63 
from multi-date AVHRR data have been used for crop 
yield modelling for wheat (Punjab, Dubey et al, 1991), 
rice (Orissa) and sorghum (Maharashtra, Potdar 1993). 
For cotton in Gujarat multi-date LISS-I data have been 
used to generate spectral profiles. In-season yield 
forecasts have not been possible using these models 
because of their multidate data requirement, and (c) 
Combination of different parameters like trend, RS and 
meteorological parameters either by including all in a 
multiple linear regression equation (wheat in Punjab, 
UP) or by optimal combination of different estimates 
(wheat in MP, mustard in Gujarat) has improved the 
forecasts. 
CROP YIELD MODELLING USING GIS 
The use of GIS (alone or with RS data) for crop 
yield modelling has not been adequately explored 
although there is immense potential for this type of 
linkage at all phases of yield modelling, i.e., preparatory, 
modelling and output phase. 
In the preparatory phase GIS is used for (a) 
stratification/zonation for use in modelling using one or 
more input layers (climate, soil, physiography, crop 
dominance etc.) with model being run separately in each 
homogenous region, or (b) preparing input data 
(weather, soil and collateral data) which is available in 
different formats to a common format for use in model. 
This may involve interpolation or weather observations 
to common grid, or use thiessen polygons (especially for 
rainfall). In case of soils, pedotransfer functions convert 
soil maps to inputs for use in models. The inputs could 
be prepared separately for homogenous regions. 
In the modelling phase the GIS linkage is mainly 
through use of RS data in form of VI profiles or multi 
layer modelling involving raster layers of crop 
distribution. The detailed crop-weather models have a 
soft-linkage with GIS and only interchange data files. 
The final output phase also could involve use of 
GIS either for (a) display of point forecasts in spatial 
context, or (b) for aggregation and display of model 
outputs for defined regions (e.g., administrative regions). 
CASE STUDIES ON CROP YIELD MODELLING 
USING GIS & RS INPUTS 
(a) Crop Productivity Mapping by linking Crop 
Simulation Data and RS inputs: Conese et al. (1986) 
describe development of an integrated data bank of soil, 
weather and RS data and its linkage with ACROM 
model of crop growth simulation to produce maps of 
crop productivity. 
(b) The CROPCAST crop assessment system of Earth 
Satellite Corporation (USA) operationally gives country/ 
continental scale assessments. The region is divided into 
24/48 km grid cells and assignments made for soil, 
cropping practices, crop phenology for each cell. Real 
time RS and meteo-rological data are used to run crop 
simulation and soil moisture budget models, whose 
predictions are evaluated against RS observations to give 
crop assessment. 
(c) Global Information and Early Warning System 
(GIEWS) of Food and Agriculture Organisation (FAO) 
of United Nations aims at providing early warning about 
food security emergencies for the entire globe. This 
assessment uses a wide variety of data including past 
and projected agricultural production, weather and crop 
growing conditions, potential food demand, food prices 
etc. It also uses the ten day data about meteorological 
and vegetation conditions from the ARTEMIS (Africa 
Real-Time Environmental Monitoring using Imaging 
Satellites). Marsh et al. (1993) describe development of 
an integrated workstation where the various software 
packages used for the above task (word processor, 
spreadsheet, RDBMS, GIS and spatial analysis) are 
linked together with a common interface to allow crop 
assessment. 
(d) Linking of Agroclimatic Yield Index and RS Inputs 
for wheat yield modelling, Haryana, India: In this 
approach spatial data layers (administrative map. 
Irrigation map (percent of crop irrigated), Soil Texture 
Map and Soil moisture map (water holding capacity) and 
crop growing season temperature) were generated in 
Arc/Info package and overlaid on each other to output an 
agroclimatic yield index (AYI) by assigning weights to 
the layers on the basis of their suitability for higher crop 
yield. The yield model was a multiple regression 
equation including both AYI and VI (area weighted 
NIR/Red radiance ratios of wheat crop in 1991-92 
season) which had higher coefficient of determination in 
comparison to simple linear model consisting of VI only 
(Sharma and Navalgund, 1995). 
(e) CGMS under MARS Project, Europe: The MARS 
(Monitoring of Agriculture with Remote Sensing) is an 
ongoing project coordinated by ISPRA for European 
Commission and includes a Crop Growth and 
Monitoring System (CGMS) which is an agro- 
meteorological modelling system to provide crop-state 
assessments and yield forecasts using a GIS database. 
The database forms an important component of the 
system and uses ARC/Info to organise spatial data and 
Oracle RDBMS to organise tabular data (Burill et al., 
1985). The databases include daily historical and current 
meteorological data base covering 350 stations for 5-15 
years, altitude, soil map and phase and attributes to 
create moisture availability. The use of WOFOST for