Studies on the Impacts of 3D-VAR Assimilation of Satellite Observations on the Simulation of Monsoon Depressions over India - Data Assimilation for Atmospheric, Oceanic and Hydrologic Applications - page 636

In the 4D-VAR method it is necessary to estimate the time evolution of the
perturbation using a linear model and to calculate the adjoint of the above linear
model. Furthermore the 4D-VAR method is considered to be computationally
intensive since this involves forward integration of the model and backward
integration of the adjoint model, over many iterations to obtain the minimization
of cost function. In this chapter, results of our studies on the impacts of 3D-VAR
assimilation of satellite observations on the simulation of monsoon depressions over
India are provided. Although 4D-VAR is superior, our study has been restricted
to 3D-VAR for the following reasons, (1) we felt that it is more appropriate and
prudent to take up 3D-VAR studies first rather than go for 4D-VAR, and (2) the
computational costs associated with the study of 4D-VAR.

26.2.2 Assimilation of Satellite Observation

The basic problem in numerical weather prediction (NWP) is that the observa-
tions are at least two orders smaller than the number of degrees of freedom
of the model. Most meteorological systems form over the sea which is a data
sparse region. Satellites provide an excellent platform to obtain observations
of the atmosphere over the sea. Unlike the conventional observations such as
radiosondes/rawinsondes, the quantities measured by satellites do not directly relate
to the atmospheric quantities such as temperature, humidity, wind direction, wind
speed, etc. What essentially satellite measures are the radiation that reaches the top
of the atmosphere at given frequencies in the case of passive radiometers and the
back scattered radiation emitted by a surface (say a sea surface) in the case of active
scatterometer.
The most common of satellite observations to be assimilated in a “data assim-
ilation” methodology is the satellite derived vertical air temperature and humidity
profiles. Other important meteorological observations obtained from satellite are
the sea surface temperature (SST), surface wind speed and wind direction over
the sea, rainfall rate, total precipitable water, cloud motion wind vector (CMV)
at different levels of the atmosphere. While the satellite derived temperature and
humidity profiles can be directly assimilated in a NWP model, the variational
method allows for the direct assimilation of satellite radiance observation. For direct
satellite radiance assimilation, the observational operator H incorporates a radiative
transfer model that maps the atmospheric profile to radiance space. The above
procedure of directly assimilating satellite radiance is better since radiance errors
are more easily characterized than the retrieval errors. The following subsections
provide brief information of the various satellite sensors (QuikSCAT, SSMI,
ATOVS and MODIS) which are normally utilized in 3D-VAR assimilation impact
studies.