DMSP/SSUSI Cal/Val

The Defense Meteorological Satellite Program (DMSP) Block 5D3 spacecraft F16, launched on October 18, 2003, hosted the first flight of two new operational ionosphere/upper atmosphere remote sensing instruments: the Special Sensor Ultraviolet Limb Imager (SSULI) and the Special Sensor Ultraviolet Spectrographic Imager (SSUSI). CPI was an active member of the SSUSI calibration/validation (cal/val) team whose work was completed in early 2006. A description of key features of the work follows. Cal/val work by CPI is continuing with the second flight of SSUSI on DMSP spacecraft F17. Participation on the F16 team followed the development by CPI of several of the SSUSI remote sensing algorithms and its active participation in the design, development, and verification of SSUSI's operational software. The work was performed for the sensor contractor, the Applied Physics Laboratory (APL). CPI algorithms include those producing the following data products:

auroral boundary
characteristic energy E_o and energy flux Q of precipitating electrons
corresponding E-layer parameters (height of E-layer peak HmE and peak density NmE)
dayside O/N₂ (column density referenced to the fixed N₂ value of 10¹⁷ cm^-2)
Q_EUV, an integrated measure of the solar EUV/XUV energy flux shortward of 45 nm (FUV dayglow is produced shortward of this wavelength)
Dayside neutral density profiles (NDPs)
Dayside NmF2 (density at F2 peak) and HmF2 (altitude at the peak)

Development of the above algorithms was made possible by CPI's state-of-the-art models for characterizing distributions of energetic electrons and using them to derive volume and column emission rates, ionization rates, and corresponding electron density profiles (EDPs). These models go by the names B3C (Boltzmann 3-Constituent auroral electron transport model) and AURIC (Atmospheric Ultraviolet Radiance Integrated Code for modeling dayglow and nightglow).

Most of our efforts under F16 cal/val were directed to the validation of the above auroral products with the exception of Auroral Boundary. Discussion is thus restricted to E_o, Q, NmE, and HmE. Team organizations that CPI worked with are the Aerospace Corporation (managing the program), APL, and AFRL/Hanscom. AFRL provided the SSJ/5 sensor that measures energetic electrons and protons once per second from 30 eV to 30 keV. This, for the first time, provides a critical test of auroral FUV remote sensing through comparison of products for SSUSI data that are coincident with SSJ/5 data. Past attempts have involved sensors on different satellites for which coincidences are always questionable. Here, coincident data are continually available over small horizontal spatial scales (10's of km for SSUSI with the higher resolution SSJ/5 data smoothed to the SSUSI resolution).

Extensive comparisons (entire cuts through the auroral oval) were made between SSJ/5 E_avg and Q values and corresponding SSUSI E_o and Q values. Characteristic energy E_o refers to Gaussian distributions for characterizing electron precipitation for which E_o is similar to the average energy. In addition, comparisons were also made between measured NmE and HmE values and those derived from SSUSI data with near-coincidence in time and location. Measurements were made by the Sondrestrom, Greenland Incoherent Scatter Radar (ISR) and the EISCAT radar at Tromsø, Norway. Discussion here focuses on the SSUSI/SSJ/5 comparisons.

Discovery of a serious complication to auroral FUV remote sensing
The remote sensing technique as developed for routine processing of SSUSI auroral data relies on two of SSUSI's five spectral channels. The five go by the designations Ly α (or HI 121.6 nm), OI 130.4, 135.6 (combination of OI 135.6 nm and N₂ Lyman-Birge-Hopfield [LBH] band emission within this channel), LBH_S [~141-150 nm], and LBH_L [~165-180 nm]. The latter two channels are used to specify E_o and Q assuming all precipitation is by electrons with a spectra behavior characterized by a Gaussian distribution with the addition of high and low energy tails whose specifications are based on past surveys of SSJ/4 data. Justification for not explicitly treating proton aurora was based on similar LBH emission yields for proton and electron aurora based on current models (although proton yields are somewhat greater). The most unexpected finding from the SSUSI - SSJ/5 comparisons is that current models (including B3C that models each type of precipitation as well as a mix of the two) seriously underestimate FUV emission for proton aurora and that this component must be explicitly treated in the interpretation of SSUSI auroral data. This seriously complicates the remote sensing method since direct information is not available on the average energy of the protons (excluding data coincident with SSJ/5) that is needed to specify emission yields (the relevant ones here are for Ly α, LBH_S, and LBH_L). If one knows the average energy, then Q for proton precipitation could be estimated from SSUSI Ly α measurements and then be used to estimate the proton contribution to LBH_S, and LBH_L and from there the electron contribution. Future algorithm development work must deal with this problem and will need to rely on statistical descriptions of proton average energy.

