A simple quadratic method of absorption feature wavelength estimation in continuum removed spectra

A simple quadratic method is proposed for estimating the wavelengths of absorption features in the short wave infrared spectral region from hyperspectral instruments ranging from 2 to 18 nm sampling intervals. The method uses a direct substitution of the three spectral reflectance bands to achieve an estimate of the wavelength of the point of maximum absorption. The method is shown to be stable across different sampling regimes and therefore across different hyperspectral instruments. The results show it is possible to accurately estimate the wavelength of spectral absorption features to an RMSE of ± 3.8 nm in the 2000–2500 nm spectral region over multiple sampling intervals and resolutions in sensors with signal-to-noise greater than 500:1.

► Accurate method of estimating absorption wavelength in the SWIR.
► Consistent across different sampling regimes e.g. different instruments.
► Is robust in the presence of lower SNR values when feature depth is greater than 0.04.
► Approaches RMSE values of high SNR instruments as absorption feature depth increases.
► Direct substitution allows fast calculation speeds.