The thesis topic is land cover mapping with pixel based and object oriented image analysis approaches for coal fire research,preformed under the project of "coal fire research in China". The Study area is Wuda, Inner Mongolia, China

Traditional methods use NDVI to investigate vegetation cover from remote sensing imagery. These methods provide per-pixel vegetation distribution, and cause a modifiable areal unit problem (MAUP), when a meaningful statistical result is issued. In this paper, a new method based on advanced segmentation techniques and classification is proposed for urban vegetation investigation extraction. This method utilizes ASTER data to build a hierarchical multi-resolution structure, so as to reflecting the inherent relationship between ground features under various scale levels.

Most of the Land Use Land Cover classifications on a scale level of about 1:100.000 are static snapshots for a single date. They do not, in and out themselves, provide the ability to perform change detection monitoring. Satellite image data has been recorded during the past 30 years from sensors like Landsat MSS, TM and ETM+. Together with data from newer sensors such as ASTER these datasets represent a reliable database for long term change detection monitoring.

Urban areas are the most dynamic region on earth. Their size has been constantly increased during the past and this process will go on in the future. Especially in less developed countries a strong trend towards concentration of people in urban areas can be observed. To understand urban growth remote sensing imagery is a reliable source. A long term growth study for the Phoenix, AZ metropolitan area was carried out on a mid scale level for a period of 30 years. This study used satellite imagery from several sensor systems and an object oriented image analysis and classification approach.

Basing on an ASTER satellite image, land use and land cover classification using an object oriented method has been performed on a study area of 795 km2 located in the central part of Poland. The satellite image has been divided into objects, which have been next classified using membership functions, standard Nearest Neighbour (NN) classification, as well as relations between the objects. 19 classes of land use and land cover have been identified. Object structure of the image after classification enabled generalisation using a working unit of 4 ha (1 ha for water and built-up areas).

Coastal marshes are an important component to the overall health of Long Island Sound. Located in the densely populated northeastern United States, Long Island Sound is subject to the extensive pressures imposed by having such a large population base living on or near its shores. Understanding the health and condition of the wetlands within the Sound plays an important part in protecting this vital resource.

La obtención de clasificaciones digitales de la interfase urbano forestal (IUF) que puedan ser empleadas por los servicios contra incendios forestales en caso de necesidad es una tarea compleja debido a la gran diversidad de elementos que componen estas áreas. Las técnicas de clasificación tradicionales, basadas en rasgos de la imagen a nivel de píxel, presentan ciertas limitaciones, como son la aparición de un característico efecto “sal y pimienta” o su reducida capacidad para extraer objetos de interés.

The tidal wetlands within the Long Island Sound estuary serve a critical role in maintaining the health of the Sound. Over the past two centuries, there has been significant disturbance and loss of tidal wetlands along the Sound due primarily to anthropogenic activities.