A CANADIAN APPROACH 
TO LARGE REGION CROP 
AREA ESTIMATION WITH LANDSAT 
by 
R.A. Ryerson 
Canada Centre for Remote Sensing 
Ottawa, Canada 
J.-L. Tambay 
R. Plourde 
Statistics Canada 
Ottawa, Canada 
ABSTRACT 
This paper describes the methodology used and results of an application 
of Landsat data to timely Canola/rapeseed area estimation in a 140 000 sq. km. 
region of western Canada. Of special importance is the integration of the 
strongest features of digital image analysis with visual image interpretation. 
The resulting methodology is particularly useful for rapid estimation of crop 
areas in large regions. 
RESUME 
Ce document décrit la méthodologie et les résultats utilisés aux cours 
d'une application des données Landsat qui consistait 3 déterminer, par 
télédétection en des temps opportuns, les surfaces destinées à la culture de 
colza canola sur une superficie totale de 140 000 km? dans l'Ouest canadien. 
D'intérét tout particulier est 1'incorporation des analyses numérique des 
images aux analyses manuelles, qui fut  accomplie en utilisant les 
caractéristiques les plus prononceés des deux analyses. Les méthodologie sui 
en résulté s'avére trés utile afin d'estimer rapidement de grandes surfaces de 
cultures. 
  
INTRODUCTION 
This paper presents an approach developed in Canada to rapidly and 
accurately determine areas of crop, such as  Canola/rapeseed, over large 
regions. The methods are based on the integration of digital image analysis 
and visual image interpretation, using the best features of both. Research and 
development in using Landsat data for potato acreage estimation began at the 
Canada Centre for Remote Sensing (CCRS) in 1975 (Mosher et al., 1978; Ryerson 
et al., 1980). Success of this R&D effort resulted in methods now proven 
capable of generating accurate and timely estimates in two successive years 
(Ryerson et al., 1981; Ryerson et al., 1982a). The method developed to 
generate timely crop estimates over a small area has been fully documented 
elsewhere (Ryerson et al., 1982a). The question addressed here is how well 
these methods, or their derivatives, work in larger, more diverse regions with 
similar constraints of timeliness and accuracy. 
The following two sections address the general approach to analysis 
adopted for agricultural projects at CCRS and the specific methods developed. 
In the fourth section, details are given on the problems encountered in the 
367 
aaa Rt 2 A CNN = ce ee a A A EH A re en em lll