NEDT influence in the thermal band selection of satellite-borne instruments
This work focuses on the impact of noise in the channel configuration of thermal sensors, in two aspects. The first one is related to the selection of an optimal band configuration to calculate accurate land surface temperatures (LSTs). The second one addresses the impact of noise level on the absolute error in temperature estimates, for the chosen band combination. These issues are studied beyond the analysis performed in previous works, where spectrally constant low-level noise values were considered. The study is based on the comparison of LST errors yielded by different potential channel configurations. The sensitivity analysis is performed using a quadratic split-window model regressed by least-squares techniques on a database containing LST and brightness temperatures for a wide range of atmospheric and surface conditions. In this analysis all sources of uncertainty are taken into account, and the influence of the procedures used to remove the effects of the Modulation Transfer Function is also considered. The results reveal that: (i) the Noise Equivalent Difference of Temperature (NEDT) should not be higher than ±0.3 K at 300 K within the 10-12.5 m window, and ±0.2 K at 300 K within the 8-9.5 m spectral region, to assure an accuracy better than ±1.2 K in LST, and (ii) the combinations (10.1-10.8)/(11.7-12.4), (10.1-10.8)/(11.8-12.5) and (10.1-10.8)/(11.5-12.5) m on the one hand, and (8.0-8.5)/(8.5-9.0) m on the other hand, are the optimal configurations within the two specified windows, respectively.