A primitive equation spectral model has been successfully integrated for five days starting with the initial data of 26 February 1982. The geopotential heights and wind strengths are well predicted up to day 3. The 24-hour accumulated precipitation of the model reasonably agrees only up to two days, with the observed rainfall under the influence of western disturbance that prevailed over Afghanistan, Pakistan and north India till 3 March 1982. For one day global integration of the model, the CPU time requirement in Cyber 170/730 is about one hour.

A number of sensitivity experiments have been conducted to investigate the influence of using synthetic data on cyclone forecasts by a global spectral model. Some well known vortices have been used and the generated wind and pressure profiles are compared. It is found that the Rankine vortex and Holland’s vortex show the best representation of cyclonic circulation. Hence these two vortices are used in the sensitivity studies to simulate two cyclones, one of May 1979 and the other of August 1979. For this purpose the FGGE level-III b data set, produced at ECM WF, UK is used. Synthetic temperature and humidity data are also introduced to make the cyclones more realistic.

With the use of Holland’s vortex the system is found to move faster than with the Rankine vortex. Also, the tracks of the cyclones simulated with Rankine vortex are found to be on the left side of the observed track while that of Holland’s vortex is on the right side of the observed track. However, substantial filling up of the systems are noticed with introduction of diabatic initialization of the mass and velocity fields and the forecasts of both the vortices behave differently. It is suggested that proper selection of synthetic vortex, initialization scheme and resolution of the model are very important for better forecast of cyclones.

The stability analyses of the stream function at the upper atmosphere have been conducted using a global barotropic spectral model with a view to examine the seasonal characteristics. The growing eigen modes are classified into three groups with periods in the range of 3–8, 11–18 and 20–50 days. This study indicates that the growth and movement of all the three types of modes are adequately sustained by the asymmetric basic flow. All the modes grow faster in summer than in winter. The meridional shear of the basic flow is the most important source of growth of the perturbations. In the absence of the meridional wind shear, the eigen modes grow slowly, in which case, the quasi-nonlinear triad interaction between the waves is identified to contribute significantly to the growth of the modes. The robustness of the eigen modes is also examined in this study using the barotropic model at different horizontal resolutions in the triangular truncation scheme.

To investigate the temperature changes at 100 hPa over Indian region from Arabian Sea (AS) to Bay of Bengal (BOB), analysis is performed using Atmospheric Infra Red Sounder (AIRS) temperature and outgoing long-wave radiation (OLR) data of 9 years (2003–2011). Fine-scale temperature variations have been studied and shown for summer (March–April–May, MAM), summer monsoon (June–July–August–September, JJAS) and winter (November–December–January–February, NDJF) months. Similarities and differences in the latitudinal and longitudinal variation of temperature and the possible causes have been examined. During MAM and NDJF, the temperature increases latitudinally by ∼2–3 K and ∼4–5 K from 3.5° to 20.5°N, respectively. However, the temperature decreases by ∼2.0–2.5 K during JJAS. A similar contrasting behaviour is observed in latitudinal temperature gradient. For MAM and NDJF, the gradient decreases from ∼0.18 to ∼0.14 K/deg and ∼0.25 to ∼0.18 K/deg, respectively, as we move longitudinally from 60° to 90°E; however, for JJAS, it increases from ∼0.10 to ∼0.14 K/deg over the same longitudes. It is found that latitudinal temperature gradient for NDJF is larger by about a factor of 1.5. Analysis suggests latitudinal change in temperature occurs due to low OLR (proxy of convection) and its northward progression during summer monsoon. Correlation coefficient (𝑅_xy) between OLR and temperature is computed latitudinally (3.5° to 20.5°N) at different longitudes and during JJAS (monsoon months), 𝑅_xy is negative (∼−0.73) over 60° and 70°E longitudes, but it turns positive (∼0.92) over 80° and 90°E longitudes (which is convectively active region), suggesting a close association between low temperature and low OLR. Land–sea contrast is also observed in temperature at 100 hPa with a slight increase (∼0.5 K) from sea to land.