A parsimonious model for prediction of monsoon rainfall in India

Recently Gowariker et al. have used multiple and power regression involving 15 independent variables for long-range forecasting of monsoon rainfall in India. They have also argued that, when most of the independent variables are 'favourable', almost invariably the monsoon rainfall is norma...

Full description

Saved in:
Bibliographic Details
Published in:Current science (Bangalore) 1991-04, Vol.60 (7), p.446-448
Main Authors: Paranjpe, S. A., Gore, A. P.
Format: Article
Language:English
Subjects:
Drought
Earth, ocean, space
Exact sciences and technology
External geophysics
Genetic research
Logistic regression
Meteorology
Modeling
Monsoons
Northern hemisphere
Parsimony
Rain
Regression analysis
RESEARCH COMMUNICATIONS
Weather analysis and prediction
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Recently Gowariker et al. have used multiple and power regression involving 15 independent variables for long-range forecasting of monsoon rainfall in India. They have also argued that, when most of the independent variables are 'favourable', almost invariably the monsoon rainfall is normal. In this note we formalize this approach using a parsimonious logistic regression model. The probability of a normal rainfall can be assessed in most cases using only five of the 15 variables. Of these, three are related to temperature and two to wind. This gives us correct results for 36 (out of 38) years, including the exceptional year 1957, when all but one independent variables were unfavourable and yet rainfall was normal.
ISSN:0011-3891