Wildfires are a yearly recurring phenomenon in many regions of the world. They partially or completely remove the vegetation layer, and as such they alter ecological processes at different scale levels. Moreover, wildfires vouch for a significant emission of trace gasses in the atmosphere. The Mediterranean basin is especially sensitive to wildfire occurrence. This results from both the human pressure on the environment and the Mediterranean climate, which is characterized by dry and hot summers. Throughout time Mediterranean ecosystems have learned how to cope with fire. The classic example of a fire-adapted ecosystem is Mediterranean shrub land. Although these ecosystems respond to fire, recent research has shown an increase in fire frequency in the northern Mediterranean rim. This increase is mainly contributed to changes in land cover which are interlinked with the recent socio-economic evolution of northern Mediterranean counties. It is however, recognized, that recent climatic change also plays a role. Improved knowledge on (changing) fire regimes is of paramount importance for ecologists, resource managers and policy makers. In this context, fire/burn severity has become a parameter of increasing interest. Fire severity is defined as the degree of environmental change caused by a fire, as measured immediately post-fire. Burn severity incorporates both the direct fire impact and ecosystem responses.Due to its synoptic nature, remote sensing can provide a valuable alternative for expensive field campaigns to assess these changes. In this research three Landsat Thematic Mapper (TM) and multi-temporal Moderate Resolution Imaging Spectroradiometer (MODIS) satellite images were used to assess post-fire effects of the large 2007 Peloponnese (Greece) wildfires. TM images provide spatial detail (30 m) at the cost of a 16-day revisiting time, whereas MODIS imagery allows temporal detail as they have daily coverage, however, with a low spatial resolution of 500 m. Additionally, field data of burn severity were acquired one-year post-fire, in September 2008. The applied field protocol was the Geo Composite Burn Index (GeoCBI). The CBI is a semi-quantitative approach that combines several post-fire factors, as visually estimated in the field, into one value which is indicative for burn severity. The 2007 Peloponnese fires consumed more than 150 000 ha of agricultural land, shrub land and forest. The fires were the worst natural disaster of the last decades in Greece, both in terms of human losses and the extent of the burned area. Also, some world heritage monuments, such as Ancient Olympia, suffered severe damage.The physical basis of optical remote sensing (400-2400 nm) relies on the characteristic spectral signatures of terrain features. Spectral indices combine reflectance data of different wavelengths. As such they form a conceptually easy tool to estimate vegetation characteristics and their changes. The main goal of this dissertation was to evaluate the potential of these spectral indices for assessing fire/burn severity. This was done by fulfilling four objectives. Firstly, three TM spectral were correlated with field data of severity for this case study in a Mediterranean ecotype. These spectral indices were the Normalized Difference Vegetation Index (NDVI), the Normalized Difference Moisture Index (NDMI) and the Normalized Burn Ratio (NBR). The NBR yielded the best results. In addition, indices were evaluated by means of the optimality concept which considers a pixel’s pre-/post-fire trajectory in a bi-spectral feature space. Results of the optimality analysis corroborated with the outcomes of the correlation with field data indicating the NBR as the best spectral index to assess burn severity.Secondly, illumination effects on the differenced NBR (dNBR) were revealed and an illumination correction method was proposed. Scene illumination is determined by topography and solar position at the moment of image acquisition. This influences any object’s reflectance behavior and the reliability of change detection. Fire severity was more reliably detected when pixels were well illuminated and the difference in illumination between the two images was small. To overcome this problem, images should be corrected for topographical effects. The best results were obtained by applying a modified topographic correction method that, in contrast with traditional correction methods, normalizes reflectance towards a maximum illumination condition instead of normalizing towards an intermediate illumination condition.Thirdly, the importance of the role of assessment timing on a dNBR assessment was verified. Due to the limitations in Landsat available imagery, this part made use of MODIS time series. For each burned pixel a control pixel was retrieved based on time series similarity and spatial context. The control pixels estimate how burned pixels would have behaved in the case of no fire. As such the control pixel selection procedure allows a reference in the study of the temporal dimension. Differing lag timing, i.e. time since fire, of an assessment can result in significantly different dNBR statistics, especially in quickly recovering ecosystems where resprouters strongly mitigate first-order fire effects by time. Seasonal timing, on the other hand, greatly determines what is actually been measured. Inappropriate seasonal timing falsifies trends in post-fire effects. This section urges for awareness for the temporal component of dNBR studies and associated terminology. In particular, this is crucial for studies that wish to compare fires across space and time.Finally, a multi-temporal burn severity approach was presented based on MODIS time series. The method accounts for both the direct fire impact and vegetation recovery processes. The multi-temporal dNBR (dNBRMT) is defined as the one-year integrated difference between the NBR of burned pixels and their associated control pixels. As such dNBRMT estimates are indicative for the change in vegetation productivity. The multi-temporal method correlated reasonably well with the Landsat dNBR. The traditional Landsat dNBR is, however, superior for spatial detail. This traditional approach is limited by Landsat’s low revisiting time, cloud cover and image-to-image normalization constraints. The time-integrated approach, therefore, present a valuable alternative for burn severity mapping at a continental to global scale. In addition, the method has big potential to enhance comparability of burn severity across space and time.It was shown that spectral indices contain valuable information with regards to fire/burn severity. To extract a maximum degree on information, however, preconditions as illumination and assessment timing should be considered. This work can have important implications for operational dNBR mapping at a regional to global scale. It is, however, recognized that the dNBR is not a perfect proxy of fire/burn severity. Further research is required to realize the full potential of multi-spectral imagery to assess post-fire effects.