Minimum temperature forecast at Manali, India

• Published in: Current science (Bangalore) Vol. 88; no. 6; pp. 927 - 934
• Main Authors: Dimri, A. P., Mohanty, U. C., Rathore, L. S.
• Format: Journal Article
• Language: English
• Published: Current Science Association 25.03.2005
• Subjects: Analytical forecasting; Correlation coefficients; Datasets; Employment forecasting; Forecasting models; Modeling; SPECIAL SECTION: MOUNTAIN WEATHER FORECASTING; Statistical variance; Statistical weather forecasting; Temperature; Weather forecasting
• ISSN: 0011-3891

Northern India is comprised of complex Himalayan mountain ranges having different altitude and orientation. Knowledge of minimum temperature in this region during winter months is very useful for assessing human comfort and natural hazards. In the present study, Perfect Prognostic Method (PPM) is used for forecasting minimum surface temperature at one of the stations, Manali, in Pir Panjal range of Himalayas. Firstly, a statistical dynamical model is developed for assessing next day's temperature category, i.e. ≤ 0°C or >0°C. Once the category is known, then temperature forecast model is developed for that category. Statistical dynamical models are developed for winter season, December, January, February and March (DJFM) using multivariate regression analysis. Model is developed with data of DJFM for 12 years (1984–96) and tested with data of DJFM for the year 1996–97. Analysis data from National Center for Environmental Prediction (NCEP), US, station surface and upper air data of three stations of India Meteorological Department (IMD), India and surface data at Manali are used. Four experiments are carried out with four different sets of predictors to evaluate performance of the models with independent data sets. They are: (i) NCEP reanalysis data, (ii) operational analyses from the National Center for Medium Range Weather Forecasting (NCMRWF) in India, (iii) day 1 forecast with a T80 global spectral model at NCMRWF and (iv) forecasts from the regional mesoscale model MM5 day 1 forecast. A comparison of skill is drawn among these four set of experiments. It is found that best prediction for temperature category is made with an accuracy of 71.2% with MM5 day 1 forecast as predictors in temperature category forecast model. Further, temperature forecast model for ≤0°C category selects only station data and shows skill of 62.1% with independent data, whereas, model for >0°C category selected predictor from numerical analysis also. Here MM5 day 1 forecast makes best prediction with 90.0% skill.