Improved Track and Intensity Predictions Using Cyclone Bogusing and Regional Assimilation - Monitoring and Prediction of Tropical Cyclones in the Indian Ocean and Climate Change - page 247

Both systems crossed the coast of Tamil Nadu, with cyclone Jal crossing the
coast north of Chennai on 7 th Nov 2010 and cyclone Thane crossing the coast
south of Chennai on 30 th Dec 2011 near Puducheri. Both these cases attained
peak intensity in the DWR range.
Radar observations are available in 250-500 km range from antenna at a
resolution of 250-500 m. Azimuth angle found to vary from 1 to 360 o at a
resolution of 1 o and elevation angle between 0.2 and 21 o . Radars employ PPI
Volume scan with two modes of scans having one long-range scan with two or
three lower elevation angles and a short-range scan for ten elevations. These
observations available in radar centric spherical coordinate system is geo-located
as part of pre-processing. The detailed description of DWR data and pre-
processing is explained in John et al. (2011). Reflectivity values lower than 10
dBZ are ignored. The observations coming in the altitude range 0.2 to 18 km
are used to avoid ground clutter and to limit observations up to tropopause
level. Figure 4 shows the typical radar reflectivity from the cyclone bands for
the two cases. For each cyclone case a control run (CTL; without DWR data),
an experiment run (EXP; with DWR data) and another experiment run (EXPRV;
with DWR radial wind data alone) are carried out. All model runs make use of
the WRF-VAR assimilation system for preparation of the high resolution
analysis (9 km). The CTL makes use of only the conventional data in the
assimilation and the EXP makes use of the conventional data in addition to all
DWR data (Radial wind + Reflectivity) and EXPRV with radial wind alone.
The model produces forecast up to 72 hours. The forecast tracks for the three
experiments for Jal and Thane cases are shown in Fig. 5. The forecast track in
the EXP and EXPRV clearly shows an improvement over the CTL run. This is
also reflected in the forecast track errors in the initial hours as shown in Fig. 6.
Results obtained from the cases of cyclone Jal and Thane show improvements
in the EXP and EXPRV run suggesting important role of DWR data in the
assimilation. Forecast errors at some of the lead times suggest that assimilation
of radial wind alone produces lower errors. However, this needs a detailed
investigation involving large number of cases for a conclusive remark.
Fig. 4: Radar reflectivity bands in cyclones (a) Jal, 7 th Nov 2010 and
(b) Thane, 29 th Dec 2011.