Interpolation errors in UV-visible spectroscopy for stratospheric sensing: Implications for sensitivity, spectral resolution, and spectral range

UV-visible measurements of stratospheric constituents require the ratio of a pair of spectra to be determined. If their wavelength calibrations differ and if an array detector is used, at least one spectrum must be interpolated. This introduces error if the spectrum is undersampled; the error is smaller if wavelength stability is good. Increasing the sampling ratio by making the spectral resolution poorer reduces the optical depths of absorption by constituents. Exact values of interpolation errors from real spectra are a difficult topic, but with a theoretical study with a simulated spectrum we show that the sampling ratio should exceed similar to 4.5 pixels/FWHM but need not exceed 6.5 pixels/FWHM. To avoid significant reduction in the optical depth of NO2, the resolution should be smaller than similar to 1.0 nm FWHM. Hence a spectrometer system that measures both OClO and NO3 by observing one order from one stationary grating should have more than similar to 1500 pixels, more than many currently available array detectors. (C) 1996 Optical Society of America