Interim ARSF 2008 Calibration Report Introduction This report is into progress made so far (as of 03/10/2008) in analysing the calibration of the Eagle and Hawk sensors carried out in July 2008. This includes both wavelength and uniformity calibration. It should be noted that the analysis is currently incomplete (in that the near-field and far-field datasets have not yet been combined to produce the final calibration, nor has the FODIS calibration been attempted) and therefore conclusions are subject to change. Wavelength calibration The wavelength calibration for both Eagle (fig. 1) and Hawk (fig. 2) appears to be good, with almost all peaks shown in the He and Hg-Ar spectra matching the wavelengths of those expected for the light source exactly (to the nearest waveband centre). There is a caveat in that the calibration curve for 2007 Figs. 1a (left) and 1b (right), showing the spectral peaks visible in the Eagle wavelength calibration against their expected spectral position for the Hg-Ar light source. Error bars show the full width half maximum of the Eagle band whose centre should be closest to the particular spectral peak. Fig. 1a shows the entirety of Eagle's spectrum, Fig. 1b is zoomed in between 735nm and 780nm to make detail clearer. shows one feature at approximately 2300nm that is offset from its apparent equivalent in the 2008 calibration by 12-15nm (see fig. 5b later). This has not yet been explained, though it is possible that it was caused by using a different band configuration for the 2007 calibration. Figs. 2a (left) and 2b (right), showing the spectral peaks visible in the Hawk wavelength calibration against their expected spectral position for the He and Hg-Ar light sources. Error bars show the full width half maximum of the Hawk band whose centre should be closest to the particular spectral peak. Fig. 2a shows the entirety of Hawk's spectrum, Fig. 2b is zoomed in between 1230nm and 1400nm to make detail clearer. Radiometric calibration The full radiometric calibration analysis has not yet been completed. Preliminary results from analysis of the uniformity dataset are available but need to be combined with the far-field and FODIS datasets to give the final calibration. In absolute terms the 2008 Eagle calibration appears to match well with the 2007 data (fig. 3). Calibration multipliers for the Eagle spectrum are within 0.01-0.02 of the 2007 values. As a percentage however this is not quite as good, constituting a 10-15% difference from the previous calibration in some regions of the spectrum (fig. 4). It will be noted from fig. 3 that while the general shape of the calibration curve from 2008 appears to be a good match for that from 2007, the 2007 data both starts and finishes at a slightly higher wavelength than the 2008 data. This has not yet been explained with complete confidence, but is believed to be because of the use of a new band configuration file during the 2008 calibration. The poor matches between the 2007 and 2008 data at the edges of fig. 4 are a result of this mismatch, coupled to a large difference at the blue end caused by the blue filter (or lack of it). Figs. 3a (left) and 3b (right). Eagle calibration multipliers for 2007 (red line) and 2008 (blue and green lines, with and without a blue filter respectively). Fig. 3a shows the entire calibration curve, fig. 3b shows that portion of the curve with a multiplier below 1.5 in order to show greater detail. Fig. 4. Eagle calibration multipliers for 2008 with and without a blue filter (green and red lines respectively) as a percentage of those from 2007. In absolute terms the Hawk 2008 calibration also matches well with the 2007 calibration above approximately 1050nm, with a difference of order 0.001-0.002 (fig. 5). This equates to a percentage reduction of up to 40% (fig. 6), but is small enough that this difference may be noise. However, below 1050nm the 2008 calibration multipliers are much higher than the 2007 ones (up to ~250%), this discrepancy has not yet been explained. If it turns out to be a genuine issue with the sensor then some 2008 hawk data may need to be re-delivered. However, Hawk is known to perform poorly at the extreme short-wave end of its range, so this may be a manifestation of that problem. Figs. 5a (left) and 5b (right). Hawk calibration multipliers for 2007 (red line) and 2008 (green line). Fig. 5a shows the entire calibration curve, fig. 5b shows that portion of the curve with a multiplier below 0.02 in order to show greater detail. Note the slightly offset spectral feature at approx. 2300nm. Fig. 6. Hawk calibration multipliers for 2008 as a percentage of those from 2007. Conclusion From these initial results the 2008 calibration of both Eagle and Hawk (Eagle in particular) appear to be sufficiently similar to that from 2007 to not require redelivery of 2008 data processed using the 2007 calibration. Although the percentage change in the calibration curves from year to year is relatively high, the absolute change is low enough that this may be within error. Further analysis of the shortwave Hawk calibration and of the feature in the Hawk spectrum at ~2300nm is needed to confirm this conclusion, as is combination of these results with the far-field radiometric dataset. Analysis of the FODIS calibration data has yet to be undertaken, since a different light source was used for this and PML are awaiting characterisation data for this light source.
© Copyright 2026 Paperzz