The comparability of aggregated emissivity and temperature of heterogeneous pixel to conventional tes methods

Unmixing in the reflective domain is a well known technique to retrieve materials and their abundances from hyperspectral imaging scene. However, for the thermal domain (8-14 μm) only few unmixing models were developed due to the complexity of retrieving three unknowns (i.e. materials abundance, temperature and emissivity) from the pixel radiance. Attempt to solve this complexity is using aggregation models. The aggregation models are physics based models which estimate the pixel radiance with the contribution of acquisition parameters and scene description including the temperature and the emissivity of materials. This paper focuses on the comparison of temperature / emissivity estimated using inhouse aggregation model with two conventional Temperature and Emissivity Separation (TES) methods: Maximum Minimum Differences (MMD/TES) and SPectral SMoothness method (SpSm). This comparison is achieved using synthetic mixed pixels of two sensors AHS-160 and SEBASS. The results show that below a temperature difference of 30 K, the aggregation models are in good adequacy with the TES estimations.