|
We present a simple, spreadsheet-based model to examine the effects of the spectral response functions of individual
instrument bands on their measurements of top-of-the-atmosphere radiances. The model uses spectral radiances at 1 nm
resolution from the near ultraviolet to the shortwave infrared at wavelengths from 300 nm to 2500 nm, convolving them
with the spectral responses of the bands to calculate band-average spectral radiances. For on-orbit calibration purposes,
the model uses nominal solar irradiance and lunar albedo spectra to provide saturation, diffuser, and lunar radiances for
the bands. For prelaunch calibration purposes, the model uses a 2850K Planck function, normalized to a maximum
value of unity, to approximate the spectral shape from a laboratory integrating sphere source. For Earth-exiting radiances,
the model uses nominal radiance spectra over a blue ocean, a desert, and a grassland. These spectra are provided
with the effects of atmospheric trace gas absorption removed. In addition, the model includes a trace gas transmittance
spectrum that can be modified as a function of airmass. Currently, a spectrum with an airmass of 2.4 is used. In the
model, this transmittance spectrum is combined with the three Earth-exiting radiance spectra to provide top-of-the-atmosphere
radiance spectra both with and without trace gas absorption features.
Here we use the model to investigate three types of spectral response features. The first study involves the out-of-band
response from one of the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) bands. Using the model, we demonstrate a
technique to correct for the effect of that response on measurements of Earth-exiting radiances. The second study shows
the effect of in-band spectral differences in an instrument band with multiple detectors. In this example, the effects are
small, but differ with the type of Earth scene and with the amount of atmospheric trace gas absorption. On-orbit corrections
for portions these detector-to-detector spectral differences are possible. However, at some level these differences
will cause a residual striping in the band's measurements that cannot be removed. The final study examines measurements
by a proposed multispectral grating-based spectrometer of the wavelength region near 760 nm, where there is a
substantial absorption feature from atmospheric oxygen. Based on the bandwidth and wavelength spacing of the instrument's
bands, we investigate the use of the absorption feature to provide a wavelength calibration for the instrument.
This model provides a tool for use in the design of new satellite instruments. In addition, it is possible to use the
model to help mitigate the effects of actual spectral response features in instrument bands after those features are
revealed during prelaunch characterization.
|
