Microwave data analysis 
Microwave data has special importance to remote 
sensing scientist owing to its all weather capabilities. It 
was observed that for the total data indented by CAPE 
Project, not more than 25% of the total optical sensor data 
was available for the year 1996-97 due to cloud cover 
problem. Detailed studies using microwave data analysis 
for crops has shown that rice crop can be discriminated at 
better that 90 accuracy, higher than optical data and early 
detection with multiple forecast is possible (Chakraborty et 
at, 1997). 
National-Level Wheat Forecasts 
The procedures developed under CAPE for regional 
and district-level estimates using single-date high 
resolution RS data have been supplemented with national- 
level forecasts. For this purpose use of multi-date WiFS 
data (coarse resolution and high repetivity) has been 
explored since 1995-96 season. This approach uses 
Arc/Info GIS software for sample design and ERDAS 
image processing software in an integrated procedure. 
Use of Arc/Info GIS in National-level wheat 
estimation is made for making area sampling frame, 
storing segment-wise crop proportions for acreage 
estimation and mean vegetation indices for condition 
assessment. The procedure adopted consists of creating the 
following polygon layers : (i) District boundaries (from 
1:250,000 scale topographic sheets and boundaries 
pertaining to 1990 in Lambert Conformal Conic 
Projection), (ii) Boundaries of Image Processing zones 
(quarter UTM zones) and (iii) a 15 x 15 km sampling grid 
generated internally in Arc/Info through fishnet function. 
The district boundaries are linked to district-wise historical 
area and production attributes for identifying wheat study 
districts (districts > 50,000 ha of wheat) and to 
meteorological subdivision for defining yield strata. Using 
WiFS images number of agricultural pixels in each 
segment is estimated and stored as an attribute and 
classified into 3 strata (>70, 30-70 and 5-30 percent 
agriculture). Sample segments are selected from each 
stratum proportional to size of stratum. The multi-date 
WiFS data is classified using a hierarchical decision-rule 
based classifier. The results for 1995-96 and 1996-97 
season have been described earlier (Oza et al., 1996, 
Bhagia etal., 1997). 
Using this approach in 1997-98 wheat season, 
multiple forecasts (I: Total rabi crop using WiFS data till 
early February, II: Mid season assessment- data till end 
February and III: Final forecast - March end) were made. 
It was also possible to detect inter-seasonal crop growth 
differences using multi-year WiFS-based change detection. 
The production forecast employed temperature-wheat 
yield regression models. 
CROP YIELD MODELLING & ESTIMATION 
The spaceborne RS observations add a new 
dimension to conventional approaches of crop yield 
modelling and forecasting (biometric observations, 
econometric and crop-weather models) due to their dense, 
repetitive observation capability of crop and its growing 
environment. Thus, a number of approaches have been 
investigated for use of RS data for crop yield estimation/ 
forecasting, namely, (a) Assessment of Crop Growing 
Condition (rainfall, temperature or soil moisture or 
vegetation vigour), (b) Development of empirical- 
statistical relations between spectral response and crop 
yields, (c) Use of RS data in physical models or as input to 
crop simulation models and (d) Use of RS in post-harvest 
yield estimation. Approaches (a) to (c) are suited for real 
time monitoring and forecasting applications and studies 
on use of GIS are available for approaches (a) and (c). 
RS in Assessment of Crop Growing Conditions 
Crop environmental parameters like rainfall, temp 
erature, soil moisture etc. can be estimated using satellite 
based RS data'. These information, along with, the extent 
of vegetation cover, received from any coarse resolution 
sensor (NOAA/AVHRR), can be used for an early crop 
condition assessment. These early assessments are used for 
early warning of any food crisis precipitated by climatic 
events. FAO-ARTEMIS provides the Meteosat-based 
rainfall estimates and the NOAA/AVHRR-based assess 
ments of vegetation cover, which are used as the inputs for 
operational monitoring of crop conditions under the Global 
Information and Early Warning System (GIEWS). An 
index which quantifies stress using RS data is Satellite 
Derived Stress Index (SDSI). It is estimated as: SDSI - 
(DT - AT) / (DT - NT), where DT and NT are satellite 
acquired day-time maximum and night-time minimum 
temperatures, respectively, while AT is day-time air 
maximum temperature from meteorological records. 
Empirical-Statistical VI-Yield Relationships 
The basic assumption in such a use of RS data is that 
plant parameters (e.g. biomass, LAI) at a critical stage are 
related to final grain yield and since RS data is strongly 
correlated with many canopy parameters, a direct 
regression between RS data and yield is feasible. RS data 
offer certain advantages over procedures based solely on 
meteorological data since RS data directly view the crop, 
respond to non-meteorological factors as well as large 
deviations from ‘normal’ weather conditions and allow 
either full coverage or stratification, leading to low- 
sampling errors. The use of RS-based direct approach for