HRD seminar – Prof. U.C. Mohanty, India Institute of Technology (IIT), Bhubaneswar, Odisha, India – 16 June 2015

Prof. Mohanty presented a seminar on “Prediction of tropical cyclones over Bay of Bengal using 2013 version of HWRF system”.


More than 80-90% of the deaths due to Tropical Cyclones (TCs) are caused by fresh water flooding and storm surge; and hence realistic TC intensity, size and structure predictions are important at landfall for disaster management. The 2013 version of HWRF system is used with multiple moving nests spanning at (cloud resolving) resolution down to 3 km to predict tropical cyclones of 2013-14 over the Bay of Bengal (BoB).

Application of two versions of the HWRF system for the Bay of Bengal cyclones demonstrated that cloud resolving version of HWRF (27/9/3 km, 3 domains) is superior than HWRF run at reduced resolution around the vortex (27/9 km, 2 domains) for TC prediction (track, intensity and rainfall) that was used until recently over the Indian seas. Structure prediction is also improved with 27/9/3 km HWRF.  Real time application of 27/9/3 HWRF system showed potential to provide better intensity and structure forecast guidance for TCs over NIO. Further experiments suggest that vortex initialization and relocation in cyclic mode has a positive impact on track and intensity over cold start, though not significant (based on 4 cyclones, need for a number of cases).

HWRF provided realistic TC predictions consistently for track, intensity, rainfall and structure. The rainfall amount and structure is improved significantly, particularly, in case of Phailin, the model showed peak rainfalls over Northern parts of Odisha, as observed. HWRF is able to predict the rapid intensification of Phailin a day before to its actual occurrence. In case of Lehar, the model predicted rapid weakening and indicated the depletion of moisture associated with the weakening over the Bay of Bengal. When compared to ARW model predictions, HWRF is superior in predicting intensity and structure, even in 4 days or longer forecast length, while, track prediction is comparable. 

The improved longer range forecast of TC track, intensity and structure helped to increase the lead time of storm surge prediction over the Bay of Bengal basin. A 3-4 day forecast of maximum surge is reasonably good in terms of location and quantity. In case of Hud-Hud, a 4-day forecast of peak surge is about 1.4 m and a day forecast is about 1.7 m which provides better guidance when compared to the observed surge of 1.4 m at Vishakhapatnam.

A video recording of the presentation is available on the anonymous ftp site:

Screen Shot 2015-06-16 at 5.47.38 PM

This entry was posted in Data Assimilation, HFIP-Hurricane Forecast Improvement Project, Modeling and Prediction, Presentations and tagged , . Bookmark the permalink.