INSTRUMENT: SUSIM UARS (Solar Ultraviolet Spectral Irradiance Monitor on the Upper Atmosphere Research Satellite) Contact: Dr. Guenter Brueckner, code 7600, NRL, Washington DC 20375 Email: brueckner@susim.nrl.navy.mil SUSIM Template Version 2.0 SUSIM UARS Data version 19 LEVEL 3 REPORTED DATA: 1. Solar Ultraviolet Irradiance gridded to 1 nm intervals from 115.5 nm to 410.5 nm. For version 19, the data represents the convolution of the true solar irradiance with the SUSIM instrument function (which approximates a triangle having FWHM of 1.1 nm). 2. Solar Indices: H I Lyman alpha, Oxygen I, Carbon II, Carbon IV, Aluminum edge, Magnesium II, and Calcium II h&k. The first four of these represent the measured irradiance found in the identifying emission line. In the case of the Lyman alpha index, the irradiance at the nearby Si III, Si II, and C III emission lines are also included. The last three represent ratios of irradiances in proximate wavelength regions and so are dimensionless. The Mg II and Ca II are core-to-wing ratios of absorption lines, while the Al edge index is an ratio of irradiances taken on both sides of an absorption edge. SUSIM Irradiance Error Estimates (5/1/97) The table on two screens hence divides the solar spectrum into wavelength ranges having similar error characteristics. We also classify the types of error as precision, relative accuracy, and absolute accuracy. Defined here, measurement precision is an estimate of the variation in measured signal from one signal to the next under the same internal and external conditions. Absolute measurement accuracy is here defined as the level of agreement between the reported and the "true" values. Relative measurement accuracy is defined as the relative change in the ratio of one day's reported values to the "true" value to that ratio for another day. The estimated error values given in the table below are a result of analysis of the first three years of data. Precision estimates are based on the calculated statistical standard deviations that are reported in the data. The relative accuracy is estimated through intercomparison of the various SUSIM calibration scans of the sun as well as with data from other solar instruments. The absolute accuracy is estimated by properly combining the errors associated with the relative accuracy with the current estimation of errors which originated during the initial calibration. The values given for each error type are given as a percentage of the total. Where a range of wavelengths is given, the estimates refer to typical errors pertain to the "continuum" irradiance values. The values for strong emission lines whose strength is given in the solar indices are given separately. Where a range of error values are given, the first is a typical 1-sigma error while the second the error's estimated maximum value. The two precision values given below correspond to a representative day near the start of observations and one for a days into the third year, but before the MgF2-1 entrance filter was permanently removed from the standard optical path. The following formula approximates wavelength dependence of the ratios referred to on the next screen in items 3), 4), 6), and 7): ratio_% = 100 * B * exp[A*w] where w is the wavelength in nanometers and for case (1): day 196, A = -0.074 and B = 2927 and for case (2): day 822, A = -0.086 and B = 17725 SUSIM UARS Irradiance Error Estimate Summary Continuum irradiance Precision Abs. Accuracy Rel. Accuracy wavelength regions 1) 115-140 nm 45-65 50-100 50-100 2) 140-150 nm (1)-(2) 6-13% 5-10% 3) 150-160 nm (1)-(2) 4-8% 3-5% 4) 160-170 nm (1)-(2) 4-11% 2-8 4) 170-208 nm (1)-(2) 3-7% 1-4 5) 208-300 nm 0.4% 3-6% 1-3 6) 300-390 nm 0.2% 3-5% 1-2 7) 390-410 nm 0.2% 3-6% 1-3 Strong Emission lines 1) H I Ly-a & others (121.6 nm) 1-2% 4-10% 3-5 2) O I triplet (130 nm) 3-6% 5-11% 4-8 3) C II (134 nm) 2-4% 6-12% 3-5 4) Si IV doublet (140 nm, no index) 1-3% 6-12% 2-6 5) C IV (156 nm) (derived from hi res) 7% 10-20% 7-10% 6) Mg II core-to-wing ratio 0.4% 1% 0.8% The 1-sigma errors are given as a percentage. ERROR ANALYSIS DISCUSSION The precision given in the table is measured statistically for every reported irradiance. It accounts for not only the statistics of the solar signals but also the statistics of the dark signals which are used to form the subtracted dark signal estimate. Since the "true" irradiance is not known, there is no firm and direct measurement of the absolute or relative accuracy