A numerical study of sea breeze circulation observed at a tropical site Kalpakkam on the east coast of India, under different synoptic flow situations

A simulation study of the sea breeze circulation and thermal internal boundary layer (TIBL) characteristics has been carried out at the tropical site Kalpakkam on the east coast of India, for operational atmospheric dispersion prediction. The community based PSU/NCAR MM5 Meso-scale meteorological model is used for the study. Three cases on typical days in summer (24 May 2003), southwest (SW) monsoon (1 July 2001) winter season (2 February 2003) with different large-scale flow pattern are studied. The MM5 model is used with 3 nested domains with horizontal grid resolutions 18 km, 6 km and 2 km and 26 vertical levels. The model is integrated for 24 hours in the above cases with initial and boundary conditions taken from NCEP-FNL analyses data. Observations of 10 meteorological stations and coastal boundary layer experiments conducted at Kalpakkam are used for comparison and validation of the simulation. The characteristics of simulated sea breeze and TIBL at Kalpakkam are seen to vary in the above cases according to the prevailing large-scale winds and surface fluxes. The sea breeze circulation is seen to develop early with larger strength and inland propagation in the summer case under the influence of moderate synoptic wind and strong heating conditions than in the SW monsoon and winter cases. The horizontal and vertical extents of TIBL are found to be larger in the summer case than in other cases. Although model parameters agree in general with observations, all the fine features are not clearly captured and some slowness in model sea breeze development is also seen. The results indicate the need to improve i) the initial conditions by assimilation of available surface/upper air observations to reduce model bias and ii) surface net radiation parameterisation. The model could predict the essential features of the local circulation and further improvement is expected with better initial condition data and incorporation of more realistic surface data.