Application of Neural Network Technique to improve the location specific forecast of temperature over Delhi from MM5 model

Authors

  • S. K. ROY BHOWMIK
  • SANKAR NATH
  • A. K. MITRA
  • H. R. HATWAR

DOI:

https://doi.org/10.54302/mausam.v60i1.958

Keywords:

Location specific forecast, Neural network, MM5 model

Abstract

India Meteorological Department (IMD) has been using direct model output (2 meters height temperature) of MM5 model as numerical guidance for forecasting maximum and minimum temperature of Delhi in short range time scale (up to 72 hours).  Performance statistics of the direct model outputs of the model for maximum and minimum temperature show that forecast skill of the model is reasonably good, particularly for the minimum temperature. For further improving the model forecast, Neural Network (NN) as well as regression techniques are applied so that  the systematic errors of the direct model output of the model for maximum and minimum temperature could be reduced. The study shows that both Neural Network approach and regression technique are capable to improve the  forecast skill  of maximum and minimum temperature. Daily modified forecasts are found persistently closer to the observations when the method is tested with the independent sample. The methods are found to be promising for operational application.

Downloads

Published

01-01-2009

How to Cite

[1]
S. K. R. . BHOWMIK, S. . NATH, A. K. . MITRA, and H. R. . HATWAR, “Application of Neural Network Technique to improve the location specific forecast of temperature over Delhi from MM5 model”, MAUSAM, vol. 60, no. 1, pp. 11–24, Jan. 2009.

Issue

Section

Research Papers

Most read articles by the same author(s)

1 2 3 4 5 > >>