Beeri, Ofer 
  
PRECISION AGRICULTURE AND REMOTE SENSING: 
VARIANCE ANALYSIS OF WHEAT CROP BY SPECTRAL INDICES 
O. BEERT, A. PELED' , S. KITAIN" 
University of Haifa, Department of Geography, Haifa, 31905 Israel 
   
  
*ofer@geo.haifa.ac.il 
**a peled@uvm.haifa.ac.il 
*** Ministry of Agriculture, Israel 
Working Group 11/2 
KEY WORDS: Precision Agriculture, Multi-temporal Aerial Photography, Vegetation Algorithms. 
ABSTRACT 
Precise agricultural monitoring throughout the growing season, is a key tool for obtaining optimal crop vield at lowest 
possible cost. The paper focuses on variance analysis of crop (wheat) yield by spectral indices. This is part of a 
multi-year ongoing research on precise agricultural monitoring. carried out at the department of geography, University 
of Haifa. This research has two main goals: (1) to test various vegetation parameters and spectral indices for 
monitoring wheat yields; and (2) to compare predicted remote sensing yield maps with combine-GPS vield-maps. The 
paper describes the data collection steps, analyses methods and the results, gained during one, particular, wheat growing 
season. The test field in this research effort, is located near Be'er Sheva, in the semi-arid region of southern Israel. The 
field was divided into 54 test plots, each given different nitrogen and water treatment. Crop was collected on three 
different occasions throughout the growing season, complimented with color and CIR air photography coverage. The 
harvesting was done in two different methods: (1) A 1.25 meters-wide combine harvested the field passing through the 
center of each plot. (2) In a second stage, the entire field was harvested by a "GPS-combine". The actual vield amounts 
were compared to the predicted-yield map, produced by processing the CIR photographs. The spectral indices, 
developed in this research, proved to have better correlation with the yield from each plot, than with vegetation 
parameters, that are used in traditional agriculture monitoring. There was not found a si gnificant difference between the 
correlation of Color-Infrared indices and Visible-Light indices of the predicted yield. The crop yield map, produced by 
modeling the remote sensing data, correlate the combine-GPS yield-map, and has better ground resolution. Thus, it 
depicts better spatial variability for each plot. This variability stems from the farmer activity and may be used for better 
agriculture management. Three of the spectral indices manifested good correlation throughout the season, and will be 
applied, for crop-vield prediction, in the continued research. Using the suggested model for producing predicted crop 
yield maps, will enable better monitoring of wheat with greater spatial accuracy. 
  
164 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000.