OPTIMAL LAND SURAFCE TEMPERATURE VALIDATION SITE IN EUROPE FOR MSG P. Dash 1 , F.-S. Olesen 1 and A. J. Prata 2 Institute of Meteorology and Climate Research, Forschungszentrum Karlsruhe, Germany 1 CSIRO Atmospheric Research, PMB 1 Aspendale Vic 3195, Australia 2 ABSTRACT Measurements utilising satellite-sensors and ground-based point measurements have different bases and dissimilar resolutions. Many variables, e.g., land surface temperature (LST), and emissivity, cannot be easily measured at the ground at a scale corresponding to that of the sensors. This makes a direct comparison of ground-based and satellite-based measurements not feasible. Consequently, it is difficult to determine the accuracy of LST and emissivity estimation models from field-data and, often, it is compelling to validate the method using synthetic data first. However, comparisons of ground-based measurements versus satellite- based measurements over homogeneous areas are necessary for ultimate validation of satellite based products. The definition of this “homogeneous area” depends on various factors such as surface type, scale of measurement of the sensor, temporal variation etc. The present work is an effort to identify potential locations for field measurements, within Europe, to validate LST measurements from Meteosat Second Generation (MSG) Spinning Enhanced Visible and Infra Red Imager (SEVIRI) data. For identifying homogeneous areas, statistical analysis is performed on AVHRR data (~ 1.1 km resolution) for 25 x 25 pixel sliding-windows over 8 months (March 2002 - October 2002). In addition to the pixel-by-pixel approach, an image-object based approach performing segmentation and subsequent classification based on class description is used. This method uses the properties of groups of pixels and inter-group relations, rather than treating each pixel individually. The homogeneous areas can be sites for validation of land parameters derived from Meteosat Visible and Infrared Imager (MVIRI) and SEVIRI. However, the homogeneity criteria must be relevant for both the satellite spatial scale and the ground measurement scale, i.e., the measurements at the identified site must represent measurements from satellite for the surrounding area. The final selection requires investigation with high spatial resolution data and field investigation. 1. INTRODUCTION For radiometric measurements, land surface temperature (LST) is defined as the “surface radiometric temperature” corresponding to the instantaneous field-of-view of the sensor (Prata et al., 1995). A similar concise definition states: LST is the “ensemble directional radiometric surface temperature” (Norman and Becker, 1995). The term “ensemble” depicts the bulk contribution of an inhomogeneous pixel. For large areas, LST can only be derived from surface-leaving radiation measured by satellite sensors. As they represent the integrated effect of the surface they are better suited for many applications in comparison to point measurements, e.g. in Earth’s radiation budget. To estimate LST from space radiometry, three main effects have to be considered and corrected for: atmospheric, angular, and emissivity effects. Apart from angular effects caused by limited view-angles and atmospheric attenuation, simultaneously unknown emissivity and LST pose an underdetermined problem. For resolving this under-determination, additional