54% of the world population lives in urban areas and a raise is expected to 66% by 2050. Moreover, in a critical scenario, a recent study pusblished in 2017 has shown that urban temperatures could rise up to 7°C by 2100 (Estrada et al, ). This rising is due to global warming but is accentuated in cities by the Urban Heat Island (UHI) effect, which is the air temperature difference between the urban area and its surrounding rural areas. In remote sensing, the key-parameter is the land surface temperature which is indirectly retrieved from radiance measurements in the infrared domain. TRISHNA (Thermal infraRed Imaging Satellite for High-resolution Natural ressource Assessment) is a joint mission between both CNES and ISRO (National Space Center and Indian Space Center respectively). TRISHNA is expected to be a multispectral mission with spectral bands in the VNIR and TIR ranges with a high temporal repetitivity (1 to 3 days) and a high spatial resolution (50m in the TIR range). The project CATUT aims to quantify the surface temperature accuracy from simulated spaceborne measurements for the TRISHNA instrumental and spatial characteristics for urban applications. In order to do so, the processing tool STRASSS (Surface Temperature Accuracy Retrieval With Spaceborne Sensor Simulations) has been developed : it computes TOA (Top of Atmosphere) radiances from DESIREX airborne measurements over the Madrid city (4-m spatial resolution for the Airborne Hyperspectral Scanner), performs spatial aggregation to 60 meters, adds the instrumental noise and retrieves the surface temperature with two algorithms called Split-Window and TES (Temperature Emissivity Separation). The Split-Window is performed over two spectral bands between 10 and 12 µm and the TES is performed for three or four spectral bands between 8 and 12 µm. Both Split-Window and TES have been calibrated on a specific database : the algorithm error is 0.97 K for the Split-Window and 0.013 or 0.012 for the emissivity for the TES algorithm (three or four spectral bands respectively). The land surface temperature retrieval accuracy is given according to the spatial resolution from 4 meters to 60 meters (4 meters, 8 meters,16 meters, 32 meters, 60 meters). A comparison with a 7-band TES applied on the airborne measurements is performed (Rosa Oltra-Carrio, 2014) as well as a comparison with in situ measurements from validation sites. This study shows the degradation of the accuracy according to the spatial resolution and helps considering unmixing methods to improve the land surface temperature retrieval.