ining 
field 
6. 
solar 
Note. 
tions 
1wal, 
5-96 
mote 
hari, 
ER; 
area 
n of 
Sens. 
002. 
and 
Sci., 
Soils 
ls of 
Use 
,000 
n of 
heat 
Tic. 
ling 
s of 
Nov 
°;rop 
ohic 
ural 
arch 
KR 
jon, 
for 
e in 
and 
tor, 
IAPRS & SIS, Vol.34, Part 7, "Resource and Environmental Monitoring", Hyderabad, India,2002 
FORECASTING DISTRICT-LEVEL SORGHUM YIELD USING 
STANDARDIZED PRECIPITATION INDEX (SPI) IN KARNATAKA, INDIA 
Sujay Dutta*', V.K. Dadhwal*# and V.S. Prakash+ 
* Crop Inventory & Modelling Division, 
Space Applications Centre (ISRO), 
Ahmedabad - 380 015 
* Drought Monitoring Cell, 
Cauvery Bhawan, 
Bangalore 
# Commision VII, Working Group VII/6 
Key Words: Standard Precipitation Index, Sorghum yield forecasting, Rainfall data 
ABSTRACT 
Sorghum (Sorghum bicolor (L) Moench) is the major rainfed kharif crop in Karnataka state. In sub-tropical regions, 
like Karnataka, changes in meteorological parameters particularly rainfall and temperature cause large year-to-year 
variation in yield. The study districts have large fluctuations in rainfall during the monsoon season. There is a need to 
develop a simple model for yield forecasting during this season. Standardized Precipitation Index (SPI) have been 
shown to have good correlation with grain yield of many rainfed crops. Here, SPI has been used for predicting sorghum 
yield at district level in Karnataka state. The variability of sorghum yield explained by SPI varied from 36 to 96 %. 
1. INTRODUCTION 
Sorghum (Sorghum bicolor (L) Moench) is one of the 
major rainfed kharif crops in Karnataka state covering 
about 50 lakh ha with production of 75 lakh tons. 
Weather plays a significant role over the growth and 
yield response crops. In sub-tropical regions, like in 
Karnataka, changes in meteorological parameters 
particularly rainfall and temperature causes large 
variation in year-to-year yield. In this area, the time 
clustering of precipitation during the growing season of 
sorghum crop shows complex behaviour of rainfall 
shortage and excess. Various models have been 
developed for assessment of crop response to 
meteorological factors and development of models for 
prediction of yield. Newman 1974, have shown 
statistical techniques for meteorological data as good 
predictor of crop yield forecasting. Dutta et al. (2001), 
have demonstrated application of regression based agro- 
meteorological model for wheat yield prediction in 
Bihar state. Similar attempt has been made by Oza et al. 
(2002) by correlating Standardised Precipitation Index 
(SPI) with bajra yield in Rajasthan and groundnut yield 
in Gujarat. Viau at al. (2000) have shown that 
appropriate use of indicators coming from 
meteorological and remote sensing observations provide 
a good tool for detection, quantification and assessment 
of drought. There is a need to develop a simple yield 
forecasting model using a realistic assessment of 
  
* corresponding author, e-mail: sujay_dutta@indiatimes.com 
available soil moisture and the water requirement of the 
crop in the region. Dutta et al. (2002), have evaluated 
sorghum yield prediction using three different 
approaches viz. water balance approach, crop simulation 
approach and SPI method. Among these SPI have been 
found to be robust under drought conditions. 
Sorghum is grown here mainly as a rainfed crop. Thus 
precipitation is the critical component among the 
meteorological variables affecting crop yield. As such 
modelling the cause and effect relationship between 
yield and this factor derived as a precipitation index has 
been attempted. 
2. STUDY AREA AND CROP 
CHARACTERISTICS 
Four districts of Karnataka state, namely Raichur, 
Bijapur, Dharwad and Gulbarga have been taken for 
yield prediction. Reported daily rainfall for have been 
collected from 1960 to 2001. Daily rainfall data from 
1960 to 2001 of rain gauge stations located at all four 
districts were converted to monthly total. These monthly 
rainfall data have been analysed to calculate SPI index 
at monthly, bi-monthly and tri-monthly time scales. The 
monthly SPI for the growing season (June to October) 
are correlated with sorghum yield. The yield level varied 
from 400 to 2000 kg/ha among the study districts. The 
significant SPI months were selected using stepwise 
regression approach at 90% confidence interval. The