Aigner, Edgar 
  
Crop Yield Estimation Using NOAA - AVHRR Data and Meteorological 
Data in the Eastern Wimmera (South Eastern Australia) 
Edgar AIGNER*, Isabel COPPA**, Friedrich WIENEKE*** 
*Technical University Munich, Germany 
Chair for Photogrammetry and Remote Sensing 
edgar.aigner@ photo.verm.tu-muenchen.de 
**RMIT University Melbourne, Australia 
isabel.coppa@rmit.edu.au 
***Ludwig—Maximilian University Munich, Germany 
f.wieneke @igef.geo.uni-muenchen.de 
KEY WORDS: Agriculture, Applications, Evaluation, Land use/Land cover, Modelling, Multi-spectral data, 
Multi-temporal, Remote Sensing 
ABSTRACT 
Having an estimate of final yield early in the growing season can be a powerful management and economic 
tool for the farming community. Therefore the possibility of using temporarily high resolution remote sensing 
data in combination with daily meteorological data for crop yield prediction on a close to field scale has been 
investigated for one of the main cropping areas in south—eastern Australia. The lack of rainfall in semi-arid to 
semi—humid climate of this region is one of the major limiting factors to crop growth. The relation of different 
parameters, such as the “Normalized Differential Vegetation Index” (NDVI), the date of the commencement of 
the grainfilling stage (GF), the water-Stress Degree Days” index (SDD), as well as the growing season 
rainfall, to yield of canola, wheat and cereals (wheat and barley) have been examined. Using information from 
1995 to 1997, a crop yield estimation model on the basis of a multiple linear regression model has been 
developed and evaluated. The following paper reports the results of a study carried out at the Commonwealth 
Scientific and Industrial Research Organization (CSIRO), Aspendale, in collaboration with the Department of 
Natural Resources & Environment (NRS), Melbourne, and the University of Munich. 
1. USING REMOTELY SENSED DATA FOR CROP YIELD FORECASTING AS AN 
AGRICULTURAL MANAGEMENT TOOL 
For farmers in the eastern Wimmera (Victoria, Australia) it is important to estimate final yield early in the 
growing season. Using their knowledge and experience about the local conditions they are able to estimate 
yield to a certain extent. Having reliable predictions of yield on a close to paddock scale could support their 
management and economic decisions. 
A basic parameter for crop yield, especially in semi-humid to semi-arid regions is growing season rainfall 
(GSR). Therefore, the GSR is often used by farmers to estimate final yield. Also, remotely sensed data have 
proved to be a good source of information for agricultural applications, in particular for yield estimation. 
Amongst others, M. S. RASMUSSEN (1997) and S. MOULIN et al. (1998) give a good review over past and 
present trends. 
One of the most common approaches is the use of vegetation indices, such as the Normalized Difference 
Vegetation Index (NDVI) as a measure for plant growth and development. M.S. RASMUSSEN (1992) 
examined the integral of NDVI from data of the Advanced Very High Resolution Radiometer (AVHRR) over 
the phenological stages of reproduction of wheat for 4 km? areas. N.A. QUARMBY et al. (1993) also 
investigated AVHRR-data and created 4 years time-series of NDVI. They point out the importance of the 
grainfilling period of wheat for final yield. M.P. CABEZON and J.C. TAYLOR (1994) analyzed correlations 
of multiple linear regressions using variant combinations of rainfall parameters and vegetation indices as 
independent variables. R.C.G. SMITH et al. (1991) found, that in a mediteranean-type environment the 
vegetation index — yield relationship does not explain yield variations significantly better than the rainfall — 
yield relationship over the growing season. However, they also point out that a combination of such 
information in a multiple linear regression model do improve the correlation significantly. 
Since S.D. JACKSON et al. (1977) it is known, that observations in the thermal band of the EM-spectrum can 
be used as an indicator for the plants' water-stress. They use the difference of observations of daily surface 
temperature and air temperature for deriving the water-stress index “Stress Degree Days" (SDD). 
To better take into account the course of vegetation development, temporarily high resolution remote sensing 
data have to be applied. Only frequent observations allow the examination of the crop’s response to changing 
agrometeorological conditions. Such data can be obtained from the AVHRR-sensor onboard the polar orbiting 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 
19