of the irradiances. The values given are based on partly subjective qualitative and quantitative comparisons with measurements made by other instruments and also with the SUSIM measurements through time using guidance from outside knowledge of the general behavior of solar variability. Complicating any presentation of these error estimates is their assured time dependence. The level of absolute accuracy shrinks (i.e., the errors grow) as the experiment progresses. The level of relative accuracy depends not only on the length of the time period but also on when that time period occurs. For example, relative errors are larger for 1991 and early 1992 at the start of the experiment for a variety of reasons, including insufficient measurements of a highly varying degradation and relatively bad solar pointing. On the other hand, the measurement precision was also relatively good during that same period because of the high signal levels, especially at short wavelengths. SYSTEMATIC EFFECTS Wavelength Registration - Wavelengths of the 1.1 nm scans used to form the gridded irradiances are corrected to within 0.06 nm (1-sigma using a spectral feature method. Stray Light - The SUSIM data are corrected for instrumental stray light. The model assumes that the input solar signal is convolved with a family of Lorentzian functions. Currently, the stray light model for the 1.1 nm scan data appears to undercorrect somewhat. This causes the continuum between emission lines at wavelengths below 140 nm to be overestimated. Further, stray light appears to be undercorrected for wavelengths above 395 nm. Field of View - There were significant solar pointing errors before UARS day 77 when SSPP pointing algorithms were improved. The software used to produce the V19 data attempts to compensate for this with unknown success. Systematic errors due to solar off-pointing are most apparent for otherwise very precise long wavelengths when the pointing behavior undergoes abrupt changes such as at UARS yaw-arounds or when onboard solar pointing algorithms were changed. Initial Optical Calibration - Problems in initial optical calibration are likely cause to absolute irradiance errors as large as 5% in selected wavelength regions. Also, wavelength registration problems may possibly cause even larger errors in the application of the initial calibration at wavelengths < 130nm, but the actual level is not known. Responsivity Degradation - The degradation of the working channel is measured through intercomparison of its solar scans and that of three other "reference" channels. Each of these is calibrated by scans of one or more onboard stable deuterium lamps. The amount of lamp output degradation is gauged through intercomparison of scans scans of the different lamps by the same optical channel. The periodic calibrations determine the working channel responsivity on calibration days. Responsivities on other days are found by interpolation using an abscissa derived from the channel's lamp and solar UV exposure. Degradation errors are generally larger for short wavelengths and early mission days because the degradation itself is larger there. Further, in the case of early mission days, degradation measurements were less frequent and were done with fewer independent reference channels. Finite Instrument Response Function and Binning Effects - The reported irradiances represent the completely resolved solar spectral irradiance that is convolved with the SUSIM instrument function, a triangle having a FWHM of about 1.1 nm. Where the solar spectral irradiance changes rapidly, at either emission or absorption lines or at absorption edges, the 2.2 nm width at the base of the SUSIM instrument function will cause an apparent "spreading" to adjacent bins. The solar indices are the recommended way to chart changes in the emission lines. Miscellaneous Systematic Effects - a) The D2 lamps used to calibrate the reference channel have a strong slope in their radiance between 160 and 170 nm in the 5nm resolution scans used to measure them. As a result, irradiances here are a bit more uncertain than elsewhere. b) Wavelength errors are sometimes introduced when scans are interrupted and restarted which current algorithms now only partly correct. c) Gain and temperature effects are modeled fairly accurately (<< 1%). Detector temperatures are measured almost continuously. Detector gains are measured at least once per day through connection to a fixed current calibration source.