Compared with conventional narrow-band-scanning absorption spectroscopy, hyperspectral absorption spectroscopy shows great technical advantages thanks to more abundant absorption information covering a wider spectral range. In practical hyperspectral absorption applications, the laser source can usually be tuned within a wide spectral span (>20 nm). However, full-span tuning puts very high data acquisition, storage, and processing requirements, leading to a high system cost and mammoth workload. Consequently, scanning band selection with proper bandwidth becomes crucial for hyperspectral absorption applications. Here, a numerical method for scanning band selection is developed based on the frequency-dependent lower-state energy. And favourable scanning bands around 1.3 and 2 μm wavebands are selected. For the 1.3 μm waveband, the selected bands are consistent with those exploited in publications, validating the feasibility and correctness of the method. For the 2 μm waveband, the superiority of the selected bands in temperature measurement is verified with experimental data recorded at given operation conditions. Advantageous scanning bands around 2 μm are provided concerning the wide operation condition in practical applications. Calculations demonstrate that proper scanning bands exist at the 1.8 μm short and 1.9 μm long waveband, with the optimum at 1.86 μm. A larger scanning bandwidth may not result in the best temperature measurement accuracy, but generally, it brings about a smaller and more controllable uncertainty over the whole spectrum. These findings are instructive for field applications of hyperspectral absorption spectroscopy in engine combustion field diagnosis.