A Missing Piece of the E‐Region Puzzle: High‐Resolution Photoionization Cross Sections and Solar Irradiances in Models

Most ionospheric models cannot sufficiently reproduce the observed electron density profiles in the E‐region ionosphere, since they usually underestimate electron densities and do not match the profile shape. Mitigation of these issues is often addressed by increasing the solar soft X‐ray flux which is ineffective for resolving data‐model discrepancies. We show that low‐resolution cross sections and solar spectral irradiances fail to preserve structure within the data, which considerably impacts radiative processes in the E‐region, and are largely responsible for the discrepancies between observations and simulations. To resolve data‐model inconsistencies, we utilize new high‐resolution (0.001 nm) atomic oxygen (O) and molecular nitrogen (N2) cross sections and solar spectral irradiances, which contain autoionization and narrow rotational lines that allow solar photons to reach lower altitudes and increase the photoelectron flux. This work improves upon Meier et al. (2007, https://doi.org/10.1029/2006gl028484) by additionally incorporating high‐resolution N2 photoionization and photoabsorption cross sections in model calculations. Model results with the new inputs show increased O+ production rates of over 500%, larger than those of Meier et al. (2007, https://doi.org/10.1029/2006gl028484) and total ion production rates of over 125%, while N2+ ${\mathrm{N}}_{2}^{+}$ production rates decrease by ∼15% in the E‐region in comparison to the results obtained using the cross section compilation from Conway (1988, https://apps.dtic.mil/sti/pdfs/ADA193866.pdf). Low‐resolution molecular oxygen (O2) cross sections from the Conway compilation are utilized for all input cases and indicate that O2+ ${\mathrm{O}}_{2}^{+}$ is a dominant contributor to the total ion production rate in the E‐region. Specifically, the photoionization contributed from longer wavelengths is a main contributor at ∼120 km.


