You are here

Warning message

Attention! This event has already passed.

Thermal subpixel estimation in urban areas with spaceborne sensing

Friday, 10 June, 2011 - 10:30
Essa Wiesam
phd defence

Openbare doctoraatsverdediging van Wiesam A. A. Essa voor het behalen van de graad van Doctor in de Ingenieurswetenschappen met het proefschrift getiteld: 'Thermal subpixel estimation in urban areas with spaceborne sensing'.

Promoter: Prof. O. Batelaan
Co-Promoter: Dr. J. van der Kwast


Downscaling land surface temperature from thermal spaceborne imagery is considered as a rather new and challenging research problem for urban areas while this research problem has been well investigated for vegetated areas using different approaches. Consequently, no methodology is presently available for bio-physical surface characterization such that downscaling land surface temperature in heterogeneous urban areas is feasible. This work therefore contributes to a better understanding of the thermal physical processes acting in urban areas. Adaptation and improvement of the DisTrad method, commonly used in vegetated areas, was applied to downscale land surface temperature over urban areas. In this study, the urban zone of Dublin city (Ireland) was selected because it shows a sufficient range of spectral and thermal remote sensing information. The first part of this thesis considers the role of 16 different spectral indices: BASVI, R, B, NDWI, NDBaI, SVI, SAVI, NDBI, NDSI, I, UI, FC, VC, V, IBI and NDVI as possible input in DisTrad. The relationship of each index with observed remotely sensed land surface temperature was used to investigate the most suitable input parameter for DisTrad in an urban environment. The impervious percentage (I) shows the best downscaling results for the urban land surface temperature patterns to higher resolutions compared to the originally used NDVI in vegetated areas. The second part investigates the feasibility of downscaling potential sources of thermal data imagery such as Moderate Resolution Imaging Spectroradiometer (MODIS). Simulation of MODIS land surface temperature images was necessary, since no coincident high resolution images were available for evaluation of the downscaling results. The adapted DisTrad method using the relation between land surface temperature and impervious percentage shows improved downscaling results compared to the original implementation of DisTrad for all evaluated resolutions. Since this study was carried out using simulated data, it is likely that larger errors can be expected when applying the procedure to real MODIS data, because of additional noise added to the temperature data particularly from sensor registration. Future studies should address this issue. The third part discusses the utility of three types of MODIS/Terra land surface temperature image products (MOD11_L2, MOD11A1 and MOD11A2) that are obtained in two different ways V4 and V5 for downscaling to different target resolutions of 60, 90, 120, 240 and 480 m. An improved version of DisTrad was proposed in which a better parameterization of the original DisTrad temperature residuals in urban areas is obtained. The improved DisTrad proofs to have a stronger capability to downscale land surface temperatures in urban areas than the original DisTrad. Validation of the downscaled products using original DisTrad, the improved DisTrad and the uniformly disaggregated land surface temperature images (UniTrad) are performed by comparative analysis with a time-coincident Landsat 7 ETM+ thermal image. Statistical results indicate that the improved DisTrad method is more robust than the original DisTrad or the UniTrad.