Mesoscale Modelling for Tropical Cyclone Forecasting over the North Indian Ocean - Monitoring and Prediction of Tropical Cyclones in the Indian Ocean and Climate Change - page 272

Fig. 1: Mean direct position error (km) from 97 cases, in 12-hr interval up to
72-hour forecast for the TCs over the NIO.
3.2 Impact of Satellite-derived Sea Surface
Winds on TC Forecast
An attempt is made to assess the impact of satellite-derived sea surface winds
obtained from Quick scatterometer (QSCAT) and Special Sensor Microwave/
Imager (SSM/I) on initialization and simulation of TCs over NIO. For this
purpose, six TCs viz., 'Nargis (five cases)', 'Gonu (four cases)', 'Sidr (four
cases)', 'KhaiMuk (three cases)', 'Aila (four cases)', and 'Laila (four cases)'
are simulated with varying initial time constituting 24 cases. Two sets of
numerical experiments, with and without satellite-derived wind data assimilation
are conducted: CNTL, in which NCEP FiNaL (FNL) analyses are used as initial
and boundary conditions and 3DVAR, in which QSCAT and SSMI winds are
assimilated using 3-dimensional variational data assimilation (3DVAR)
technique to improve the model initial condition.
The assimilation of above mentioned data improved the initial vortex
structure and the initial vortex position of TCs significantly in 21 cases out of
total 23 cases as shown in Fig. 2 (no observation is available for case 17). The
mean initial vortex position error is reduced from 72 km (FNL analyses) to 48
km in 3DVAR analyses (by 34%). The improvement in initial vortex structure
can be obtained from Osuri et al. (2012b) after the assimilation of satellite-
derived winds. 3DVAR experiment improves intensity in most of the cases
significantly. Out of 24 cases, 3DVAR shows better intensity predictions in 18
cases (in terms of minimum sea level pressure, MSLP) and in 19 cases (in
terms of 10 m wind). There is an improvement of 10% in MSLP prediction and
21% in 10 m wind prediction with 3DVAR experiment.
From Fig. 3, the vertical structure of wind distribution is better simulated
with 3DVAR as the depth of stronger wind (>55 m s -1 ) extends up to 700 hPa
level. In both CNTL and 3DVAR experiments, the wind distribution is not