PDS_VERSION_ID = PDS3 RECORD_TYPE = STREAM OBJECT = TEXT PUBLICATION_DATE = 2007-09-26 NOTE = "Description of contents of RAWDATA directory" END_OBJECT = TEXT END We have included the raw instrument data for those who desire to check the supplied corrected reflectances and/or apply a different algorith to remove the effects of scattered light. This data is in 6 columns: incidence angle (from nadir), emission angle (from nadir), azimuth angle (see notes in the geometry section of the archive), phase angle, raw instrument voltage (volts), and normalized reflectance (raw instrument voltage divided by the instrument voltage of a spectralon sample illuminated at i=0, e=5 deg, rounded to 5 significant figures). You will note that these files have a number of additional data points not included in the corrected files. The extra data is at the end and consists of a number of measurements at i=0. The purpose of these measurements is to measure the magnitude of scattered or secondary light on the overall reflectance of the sample. We found that light scattered from the sample illuminates the stage mechanism and is therefore visible to the detector. As the stage rotates in azimuth, more or less of it is visible to the detector. The best case is at az <= 0 where virtually none of the sample stage is visible to the detector. This is our baseline reflectance. As the stage rotates in azimuth, more of it becomes visible to the detector until az~90 deg where the secondary reflection peaks. The effect can be observed and measured by holding incidence angle constant, emission angle constant, and rotating in azimuth. There should be no change in reflectance, but we observe small changes as described above. At larger emission angles, the effect is more pronounced because the projected area of the stage is larger while the projected area of the sample is smaller. Although these effects are small (<10%), they need correction because they are systematic. To correct for secondary reflection, we take a set of measurements at i=0, e=constant (for this example, say 40 deg), and different azimuth angles, including a few 'negative' azimuths. We take the azimuth with the lowest reflectance to be our baseline (typically az = 0, or one of the two 'negative' azimuths when little or no stage is visible to the detector) and divide the other reflectances at i=0, e=40 by this baseline value. This gives us a reflectance normalized to one which we assume has no scattered light. We plot these normalization constants versus azimuth and fit a 3rd order polynomial to them. We repeat this for the other emission angles. We then generate a look-up table of normalization constants for any integer emission and azimuth angle by also interpolating between constant azimuth angles and varying emission angle. Now we take all our measurements (other than these latter ones at i=0) and divide those reflectances by the normalization constant at heir respective emission and azimuth angle. For example, suppose we find the detector voltage (proxy for reflectance) at i=0, e=40, az=0 to be the lowest at 0.2250 V. When observing at i=0, e=40, az=90, we find the detector voltage to be slightly higher at 0.2277 V. Our lookup table will show the normalization constant for this geometry to be: 0.2277 / 0.2250 = 1.012. This means the reflectance at this geometry was 1.2% higher than when az = 0. We measured the reflectance of our sample at i=30, e=40, az=90 to be 0.0456. We correct this reflectance to 0.0456/1.012 = 0.0451 to remove the stray light. This value is then multiplied by cos(i)/cos(e) to convert it to radiance factor as explained in the bugmars_ds.cat. This algorithm is not perfect because we have not include variations in incidence angle in the scattered light correction. However, we would only expect this to be a problem if the sample has unusual scattering characteristics such as an increasing side-lobe component with increasing incidence (the most visible part of the stage - its back - is always 90 deg from the incident arm plane). For the more likely cases (i.e. physically realistic for powdered samples) of an isotropic or a strong forward or back-scattering surface, this algorithm should be sufficient. For the user who desires to look at the radiance factor of the sample WITHOUT the secondary reflection correction, you may do the following. Take the last column of the raw data file (raw instrument voltage divided by the instrument voltage of a spectralon sample illuminated at i=0, e=5 deg, rounded to 5 significant figures), multiply by cos(i)/cos(e).