2004 
| 
ORTHORECTIFICATION AND GEOMETRIC QUALITY ASSESSMENT 
OF VERY HIGH SPATIAL RESOLUTION SATELLITE IMAGERY 
sinn FOR COMMON AGRICULTURAL POLICY PURPOSES 
e for 
vould J. Chmiel ^^, S. Kay, P. Spruyt* 
their 
uon. | * European Commission DG-JRC, IPSC, MARS Unit, TP266, I-21020 Ispra (VA), Italy - 
simon.kay(@jrc.it, peter.spruyt@jre.it 
® Warsaw University of Technology, Inst. of Photogrammetry and Cartography, 00-661 Warsaw, Poland 
— Jerzych@gik.pw.edu.pl 
KEY WORDS: IKONOS, QuickBird, EROS, geometry, errors, quality, agriculture 
ABSTRACT: 
ror in 
ct on 37 sites for the checking of farmers’ aid applications under the Common Agricultural Policy (CAP), covering an area 
of over 10,000 km”, were imaged with very high resolution (VHR) images (Ikonos, QuickBird, Eros) in the framework of an 
operational testing programme. Determining the performance of VHR imagery orthorectification, and geometric quality 
tainty assessment of the results in relation to the different factors, were important goals of the programme. 
pases. The study, located throughout the extended European Union (15 Member States and 10 accession countries) and 
managed by the European Commission in parallel to its operational work, was intended as a broad validation programme of 
VHR data use. Two sources of results and experiences — one delivered by contractors and a second one from internal 
‘or in reprocessing of orthorectification done at the JRC — were compared and evaluated. 
netric This paper presents results recorded for Ikonos (Geo Ortho-kit), QuickBird (Standard OrthoReady) and EROS 1A 
level data, applying different approaches (variants, models) to orthorectification. The results show that VHR orthoimage 
products generally meet the geometric specification of 2.5m (1D) RMSE corresponding to EU technical requirements. 
elling Existing limits of geometric accuracy under certain circumstances are identified and described. Quantitative and qualitative 
evaluations of orthorectified imagery were carried out using independent check points, according to a standardized protocol, 
helping to identify the influence of different factors on the geometric accuracy level of orthoimage products. A positive 
mual, outcome of this assessment is a demonstration of the validity of the use of VHR remote sensing inside the CAP for control 
The 
USA 
purposes, as well as for the use of spaceborne VHR in CAP applications in general. 
1. INTRODUCTION 
applied in GIS or mapping oriented area applications e.g. 
terrain analysis, thematic information extraction, area 
ex Very high resolution satellite image technology provides a rich measurements, etc. 
source of up-to-date, large scale, geospatial Earth-observation The theoretical basis for orthorectification methods of VHR 
ces m data. Such images have already confirmed their usability in satellite imagery, as well as some research based results, are 
nesia. 3ented.. annliicati vat - arameter P 
l many mapping oriented application areas. The parameters well documented in the literature although the VHR technology 
a (image resolution, repeat cycle, etc.) of VHR sensors combined is still relatively new (e.g. Grodecki, 2002, Toutin et al., 2002, 
with metric image properties have strong potential for Control Jacobsen, 2002). Consequently, the main geometric accuracy 
y on "m x n Canal M ^ 4 > x 2 far vapMmail 3 : a > 2. T a 
ation with Remote Sensing (CwRS), one of methods for verification ^ constraints for orthorectified products and approximately which 
' ^ FE ' ir ‘ D inu ‘ Où aid « ni 3 ; e : " 
nental of EU Common Agricultural Policy (CAP) aid applications by conditions (especially for ancillary data such as ground control) 
2 g 
farms in Europe. Such images can be used for determination of 
the area and land use of declared agricultural parcels. To be 
applicable for this purpose the images have to meet first certain 
requirements defined in relevant guidelines stipulated by 
European Commission (EC, 2004). One of the important 
criteria is the orthoimage geometric accuracy at least 2.5m as 
RMSE, measured on a set of independent check points. 
Taking into account the fine spatial resolution of 
panchromatic VHR satellite images and their information 
content, it can be noted that the planimetric accuracy of the 
delivered raw image, in comparison to ground sampling 
distance (GSD) (or pixel size) is relatively poor, due to a 
number of geometric distortions. The inherent geometric 
accuracy of images and their approximate georeferencing 
(computed for some product levels from the source satellite 
orbital position and imaging geometry) needs to be improved 
for CAP applications by applying 3D geometric rectification 
(orthorectification), to be able to achieve the 2.5m RMSE 
accuracy. Orthorectification is a process which converts images 
into map-like (metric quality) form by accurately removing 
from it satellite, scanner (camera), and terrain related 
distortions. The resulting orthoimage can then be directly 
are required to achieve a certain level of accuracy are well 
known. Nevertheless, it is often more difficult to meet all the 
requirements in operational production conditions, unlike 
research or scientifically oriented study conditions. In a large 
scale operational mode, therefore, practical questions can arise 
concerning the availability of good quality ancillary data, cost 
effective procedures and methods in given circumstances, and 
how far the departure from the best practice type and quality of 
input data can still give acceptable results. These issues are 
important for the preparation of correct technical 
recommendations, as well as with respect to the success of a 
given project or campaign. 
The goal of this validation programme is directly linked to 
VHR orthoimagery use in the CAP, and recommendations to a 
large community of user. The paper is focused on geometric 
accuracy issues of image orthorectification, and describes the 
results obtained from the programme in this respect. À positive 
result of the assessment for images acquired throughout 
different geographical sites in the Europe can confirm the 
validity of the use of VHR satellite images for control purposes 
and other applications related to the CAP. 
1019