Selected Interpretation Results:

CAMEROON

The republic of Cameroon has a surface area of 475,442 km² and a population of 12,1 million. Its major income is derived from the exploitation of oil fields and forests. Crops are only cultivated for the country's internal needs. Selected food products are imported. Figure 6 shows an overview map of Cameroon. As the use of ERS-SAR data for coastal zones will be demonstrated over Gabon and Cabinda, Cameroon has been chosen to demonstrate the SAR's potential for the management of hydrological resources. Together with Nigeria, Niger and Chad, Cameroon shares the resources of Lake Chad, which is listed as the fourth largest lake in Africa at between 10,000 and 25,000 km² by different literature sources.

Figure 6: Geographic map of Cameroon

One can immediately see that the simple spread of Cameroon across different climatic zones calls for very sophisticated regional planning and management. Both have to be adapted to the different climatic zones, from the humid and almost tropical South (ocean) to the very dry North (Lake Chad). In the North, desertification is an important issue and is also one of the major factors in the decreasing size of the lake that has been observed in the past 20 years. For this reason, a lake commission was funded by the different countries involved to coordinate work to re-establish the lake's level or, at least, to deal with the new economic and ecological situation. The most important need is therefore for up-to-date maps and information that show the changes in the lake's level and size from one year to the next.

Through these observations, trends can be established and used for medium- and long-term planning for infrastructure items such as roads, etc.

Turning to the environmental issue, the smaller the lake surface becomes, the higher will be the salinity of the lake (through evaporation of the water). This leads to lower amounts of fish (fish normally adapt to a certain level of salinity; changes in these conditions lead to their death). In addition, the original lake bottom which emerges when the lake is decreasing and which is saturated by salt, is blown away and "pollutes" the surrounding agricultural area.

This desertification effect is very difficult to overcome (recultivation of fields) as the salt needs to be washed out or kept below the fertile soil. This is a very expensive and not always successful process.

Image 14 shows the southern border of Lake Chad in a light grey. Some parts of the coastal zone, being in the wind shadow of islands, appear black, due to a flat water surface (no waves). On the left hand side of the image one can see the river Chari. The white spot on the left of the river is a smaller city. Grey level differences over the land surface are due to changes in the composition of the vegetation.

Image 14: ERS-SAR scene acquired on August 1, 1994

Images 15 and 16 show the gains that can be achieved through the use of a multitemporal approach and image filtering.

Image 15: Multitemporal subset (non-filtered)

Image 16: Multitemporal subset (filtered)

A number of simple rules can be applied for the interpretation of these multitemporal data sets. First of all, one can see that some areas are black or white, which means that the surface did not change from one image to the next (difference of 20 days). As the agriculture surface changes very quickly due to the growth of plants or the increase or decrease of humidity in the soil, those areas which tend to keep their grey level can be considered to be stable and covered by material which does not change its spectral properties very quickly.

Different investigations have shown that forests generally appear in a darker colour in ERS-SAR images. Therefore these areas in the lower part of the images must be interpreted as wooded surfaces. White areas are probably sand dunes. Those areas with different grey values between the first and second dates are potentially cultivated.

Image 17 shows a full resolution filtered multitemporal subset of the coastal zone. This example demonstrates that it is possible to map the shoreline of the lake. One needs to apply the multitemporal approach as, depending on the wind speed and direction on a particular date, land surface and large water bodies like Lake Chad tend to exhibit similar grey levels.

Image 17: Multitemporal subset showing the shoreline of lake Chad

Image 18 shows a phenomenon which is probably due to the extension of the lake's surface. Apparently, the lake has been growing in the past year and sand dunes (former land) have been submerged by it. Only the peaks of these dunes can still be seen.

Image 18: Submerged sand dunes in lake Chad seen from ERS-SAR

Image 19, covering an area south of the lake, was acquired in order to study the potential of ERS-SAR for the monitoring of different land cover classes and general topographic mapping in this semi-arid zone.

Image 19: Full image (processed with average filter) acquired on August 12, 1994

Features that can be interpreted in the image include a large number of grey lines on the left side which correspond to the old river beds. The feature in the lower region expressed as a large number of black spots can be explained by metallic deposits which are less eroded than the surface. As the lake surface is reported to have been much larger in former times and these "hills" were considered to be above flood level, the housing areas on these "hills" (white spots) can be explained and easily mapped. A more detailed interpretation of cultivated areas is possible from images 20 and 21. On image 20, one can see a number of smaller fields on the left side which are probably enclosed by bushes in order to reduce wind erosion. The large white area in the middle of the image potentially corresponds to a dry field (smooth surface). Small white spots in the lower part are houses or villages located on the small hills described above.

Image 20: Subset of an ERS-SAR scene over a cultivated area south of lake Chad

Image 21 shows another subset of the scene with some agricultural fields visible as well as a road with villages to the side, which probably leads to the lake located in the north of this area. This subset clearly demonstrates that cartographic map updates can be performed with ERS-SAR data. New housing areas as well as roads can be easily detected. For the mapping of agriculture, one should have at least two images of the same area (multitemporal approach).

Image 21: Subset with road and housing features

The images shown above demonstrate very impressively the capabilities of ERS-SAR data for basic cartographic mapping in this climatic zone. Specifically this will consist of mapping the lake extension and the river network. Changes in surface area and the drying-out or flooding of river beds can be quickly identified. The low price of the images and the need for relatively limited processing software and hardware might be a further advantage and argument for the integration of these data into the operational cycle of map updates. Of course, the images also show that very precise cartography from "scratch" is very difficult to perform with radar data. Assuming that a much higher level of expertise on local phenomena is available in each country, then mapping at a scale of 1:100,000 to 1:500,000 should be feasible. Studies requiring a larger scale (e.g. 1:50,000) are possible, but class affiliation to land cover types will become more difficult to handle.