ROLO Data Reduction Procedures
ROLO procedures for removing detector artifacts are straightforward. For the VNIR CCD, the bias level is measured from extra pixels in the CCD readout row. Dark current is determined as a function of exposure time from fitting unilluminated images acquired at the start and end of each observing night. For the SWIR camera, the detector is clocked out continuously, and the last frame before exposure is saved, to be used for detector readout characterization. Additionally, a dark frame image is acquired as part of each SWIR 9-filter sequence, to augment the twice nightly dark exposure time series. Linearity corrections for both detectors have been developed from special sets of observations. Flatfield images are acquired 2-3 times each lunation by viewing a spectralon plaque illuminated by a 1000 Watt FEL lamp. These lamp images are processed to remove detector artifacts, then normalized to generate flatfield correction images.
Atmospheric Extinction Correction
Correction of ROLO lunar radiance measurements to exoatmospheric values is based on the extinction observed in the star images acquired through each observing night. The observed stellar irradiances are fitted to an extinction model that is formed from MODTRAN atmospheric transmission spectra, and thus is coupled across all ROLO bands. The fitting algorithm produces a set of parameters that scale the individual transmissions as a function of time and zenith angle through the night. These parameters are used in turn to generate extinction correction values for the Moon images. Atmospheric correction is one of the most important aspects of ROLO data reduction, and the processing algorithm is undergoing continued refinement.
The stellar extinction fitting procedure handles all the star observations acquired in a night as an ensemble, typically ~12 stars imaged 10-15 times in all 32 bands. Atmospheric transmission spectra for 6 separate constituents: “normal gases” (N2, O2, CO2 and trace gases, plus Rayleigh scattering), water vapor, ozone, and three aerosols, are convolved with the ROLO spectral response functions to develop extinction coefficients for each constituent on each band. These are scaled by a “relative abundance” factor and summed to give an effective optical depth for a given slant path length (airmass). The relative abundances are the free parameters of the fit; they are allowed to vary independently among the constituents, but are coupled across all the bands. The abundances are time-dependent, modeled by a 2nd-order Chebyshev polynomial, thus there are 3 free parameters for each of the 6 constituents. In practice, additional parameters are included to model time dependent instrument gain effects as well. For each lunar image, an extinction correction factor is constructed from the fitted parameters and the time and zenith angle of the image.
Currently, the absolute scale of ROLO radiance data is tied to astronomical photometry of the star Vega. Published absolute energy density measurements were used to scale a synthetic spectrum for Vega, which was then convolved with the ROLO instrument spectral response functions to give the photon flux in each band. A dedicated reprocessing of all ROLO images of Vega produced exoatmospheric instrument response values, which were ratioed to the absolute fluxes to form the calibration.
In an effort to reduce uncertainties in the ROLO absolute calibration, an alternate method is being developed with involvement of the radiometry group at NIST. A large, on-axis collimated source that overfills the ROLO telescopes' field of view was designed and built at USGS. At the collimator focus is a uniform source that has been calibrated at the NIST FASCAL facility. This source is monitored via transfer to a NIST sphere source during field calibration work at ROLO. This is current work in progress.
ROLO Metadata Table
The final step in ROLO data processing is construction of a table of parameters for each lunar image, extracted from the image cube header information. This includes all observation geometry and ephemeris information, the extinction correction and calibration factors, the disk-integrated irradiance and sky background level (in instrument units), and error values for all derived quantities. The table is ordered chronologically by the acquisition time of each band. This table forms the basis for inputs to the lunar irradiance model.
The input data for spatially resolved lunar radiance modeling are the processed, calibrated, and ALEX-projected (see Overview) lunar radiance images. Since each image contains roughly 250,000 pixels on the lunar disk, and there are over 65,000 processed images, the processed archive was re-sampled for handling within typical computer memory capacity. The images were separated into wavelength bands and subsampled, 3 lines by the full 576-pixel image width. Cubes were built up of the subsamples, with each new band corresponding to a processed image. These transposed radiance data are called ROLO “bricks”.
U.S. Department of the Interior | U.S. Geological Survey | U.S.G.S. Astrogeology
Top | Home | Search | Disclaimers & Privacy | Special Printing Instructions
Send Science Questions to: email@example.com
Send Website Questions to: firstname.lastname@example.org
Last Modified: Fri, 05 Dec 2014 08:33:25 -0700
- - -
The USGS is committed to providing valid and Section 508 compliant content.
If you feel that this page does not meet any of the standards, please email at email@example.com.