Calibration of NIS data ======================= Calibration of NIS data follows a standard step-wise process originally defined in [WARRENETAL1997] and modified by experience with in-flight data. The calibration pathway follows photons in reverse from the detector to the source of reflected light (e.g., a geologic surface or caltarget). The calibration equation for converting raw spectral signal in counts or data numbers (DN) to physical units of radiance (W m-2 sr-1 micro m-1) has the form: Radiance(l,t) = DN(l,t,T,M,s,i) { { --------------- - Dark(l,t,T) } - Xtalk(l) } * BulkCor Obs ----------------------------------------------------------- (1) Gain(l) BulkCor = Mir(l,M,s) * Phot(l,i) * Resp(l,s) * Pol(mir) * Ecor (1a) The variables are defined below, and their values are given in appendices A and B. DN(l,t,T,M,s,i) Raw analog-to-digital count level at each detector channel of wavelength l at time t Obs Number of one-second observations summed to make a single spectral measurement Dark(l,t,T) One-second detector dark signal, at time t close to the time of the data spectra, including possible bias induced by instrument temperature T Gain(l) Corrects high-gain (10x) Germanium channel data to low (1x) Xtalk(l) Correction applied to the first five Ge channels from second order light Mir(l,M,s) Correction factor for scan-mirror position M at slit position s Phot(l,i) Photometric correction for caltarget observations only at solar illumination angle I Resp(l,s) Per-channel DN-to-radiance conversion factor determined on the ground using slit s (narrow or wide) Pol(m,t) Change in detector response due to polarization effect on some mirror positions Ecor Empirical correction to remove nonvariant high-frequency noise in spectrum [IZENBERGETAL2000] (see Table IV) detail the evaluation of variables used in the NIS calibration, from ground tests and operations in space. More details on the ground calibrations are presented in [WARRENETAL1997]. The following table outlines the NIS calibration process, and lists possible sources and effects of errors. The steps in the pathway are discussed in the following sections. NIS Calibration Pathway ======================= Step: Step Vars: Variables Outp: Output; intermediate .filetype ESrc: Error source EffE: Effect of Error Step: 1. Start with Raw data Vars: l (wavelength) Outp: - ESrc: - Effe: - Step: 2. Spectral Averaging Vars: # spectra per observation Outp: Raw DN; .rfd ESrc: Incorrect # spectra per observation EffE: Values off by factor Step: 3. Dark Subtraction Vars: l,T Outp: Dark subtracted DN; .dks ESrc: Dark spec. temporally distant from target spec.; Motor bias Effe: Variable, greater in longer l Step: 4. Gain correction Vars: l,Gain Outp: N/A (internal step) ESrc: 1x or 10x Gain Effe: Systematic error in Ge values Step: 5. Crosstalk correction Vars: Ge l, Corresponding InGaAs l Outp: N/A (internal step) ESrc: Coefficients for 2nd-order light Effe: Error in lower Ge l Step: 6. Mirror pos. correction Vars: l, Mirror pos. Outp: Calibrated DN ESrc: Wrong mirror pos. or coefficients Effe: Variable Step: 7. Photometric correction for caltarget Vars: l, Off-Sun angle Outp: Calibrated DN (caltarget obs.) ESrc: Improper model; s/c pointing error Effe: Poor correlation of caltarget obs. Step: 8. Radiance calibration Vars: l, Slit Outp: Radiance; .rad ESrc: Artifacts/inaccuracies in coefficients Effe: Propagated artifacts in spectra Step: 9. Radiance Factor conversion Vars: l, Solar distance Outp: Radiance factor (I/F); .iof ESrc: Incorrect solar distance/solar model Effe: Non-realistic radiance factor Step: 10. Empirical correction Vars: none Outp: Radiance factor (I/F); .iof ESrc: Variation in effect Effe: Added nose Step: 11. Polarization correction Vars: l, Mirror position Outp: Radiance factor (I/F); .iof ESrc: Incorrect polarization effet Effe: variable 1. Raw NIS Data Raw NIS data are represented in counts or data numbers (DN). For each channel, a DN corresponds to one least significant bit from the 12-bit analog-to-digital converter of the individual detector and represents a certain number of photons of the appropriate wavelength passing though the slit. A DN has a value of approximately 1.22 mV output from the detector pre-amplifiers from the InGaAs channels and the Ge channels when at gain 1x (0.12 mV at gain 10x). [WARRENETAL1997] discuss the electronics governing detector response in greater detail. 2. Spectral Averaging The basic unit of spectral data for any given NIS channel is the integrated DN accumulated by each detector element. For any single observation, up to 63 one-second integrations can be summed. To get the average raw signal per second for each channel, the total signal is divided by the number of seconds in the observation. Available proxies for the standard deviation of the NIS signal include noise values derived from proximate dark observations and reference in-space sequences of dark spectra obtained during cruise ([WARRENETAL1997] and [IZENBERGETAL2000].) 3. Dark Subtraction Dark data are spectra taken with the slit closed, or open and looking into empty space. The signal in a dark spectrum represents the background electronic and other noise inherent in the instrument (e.g. the voltage applied to the NIS shutter motor) and is removed as part of the calibration process. Dark spectra are derived from interleaved darks and spectra, bounding observations before and after spectra, or as a running fit of darks between many data observations. Dark spectra are subtracted from the averaged target spectrum. The standard deviation of the dark data is a representation of the noise for the associated spectra in the sequence. 4. Gain Correction The Germanium detector has a selectable gain of 1x or 10x. If the gain is set at 10x, then the Ge signal needs to be corrected to 1x in order to use radiometric coefficients generated for 1x gain. The observed effect on the Ge gain ratio in-flight vs. on-ground is approximately a 0.9% decrease. This variation does not affect the relative channel-to-channel response and is within the noise when comparing Ge and InGaAs spectra. Currently, NIS calibration uses the in-space-derived coefficient for 10x gain. TheGain correction factor is 9.843 +/- 0.030. 