Record Details

EVALUATION OF STATISTICAL MODELS FOR FORECASTING OF COTTON YIELD IN GUJARAT STATE

KrishiKosh

View Archive Info
 
 
Field Value
 
Title EVALUATION OF STATISTICAL MODELS FOR FORECASTING OF COTTON YIELD IN GUJARAT STATE
 
Creator NARBHERAM J. RANKJA
 
Contributor Upadhyay S.M.
 
Subject COTTON YIELD
Agricultural Statistics
 
Description ABSTRACT
The present investigation was undertaken to identify efficient cotton growing zones (districts) in Gujarat state and to develop pre-harvest forecasting models for cotton yield based on weekly average weather data over a period of 32 years (1975 to 2006) covering five selected cotton growing districts of Gujarat state.
Timely and reliable forecast of crop yield is of great importance for monsoon dependent country like India, where the economy is mainly based on agricultural production. Cotton being an important cash crop mainly grown under rainfed condition, the fluctuations in yield levels over the years are due to weather behavior.
For identification of the efficient zone of cotton cultivation, the collected data were used to compute Relative Yield Index
(RYI) (Kanwar,1972). The statistical parameters like mean, standard deviation and coefficient of variation (CV%) were calculated to interpret the results and to determine the efficient cropping zone. According to RYI value of selected districts, Junagadh and Rajkot districts identified as the `most efficient` zone, Banaskantha and Kheda districts fall in `efficient` zone and the Ahmedabad district fell in `not efficient` zone for cotton crop.
Weather is a major factor affecting crop production in advanced agricultural systems in our country. The large variation in yield is predominantly due to weather parameters in a climatic region.
The weekly average of weather variables used were rainfall, maximum and minimum temperature, morning and afternoon relative humidity and sunshine hours from 22nd meteorological standard week ( MSW) to 47th standard week of each year. The data on area, production and productivity of cotton from 1975 to 2006 were collected from the publications of Directorate of Agriculture, Gujarat State, Gandhinagar.
Three approaches were used for forecasting yield (1) original weather variables, week wise approach (2) Use of
generated weather variables, week number as weight approach (3) Use of generated weather variables, correlation coefficients as weight approach. The time trend was included as a explanatory variable in all the approaches. For early forecast, interval of 17, 20, 23 and 26 weeks were considered. The stepwise regression procedure was adopted using 32 years data for selection of variables. The prediction equations and forecasts of subsequent years were obtained separately for 28 to 31 years data set.
The positive and significant effect of time trend explanatory variable on cotton yield was observed in all the districts, approaches and each of the model fitted. The effect of all the weather variables, in relation to their quantum and direction, differed over the district. However, they were found important for prediction point of view in cotton productivity.
For most of the districts, week wise approach, using original weather variables was found to be superior over other approaches. This approach provided suitable pre-harvest forecasting models for Junagadh, Rajkot, Ahmedabad, Banaskantha and Kheda districts. These models for respective districts can be used for providing pre-harvest forecasts, 4 weeks before expected harvest in case of Junagadh,
Ahmedabad and Banaskantha districts, while 10 and 13 weeks before expected harvest in case of Rajkot and Kheda district respectively. The use of original weather variables, weekwise approach was observed to be the best model for forecast followed by generated weather variables, correlation coefficient as a weight approach model. The approach of week number as weight in generated variables was observed to be minimal importance for predicting and forecasting of cotton crop yield in present investigation. Finally, selected models for pre-harvest forecasts for each districts explained more than 90% variation in cotton yield and the error of simulated forecasts were below 20 per cent in each case.
 
Date 2016-09-20T18:25:15Z
2016-09-20T18:25:15Z
2009-03
 
Type Thesis
 
Identifier http://krishikosh.egranth.ac.in/handle/1/77827
 
Language en
 
Format application/pdf