STER 98 Remote Sensing Project Tutorials 1 Human Activities and Environmental Risks Natural Hazards and Urban Development Issues Vallée de la Bruche, Alsace Stephen Clandillon, SERTIT, Parc d'innovation, Boulevard Sébastien Brant, 67400 Illkirch-Graffenstaden, France, (http://sertit.u-strasbg.fr) Context Cities are often built on land communications cross-roads but most often where fluvial communications are possible. Due to this many urban developments have undergone or undergo episodes of flooding. Recently, towns and cities are experiencing rapid growth rapid growth and this, in particular since the Second World War. Potentially, constructable lands are often excluded by administrative boundaries, geomorphological characteristics, etc Hence, part of the urban growth has occurred in areas that historically were not considered as correct lands for urban development, and therefore, were developed while taking risks. An example of this is the European-wide urban development which has occurred on the flood beds of rivers and the ensuing media reporting once these areas are affected by flooding, mud slides and mud lows. Strasbourg (Figure 1) is upstream of a number of rivers prone to flooding, one of whom is dealt with, the Ill, with its flood plain in the flooding and agricultural practises projects in this course. The other major river converging on the Ill in the south-west suburbs of Strasbourg is that of the Bruche. The flood bed of this river is considerable and it links up with the Ill s potential flood bed in these same suburbs. 1. Project Objectives The area has been recently administratively classed as an area at risk from flooding. So, the main aim of this project is to analyse the urban development of this area over a 30 year period and to see whether development has occurred within the flood risk zone. To accomplish this a multi-source, multi-temporal and multi-resolution database will be used, involving historic 1962 declassified US Defence Corona Programme data for a privileged look at the landscape 3 decades ago and modern civil data, XS and PAN data, from the current SPOT programme. Moreover a scanned administrative flood zone map will be used.
2 STER 98 Remote Sensing Project Tutorial 2 Database Panchromatic Images: SPOT PAN 1996 US Defence Corona Image 1962 Multi-spectral Images: April 1995 SPOT XS July 1996 SPOT XS USGS Figure 2. Geo-referenced 1962 US Defence, Corona Image Subset of Official Flood Zone (OFZ), Plan N 5, the Bruch Valley (Figure 3) Current French Ordinance Survey Hardcopy Maps 3. Procedure This project involves the building of a geo-referenced database, the multitemporal classification of recently acquired SPOT XS data (1995-96) and its comparison with the 1962 image to develop a dynamic analysis of the urban infracture in and around the Bruche Valley flood zone. Finally, using the Plan N 5 of the flood zone, it will be demonstrated whether urban development has occurred within the flood zone. 3.1 Geo-referencing data The 1962 Corona panchromatic data must be geo-referenced to the Lambert 2 projection using the 1996 panchromatic image as a reference for ground control point (GCP) extraction. Similarly, the scanned Plan N 5 flood risk map must be geo-referenced. Try to find GCP s at the extremities of the image to be georeferenced and in the middle in order to optimise the transformation coefficients and the geometric quality of the resulting image.
STER 98 Remote Sensing Project Tutorials 3 3.2 Classify the Multi-temporal 1995-96 SPOT XS images Figure 3. SPOT XS April 1995 Image with OFZ Plan N 5 Overlaid Figure 4. SPOT XS July 1996 Image Using either a supervised or an unsupervised multi-temporal classification method derive a six theme classification containing the following classes, water, rural mineral surfaces, croplands, woodlands, grasslands, and urban area. Use the April 1995 and July 1996 SPOT XS images for the classification, with the urban mask image, which is provided, to dissociate rural mineral surfaces from urban area. Do any necessary modifications after photo-interpreting the results using the 1996 panchromatic image as backup. If you find a method to incorporate the panchromatic image (Figure 1) into the classification by all means do so. 3.3 Urban Area Analysis of 1962 Corona Image Extract from the 1962 image the area represented by the 1995-96 urban area class and either photo-interpret the 1962 urban boundary or try to extract it texturally. The former method does save time and frustration. 3.4 Dynamic Analysis of the Urban Area 1962-96 Compare the urban area extents and map the areas that have built up between 1962 and 1996. Then, create a mask of the Official Flood Zone from the scanned Plan N 5 and deduce the new urban area that was developed inside the Official Flood Zone.
4 STER 98 Remote Sensing Project Tutorial 4. Selected Illustration of Results Figure 5. Six class multi-temporal (1995-1996) classification of the study area with the application of the rural/urban mask, SERTIT 1998 In Figure 5 is shown the results of the multi-temporal classification using an unsupervised method of classification and a supervised regrouping of classes using the database and topographic maps as a ground truth aid. USGS Figure 6. Thirty four years of urban and rural change encapsulated in two panchromatic satellite images from Corona 1962 and SPOT PAN 1996.
STER 98 Remote Sensing Project Tutorials 5 Above in Figure 6 is an illustration of the dramatic changes that have occurred in the urban and rural landscapes since 1962 with an expansion of the urban area, the regrouping of land parcels and the intensification of agriculture. Figure 7. Urban Change Inside and Outside the Flood Zone between 1962 and 1996, SERTIT 1998 Figure 7 illustrates the urban dynamic between 1962 and 1996 in and around the Official Flood Zone. Much change has occurred some of which has happened within the designated Flood Risk Zone. Thirty years ago some development had already occurred within the flood zone and this phenomenon continued until this day. Another aspect that is remarkable (Figure 6) is the change is the agricultural landscape with the growth of tillage to the detriment of the grasslands and the increased size of cultural land parcel. This project has illustrated that much change has occurred in the last three decades and that some has occurred in potentially risky areas, though this is said whilst not having incorporated a digital terrain model study to account for dykes that may have been built to protect recently developed urban areas. Finally, The time span of 34 years is quite remarkable and illustrates the richness of satellite remote sensing data sources and off course these same data may be used to derive much interesting information on various subjects rendering their use interesting from a technical and economic perspective.