Statistical Assimilation of Satellite Data: Method, Algorithms, Examples

Abstract

There were several reasons that caused an intensive development of the 4-D methods for the assimilation of remote sensing data in meteorology. One reason is that the satellite data are continuous in time, in contrast to conventional network observations carried out at prescribed standard terms. But the most part of remote sensing data arrives just between such terms. Another reason is that the remotely sensed meteorological parameters have to be derived from the solution of the ill-posed inverse problems. This last occasion is of special attention, because instead of conventional direct and, therefore, point-wise parameter measurements, in the case of remote sensing we have to deal with some functional of spatial field for this parameter. For example, in the case of atmospheric thermal remote sensing we cannot to retrieve temperature magnitudes at some vertical levels or at some spatial points, but rather averaged values related to some not fully certain weight functions. The third reason is that, actually, the operator of the inverse remote sensing problem does not maintain some constant magnitudes in spatial and temporal coordinates, but, really, it is subjected by disturbances originated from permanent changes occurred in atmospheric optical properties.