VISUAL IMAGE INTERPRETATION OF HIGH RESOLUTION IMAGERY 
FOR REAL-TIME AGRICULTURAL MONITORING IN EUROPE 
M. Sharman 
Agriculture Project 
Institute for Remote Sensing Applications 
Joint Research Centre, Ispra Establishment, Italy 
ISPRS Commission Number VII 
ABSTRACT CONTEXT OF ACTION 4: RAPID ESTIMATES 
The Pilot Project of Remote Sensing applied to 
Agricultural Statistics (EEC) has been 
established to demonstrate a methodology for 
monitoring crop acreages and yields in Europe by 
remote sensing. One of the Actions of this 
project (Action 4) involves the visual 
interpretation of high-resolution (SPOT and TM) 
images for real-time monitoring and estimation of 
changes in crop acreages. The estimates are made 
from images collected repeatedly over sample 
sites scattered throughout Europe, of which there 
will be 50 at the end of the demonstration phase. 
Although ground data are collected at the sample 
sites, they are not made available to the image 
interpreters until the end of the cultural year. 
Estimates must therefore be made in the absence 
of current ground data, and operational 
requirements demand that results are available 
within two or three weeks of image acquisitions. 
The sampling strategy and the methodology of 
image interpretation assisted by software are 
discussed. Preliminary results are presented and 
the significance of these results to the wider 
aims of the Pilot Project, and the means by which 
they are to be integrated into an Advanced 
Agricultural Information System are pointed out. 
Key Words: 
Agricultural Monitoring, High resolution. Multi- 
temporal. Image interpretation 
CONTEXT OF PROJECT 
The Directorate General VI (Agriculture) needs 
precise and accurate information on agricultural 
statistics in order to direct the common 
agricultural policy of the European Economic 
Community. The data it requires include: (1) the 
areas under various crops of economic importance, 
(2) the production of those crops, and, if 
possible, (3) a forecast of EEC production, and 
if possible (4) a forecast of foreign production. 
Such data are also required by the Statistical 
Office of the European Communities (SOEC). 
The Commission has therefore set up the Pilot 
Project for Remote Sensing Applied to 
Agricultural Statistics, which uses both remote 
sensing and agro-meteorological modelling to 
achieve its aims (Meyer-Roux 1987). The project 
is directed by the Joint Research Centre under 
the supervision of the DG VI and the OSCE. 
Financed for 5 years, its work is carried out 
largely by means of contracts with national and 
private organisations throughout the European 
Community. For convenience, the project is 
normally referred to as the "Agriculture 
Project". 
Amongst its many activities, 
undertaking the ambitious task 
resolution satellite imagery 
estimates of changes in area, 
the project is 
of using high- 
to provide rapid 
from one year to 
the next, of a number of selected crops. As a 
secondary aim, the images are to be used to 
provide estimates of the potential yields of 
those crops. The imagery is to be interpreted 
visually by experts assisted by computer. 
Called Action 4 in the nomenclature of the 
project, the high-resolution monitoring is 
designed to give results purely at the European 
scale; that is, it is not intended to provide 
accurate statistics at a more local level. It is 
also intended to lead to the definition of a 
refined and improved methodology for high- 
resolution monitoring of agriculture. 
Action 4 has two distinct parts; the first being 
the ground work carried out for validation, and 
the second, which forms the main subject of this 
paper, being the image interpretation. The 
remote sensing part of Action 4 is undertaken by 
a consortium headed by a French consulting firm 
called "Services de Consultation de l'Observation 
de la Terre" (S.C.O.T.), a wholly-owned 
subsidiary of the Centre National des Etudes 
Spatiales (CNES). The remainder of the 
consortium is made up of the Service Central des 
Enquêtes et Etudes Statistiques (SCEES), a 
software house called Geosys, and SGS Qualitest. 
It is associated by convention with the LERTS. 
It is here referred to for convenience as 
S.C.O.T. Based at Toulouse in south-west France, 
the consortium benefits from its situation at the 
heart of the European industry of space-related 
research . 
SAMPL ING SCHEME 
t o 
Clearly it is out of the question to attempt 
monitor all of the 2.25 million square kilometers 
of the EEC with high-resolution imagery. The 
cost and volume of data make it necessary to use 
a method based on a set of sample sites. The 
project will demonstrate Action 4 on 50 sample 
sites throughout the European Community, but will 
work up to that number over several years 
described below. Each site is 40 km on a side, 
and can therefore be covered completely by a 
single SPOT (60 x 60 km) scene or TM (90 x 90 km) 
quarter scene. 
', SMI 
WWKÈSfèv. 
■