Sinai peninsula is considered an important region for Egypt at both the national and strategic levels. Roads network is necessary for urban planning. Also base maps forms a base for determination of the soil suitability for reclamation, urban development and selection of the suitable type of development and investment that could be made in the area. An accurate and up-to-date road network database is essential for GIS (Geographic Information System) based applications such as urban and rural planning, transportation management, vehicle navigation, emergency response, etc. Since Sinai have rough terrain (hilly and mountainous areas), when producing base maps (orthoimages) the topography should be taken into consideration. DTM and DSM will be produced from the stereo satellite images of SPOT4 so we will get a representation of height. The production of DSM and DTM have been performed on the digital photogrammetric workstation (Leica Photogrammetric Suite) LPS. The digital photogrammetric procedures include collection of GCPs,tie point measurements, aerial triangulation,block adjustment,manual DTM creation for the bare land, automatic DSM creation andDEM editing. The quality of the generated DTM has been validated. After that orthorectification of stereo satellite images has been performed on the digital photogrammetric workstation LPS. Orthorectification of mono multispectral images that is available for the study area has been implemented on the Erdas imagine. This has been followed by assessment of the quality of orthorectification using check points after that a mosaic of images has been made which has been used as base map. The results of producing the orthoimage from digital photogrammetric workstation and from image processing software have been evaluated. Spectral Angler Mapper and maximum likelihood classification algorithms have been comparedfor classifying fused SPOT4 mosaic and SPOT5 imagewith and without incorporation of DSM as an additional channel. It was found that Spectral Angler Mapper was considerably more accurate thanmaximum likelihood classification. After producing base maps roads have been extracted from manual digitizing of orthoimages and from automatic classification usingSAM taking into consideration DSM that have been produced from SPOT 4 stereo satellite images as a channel with the satellite image. In comparing between high and medium spatial resolutions for the classification using SAM algorithm, it was found that even if SPOT4 has a medium spatial resolution and that sub-pixel contamination from different land cover is evident while selecting endmembers, it has given good results. On the other hand, SPOT5, which has a fine spatial resolution there are no sub-pixel contamination, gave lower accuracy results. The majority of roads can be detected even without a DSM, though there are a relatively high number of false positives, mostly urban area. Both the completeness and the correctness values have notably improved compared to the results without the DSM. Using a DSM improves both the completeness and the correctness of the results, primarily because urban can now be clearly separated from roads. The correctness is improved because urban are not extracted as false positives.