale satellite data assimilation and initialization

The problems of mesoscale satellite data assimilation were examined. Assimilation of satellite data to improve the forecasts made by mesoscale forecast models was undertaken. Assimilation of high resolution satellite derived temperature data into a mesoscale model with horizontal resolution of 50 to 60 km is reported. Unlike global assimilation, in which a small portion of the forecast model domain is subject to data insertion at virtually every time step, the mesoscale assimilation virtually all of the forecast model domain is subject to data insertion at one time step. The mesoscale problem lends itself naturally to intermittent data assimilation and the forecast model is reinitialized whenever a new satellite pass covers its domain with data. The satellite data assimilation as an initialization problem are discussed.