Assimilation of Doppler Weather Radar Data in WRF Model for Numerical Simulation of Structure of Cyclone Aila (2009) of the Bay of Bengal at the Time of Landfall - Monitoring and Prediction of Tropical Cyclones in the Indian Ocean and Climate Change - page 307

Fig. 2: The vertical distribution of the number of radar observation data
around 0900 UTC 25 May 2009.
resolution of 9 km with 125 × 125 grid points in X and Y directions. This
domain covers an area of 1125 × 1125 km with Kolkata located at the centre.
The vertical grid stretched from surface to model top, which is located about
20 km height at a vertical resolution of 500 m. NCEP global forecast system
(GFS) data at 1° × 1° resolution are used to provide initial and boundary
conditions.
3.1 ARPS Data Analysis System (ADAS)
Brewster (1996) developed ADAS for assimilation of data such as radar, satellite
etc. in numerical model. ADAS uses a successive correction scheme, known
as the Bratseth method (Bratseth, 1986), and includes radial velocity adjustment.
Like Optimum Interpolation (OI) method, here also correlations among the
data must be specified. Typically the correlations are a function of spatial
distance. In ADAS, the correlation, U, as a function of horizontal spatial
separation is modelled as a Gaussian:
U ij = exp (- | r ij | 2 / R 2 )
where r ij is the displacement between two locations and R is the correlation
distance factor. The total correlation is also affected by separation in the vertical.
ADAS allows the vertical correlation to be specified as a function of height
separation, z
U'z = exp (-' Z 2 / R z 2 )