Brief description of SSUSI - SSJ/5 comparisons and findings
For each SSJ/5 measurement (once per second), field-line tracing is performed from the satellite down to 120 km. We assume that the peak of volume emission corresponding to SSJ/5 measurements is at 120 km although the resulting SSUSI look angle is not sensitive to the nominal range of peak altitudes from, say, 150 to 100 km. The derived location serves as the center of a circle with a diameter of ~38 km within which SSUSI pixels are summed for each of the two LBH channels. The size of the circle is a compromise between loss of S/N and loss of spatial resolution. Assuming electron aurora, E_o and Q are then derived from the averaged LBH values. Each SSJ/5 measurement is separated from the next by ~7.5 km. Five-point smoothing is applied to SSJ/5 E_avg and Q to place the measurements at approximately the same spatial resolution as SSUSI E_o and Q.

We now illustrate cases of good as well as poor agreement between the two sets of products. We begin in Figure 1with images of Ly α and LBH_L over Antarctica from orbit 3962 on July 25 2004. For this case, there is essentially no proton aurora present. Figure 2 shows product comparisons along the white cut seen in the images in Figure 1. Overall acceptable agreement is achieved for this case of pure electron aurora. Further details are given in Figure 3 whose caption explains the various panels. A case is now addressed where strong proton aurora is present. This is from orbit 5458 on Nov 8 2004 where images like those in Figure 1 may be seen in Figure 4 except at Northern rather than Southern latitudes. The signature of strong proton aurora is seen in the Ly α image on the left in the vicinity of the white cut. Figure 5, similar to Figure 2, allows for product comparisons along the cut. Serious disagreement is seen in the fourth panel from the top between Q values. There are two source of the disagreement: SSJ/5 is missing significant energy flux that occurs above 30 keV and the SSUSI algorithm is assuming all LBH emission arises from electron aurora. A significant fraction of LBH is coming from proton aurora for which our overall analysis demonstrates that an LBH yield for electron aurora cannot substitute for the proton yield (the latter being significantly greater). This will require a new algorithm beginning with upward revisions to our LBH cross sections for proton and H-atom impact.

Figure 1. SSUSI auroral images of Ly α and LBH_L during orbit 3962 on July 25, 2004. The white arrow identifies a cut along which coincident SSUSI and SSJ/5 data products can be compared (see next figure). The contours labeled 90 and 100 refer to solar zenith angle.

Figure 2. SSUSI and SSJ/5 auroral observations and data products as a function of time during the southern auroral pass for orbit 3962 on day 207 (July 25), 2004. (a) The LBH_L intensity. (b) The LBH_S/LBH_L ratio. (c) Comparison of E_o derived from SSUSI and SSJ/5 measured E_avg. (d) Comparison of Q derived from SSUSI and SSJ/5. (e) Geographic latitude/longitude and geomagnetic longitude of the field line traced pixel used for comparisons. (f) The inclination angle of the geomagnetic field lines relative to vertical. Vertical lines indicate when the SSUSI and SSJ/5 data are actually coincident in time. In (d), the dashed line is the apparent Q associated with high-energy penetrating particles, as determined by SSJ/5 background signal measurements. For this particular orbit the contamination index is near zero.

Figure 3. Additional SSUSI and SSJ/5 auroral observations and inferred data products as a function of time from the same orbit as in Figure 1. (a) Comparison of SSUSI E_o and SSJ/5 average electron energy determined by integrating the SSJ/5 spectrum from 1 to 30 keV. (b) The average proton energy measured by SSJ/5 calculated using three different energy ranges. (c) Comparison of SSUSI Q and SSJ/5 total (electron + proton) Q determined by integrating the SSJ/5 spectrum from 1 to 30 keV. (d) The SSJ/5 Q associated with ions only calculated for the same energy ranges as in (b). (e) The ratio of SSJ/5 ion Q to total Q for two of the energy ranges

Figure 4. Similar to Figure 1 except for orbit 5458 on Nov 8, 2004. The Ly α image on the left shows the presence of strong proton aurora in the vicinity of the cut for which data products are compared in the next figure. The contours labeled 90 and 100 refer to solar zenith angle.

Figure 5. SSUSI and SSJ/5 auroral observations and inferred data products for orbit 5458 on day 313 (November 8), 2004. Individual panels are as described in the caption of Figure 2. Here, there is significant proton precipitation and serious disagreement between SSUSI and SSJ/5 Q values.

Figure 6. Additional SSUSI and SSJ/5 auroral observations and inferred data products for orbit 5458 on day 313 (November 8), 2004. Individual panels are as described in the caption of Figure 3.