5. Spectral Crosstalk Correction Spectral crosstalk in NIS results from first-order wavelength light falling onto and being detected by the shortest-wavelength Ge detector elements. The first five Ge channels have measurable crosstalk signals, which can be removed by subtracting the proper percentage of the corresponding InGaAs signal from the Ge signal. The coefficients for the crosstalk correction are given in Table A-I. 6. Mirror Position Correction The NIS scan mirror gives the instrument a 140 degrees wide, selectable field of view with 350 distinct mirror positions. Varying mirror position affects the light coming into the detectors in a wavelength-dependent manner as a result of polarization and reflection efficiency variations. An interpolated curve-fit correction for each channel based on mirror position is applied to NIS data as part of the calibration process. Equations and coefficients describing relative response as a function of mirror position are shown in Appendix B. 7. Caltarget Photometric Correction The NIS caltarget consists of a roughened surface coated with a smooth deposit of gold (Infragold(tm)). The target is fixed and viewed by the instrument at a constant emission angle, e, of 75 degrees (mirror position 0), with 5 degrees of yaw, and with an incidence angle i = 50 degrees. In caltarget observations, small differences in observing geometry caused by small variations in spacecraft pointing are compensated by a photometric correction. The equation for the correction is: RRS(l,yaw) = exp( -0.15764 * sqrt(30 - yaw)} (4) where RRS is the reflectance relative to specular geometry for wavelength l at the given spacecraft yaw off-Sun. The photometric correction is the ratio of the RRS (at a yaw of 5 degrees) with the RRS at the actual yaw angle of the caltarget ([WARRENETAL1997]). 8. Radiance and Reflectance Conversion Calibration of NIS data to absolute radiometric units enables quick and easy comparison with other spectroscopic datasets, essential to creating complete spectral/spatial data cubes for mineral identification and characterization. The data are converted to radiance (W m-2 sr-1 micrometer-1) using APL OCF-determined tables (Appendix A). The radiance factor (I/F; [HAPKE1981]) of the area targeted by NIS is calculated as the ratio of the target radiance observed by NIS to the radiance of a solar-illuminated Lambertian surface of albedo = 1 observed at 0 degrees phase angle and located at the same distance from the Sun. For generation of NIS I/F spectra, we used the World Meteorological Organization 1 AU top-of-atmosphere solar flux spectrum [WEHRLI1986] convolved to the NIS bandpasses and scaled by the square of the heliocentric distance of the NIS observations. 9. High Frequency Noise Correction This corrects a channel to channel systematic variation that is constant throughout the orbital mission. The smoothes the spectrum considerably by removing an invariant noise signal. 10. Polarization Correction Data at certain mirror positions (usually high positions e.g. low phase flyby) are affected by the polarization properties of the NIS Mirror. A polarization correction is implemented to solve this. Calibration Error Sensitivity See [IZENBERGETAL2000] for a discussion of possible effects of calibration error by source. Validation of Calibration Coefficients Validation of the absolute calibration was performed through observations of the Earth and Moon and of Eros itself during the initial flyby (see [IZENBERGETAL2000].) NISCAL The calibration of NIS described in this paper has been implemented in a command-line driven set of algorithms called NISCAL. The program, coded in Interactive Data Language (IDL(r)), reads an editable configuration file and processes ASCII lists of NIS raw FITS files. Each sequence in the input list is calibrated to the selected level (raw, dark-subtracted, or calibrated DN, radiance, or I/F). The calibration software used to produce the PDS-archived data sets will also be available from the MSI/NIS team. An example of this software (unsupported) will be provided in the document area of Level 2 NIS data. Appendix A. NIS Calibration Coefficients Table A-I presents basic wavelength and calibration data for the NIS detectors. The crosstalk correction is the value by which the signal in the crosstalk channels is multiplied before subtracting from the respective Ge channel. The slit ratio is the factor by which the DN values taken with the wide slit are divided to get equivalent values to the narrow slit. The DN-Radiance coefficients are in units of DN/(W m-2 sr-1 micrometer-1). Coefficients are derived from in-lab measurements using a gold integrating sphere. Noise-equivalent spectral radiance (NESR) values are derived from 2112 10-second observations of dark space and given in radiance units of W m-2 sr-1 micrometer-1 as discussed in section 5.4.2.2 of [IZENBERGETAL2000]. Channels that are in cutoff and receive little or no signal are italicized. Appendix B. NIS Scan Mirror Coefficients Coefficients for the polynomial fits to the relative response of spectra at different mirror positions for each NIS channel are shown for the Narrow slit, along with the goodness of fit (r2), in Table B-I. The equation followed is: 5 i Relative Response = Sum( m * x ) (B1) i=0 i where x is the mirror position. Relative response in the lab was normalized to 1.0 at position 188 (the mirror position of the primary radiometric calibration measurements). Error values are well under 1% in most channels, the exceptions being those in cutoff. Selected channels are plotted in [IZENBERGETAL2000]. Wide-slit corrections are similar, though error bars are smaller. ***N.B. Tables A-I and B-I have been removed from this text and now reside in the CALIB/ directory of this PDS volume. The filenames of the table together with the PDS labels should enable the use to correlate each file with its relevant steps in the calibration procedure described in this document. References [HAPKE1981] [IZENBERGETAL2000] [WARRENETAL1997] [WEHRLI1986]