The retrieval and application of emissivity spectrum and temperature are key issues in thermal infrared remote sensing. Thermal airborne hyperspectral imager has 32 bands from 8 μm to 11.5 μm, which can provide abundant useful information for the retrieval of emissivity spectrum and temperature. This paper establishes regression between MMD and β min using 274 laboratory refl ectance and fi eld emissivity spectra, analyzes its accuracy using the data fi eld measuring, and evaluates urban surface diurnal temperature range. The result shows that: (1) the average absolute difference between the temperature of retrieval and the measurement is 1.8 K, and the relative difference is 0.59%; (2) the mean difference between the broadband emissivity and the mean emissivity from TASI is 0.036, and the standard error of difference is 0.032. Because of the effects of scale and atmosphere, the predicted value does not equal to the laboratory measurement, but the pattern of predicts is similar to the laboratory measurement; (3) the result of diurnal temperature range is reasonable. Therefore, the accuracy of this method can satisfy operational application and it is feasible to retrieve the emissivity spectrum and temperature for TASI data.