Context and objectives
Some authorities (DGPL) of the Walloon Region (Belgium) are involved in the realization of a complete regional coverage (16,844 km2) of natural colour ortho-photomaps to be used at large scales (up to 1:4000). These are made of very high resolution digital ortho-rectified aerial photos presented as raster files. Those rasters are grouped by corresponding administrative boundaries and distributed to the local authorities in order to be used for various land and urban planning purposes. One goal of the DGPL is to provide the local administrations with up to date image data, within the limits of some obvious financial constraints. Because significant changes in the landscape appear at different rates over time and space, it was suggested to devise a methodology to find out where these changes would require a rapid update of the imagery. In regard to this situation, a pilot project has been set up in the year 2000, aiming at detecting the location, the extent and the nature of landscape changes in contrasted areas – urban, suburban, rural – from adequately processed very high resolution (VHR) satellite images – typically Ikonos multi-spectral (4m) and panchromatic (1 m) images. Hence, this project deals with various issues related to multi-source analysis, geometric and radiometric aspects of processing VHR data, diachronic analysis, in order to provide the administration with the basic knowledge and the appropriate tools to carry out the proposed task.
Expected scientific results
The topics discussed in this paper show that in order to define a policy for the progressive renewal of the coverage of ortho-photomaps derived from aerial photography, a viable alternative is to proceed to a diachronic analysis for change detection between the actual cover and VHR satellite imagery. The resulting analysis provides an indication of where a coverage renewal would be appropriate and constitutes thus an aid to any decision process for this matter. It is of course now up to the DGPL administration to proceed to the diachronic analysis and to produce the concomitant change detection map for the whole region.
For now, part of the proposed methodology rests on a somewhat tedious manual screen digitising procedure. However, in a near future, with increasing quality of the acquired orho-photomap coverage, as also with a progressive improvement of the organization of - and the access to - the available ground occupation databases, this task of updating will certainly evolve towards more automation. A particular attention should be paid to the existing PICC database which unfortunately could not be incorporated in time into this case study.
The proposed methodology relies on sophisticated technologies involving digital image and GIS processing for which specific training is required. This expertise could be provided with the aid of appropriate consultancy such as can be offered by university departments involved in remote sensing and GIS related matters. A symbiosis with some research departments would be an efficient manner to assist the introduction of high technology into the administrations and to keep the necessary knowledge and skills up to date. It would result in improved product and service quality and also enable the administrations to become less dependent of some monopolistic private companies.
Another topic of vital interest in relation with our subject is the rapid evolution of spatial imagery products. Although currently available satellite images can serve many purposes, 1 m resolution panchromatic imagery still doesn’t compete favourably with the 40 cm resolution colour images stemming from aerial photography, and are therefore less suitable for many spatial planning surveys and management activities at large scale. But that was for yesterday. Today, new generation satellites are planned, some are already orbiting, and start producing higher resolution images - for now up to about 70 cm resolution - at a very competitive price of a few tens of dollars per km2, depending upon the quality of the ordered product. Such images would almost certainly provide a viable alternative for replacing aerial ortho-photos in areas of slow rate of change such as rural and forest land. Whatever, these facts clearly show the evolutionary trends: better image quality, product homogeneity, reduced costs, complete coverage acquired within reasonable delays, and rapid revisit times. What else could we ask for ? Within this scenario the answer for updating an image database is of course immediate. And how logical and cost-effective would it be to get all the administrations concerned to agree on a common acquisition of such an image cover, all of them are hungering to use for their own needs. It would benefit to all.