The city of Caracas, as many others Latin-American cities, has experimented a fast growth in the late 20 years, demographic pressures, and the lack of an appropriate urban planification, and others socio-economic problems, tend to reinforce the urbanization phenomena, that has transformed the environment, and the quality of life in these cities. Geospatial tools has provide an interesting perspective to understand the dynamics of these phenomena, the use of thermal band to measure the extension an intensity of urban heat island has been used combined with terrestrial observations, to explain the changes in the urban surface patterns. Combining radiometric, resampling and geometrics corrections techniques, and integrating this information into a GIS, it is possible to compare urban land use to urban surface temperature and identified urban heat critical areas, more accuracy. These works shows the result of the observations develops in the city of Caracas. The geospatial analysis was developed, using LANDSAT 7 ETM + images for the period of selected, ERDAS 11 for image processing, and Arc-Gis 10 for cartographic development. Geometric correction (pixel by pixel) with ERDAS, allow us to work at urban scales, in order to observe the variations in the urban canopy related to the urban surface temperatures patterns. Results of the study were useful to identify critical areas and urban structures related to these thermal patterns. This information will be use by urban planners, to develop mitigations and adaptation strategies, in order to prevent the intensification of the urban heat island, during the occurrence of an strong dry season, or heat waves, which might affect the city, the populations and the environment.

Key words: GIS, urban heat patterns, adaptation-mitigation strategies.

In this paper, we investigated the joint use of both a pair of high resolution WorldView-2 optical satellite images and the TerraSAR-X synthetic aperture radar satellite images, to extract 3D building information (including 2D building footprints and the third dimension building height) in high density urban areas. The main idea of the proposed fusion model is to take full advantage of both datasets for building extraction. Compared with SAR images, optical satellite images are more suitable for extracting building footprints, and can be applied to retrieve the heights of low buildings with higher accuracy. Whereas SAR images perform much better in retrieving the heights for tall buildings. But due to positioning errors and mutual interference of surrounding buildings, SAR images cannot be applied to retrieve the heights of low buildings efficiently. Therefore, in this study, both dataset are combined to generate 3D building product for high density urban area, where a large number of both tall and low buildings are included.

The proposed approach includes two main stages. Firstly, building footprints extraction. Optical satellite images are utilized to extract building footprints by using object-based analysis method, in which both building spectral, texture, contextual and elevation information are used. Secondly, building height retrieval. Initial heights of buildings are retrieved from both Stereo images and SAR images using photogrammetry and SAR interferometry techniques separately, then both initial results are combined with a novel object match based fusion method, in which heights of points for the same building footprint are retrieved and integrated. The proposed approach is especially suitable for building extraction in high-density urban area where single satellite data has certain limitations. Experimental tests on Mong Kok area of Hong Kong city showed that the proposed approach with both stereo images and SAR images can achieve a mean absolute height retrieval error of 6.8 m, superior to the results from stereo images or SAR images with a mean absolute error of 9.4 m and 12.4 m, respectively. Accuracy of the extracted 3D building product meet the requirement of urban climate simulation, which has also been verified in our urban climate model at different scales in Hong Kong study area.