Introduction
Accurate characterization of photoionization processes (including ionization frequencies and rates) is vital for understanding gases in the terrestrial atmosphere and its response to solar ionizing radiation.Ab initio models use solar spectral irradiances and photoionization and photoabsorption cross sections to compute the deposition of energy as a function of wavelength and altitude, eventually leading to the production of dayglow.However, essentially all ionospheric E-region models underestimate electron densities and fail to match observed electron density profiles (EDPs) (Fang et al., 2008;Sojka et al., 2014;Solomon, 2006).Attempts to mitigate this puzzling outcome are often attempted by ad hoc modifications to the solar soft X-ray flux; however, these short term solutions do not properly address the underlying issue.For example, Solomon et al. (2001) used Student Nitric Oxide Explorer satellite (SNOE) data to scale the extreme ultraviolet irradiance by a factor of 4 to achieve agreement with E and F1 region observed electron densities.Titheridge (2003); Titheridge (2000Titheridge ( , 1997) ) also found that EUVAC (EUV flux model for aeronomic calculations; Richards et al., 1994) would need to be scaled upward to agree with measured values.Sojka et al. (2013) and their references have suggested that modifying the observed solar soft X-ray flux could explain the differences between measured and modeled EDPs.We believe this deficiency is related to the poor physical representation of the deposition of solar ionizing radiation in the lower thermosphere (90-150 km).This is a direct consequence of the low-resolution cross sections and solar spectral irradiances that are implemented in models, typically with spectral resolutions of ∼1 nm (Sojka et al., 2014;Solomon, 2017).Low-resolution cross sections and low-resolution irradiances do not permit sufficient penetration of solar radiation to lower altitudes and thus lead to a severe underestimation of E-region electron production rates (Meier et al., 2007).The scarcity of high-quality photoionization and photoabsorption cross sections and inadequate spectral irradiances have hampered the ability to properly model observations.In another example, the Chandra and XMM-Newton observatories currently provide an abundance of X-ray spectra of astronomical objects, but a lack of high-quality atomic data impedes the interpretation of these spectra (Hasoglu et al., 2010).It is therefore necessary to produce and utilize high-resolution cross sections in combination with high-resolution solar irradiances to accurately model observations and interpret spectra.
Photoionization cross sections contain hundreds of narrow autoionization lines which current ionospheric models neglect to include.These autoionization features are present as the result of photon absorption into the semibound states of an atom/molecule that reside above the ionization limit.The semi-bound electron can either return to a bound state with no ionization or jump into the continuum as an autoionization.The O cross section from Meier et al. (2007) included hundreds of autoionization lines that exhibited Fano-type line shapes characterized by very large increases adjacent to major dips at discrete wavelengths.These autoionization features produce windows and peaks in the cross sections that enable a net increased penetration of solar EUV radiation.Therefore, use of high-resolution cross sections which can resolve the line profiles and preserve the dips and peaks are needed to properly account for the solar ionizing radiation at lower altitudes.
Solar spectral irradiance models often bin the flux into 1 nm or broader bands in order to maximize computational efficiency at the expense of preserving the structure in the solar spectrum (Solomon & Qian, 2005).Models such as the Time Dependent Ionospheric Model (TDIM; Schunk, 1988;Schunk & Walker, 1973) which use Extreme ultraviolet Variability Experiment (EVE) solar flux measurements at ∼0.1 nm resolution, predicted electron densities that were 20% lower than observed implying a photoionization rate that is low by about 36% (Sojka et al., 2014).Meier et al. (2007) showed that such models significantly underestimate ionization rates in the Eregion.Their analysis of high-resolution (0.001 nm) O cross sections that include more than 300 autoionization lines, demonstrated that photoionization rate estimates in the E-region can be low by as much as a factor of 3 when using solar spectral irradiances and cross sections modeled at 1.0 nm as opposed to a spectral resolution of 0.001 nm.The high-resolution cross sections possess autoionization features and dips in the continuum that allow photons to "leak" through to lower altitudes enabling a net increased penetration of solar EUV radiation to the Eregion and thus increased ionization.To examine the increased ionization, sufficiently high-resolution solar spectra of the same spectral resolution as the high-resolution cross sections are necessary to more accurately determine the E-region electron density.
Determining the appropriate atomic cross sectional data over a range of energies and temperatures necessitates a predominantly theoretical approach.Experimental measurements are essential to validate theoretical studies, particularly in the case of complex heavy ionic systems.However, experimental studies are limited in the range and resolution of energy, and charge states investigated (Angel & Samson, 1988;Feldman et al., 2001).In order to model the dynamics of the interaction of photons with atomic systems, large R-matrix (Abdel Naby et al., 2013;McLaughlin et al., 2011McLaughlin et al., , 2013;;Scully et al., 2006;Tayal & Zatsarinny, 2016) calculations have been performed to accurately predict both the whole state energies and the resulting photoabsorption (Sant'Anna et al., 2011) and photoionization spectra (Gharaibeh et al., 2011).Bautista et al. (2022) and Bergemann et al. (2021) show R-matrix calculations of O photoionization cross sections, they are limited to a few of the O I lines while our work provides a more extensive analysis including 34 atomic oxygen states.Gorczyca et al. (2013) also presented an O I photoabsorption cross section, determined via R-matrix calculations, for energies of interest in X-ray spectral modeling in order to resolve discrepancies in molecular O abundances between spectral models.Tashiro (2010) used the United Kingdom (UK) R-matrix codes (Morgan et al., 1998) to calculate N 2 photoionization cross sections; however, it was simply intended to demonstrate the ability of their newly implemented photoionization code, and the photon energy range was limited.Therefore, until now, no detailed high-resolution O and N 2 theoretical photoionization cross sections have been available to examine the discrepancies in the photoionization rates, and thus electron densities.This work resolves this deficiency.

While other works by
In this paper we first discuss the photoionization and photoabsorption cross sections used as inputs in photochemistry models (with a detailed description pertaining to how these are extended to shorter and longer wavelength regimes provided in the Supporting Information S1).Furthermore, details regarding the R-matrix model calculations of the cross sections are relegated to the appendix.We then discuss the solar flux models and methods used to generate high-resolution solar spectral irradiances.In the following section we examine the model outputs of two radiative transfer/photoionization codes using the new high-resolution inputs.In the final section we summarize the most significant insights from the study.

Cross Sections
In this work, the R-matrix method is used to treat photoionization of atomic oxygen and molecular nitrogen.Details regarding this method are provided in Appendix A. New high-resolution atomic oxygen and molecular nitrogen photoionization cross sections are produced using the R-matrix method with a spectral resolution of 0.0001-0.02nm, hereafter referred to as R-matrix resolution (Table 1).The cross sections can be implemented in aeronomy models to compute volume production rates, electron density profiles, and additional model outputs.This work tests the impact of the combined high-resolution solar irradiances and photoionization and photoabsorption cross section data (0.001 nm wavelength resolution) on E-region ionization rates determined by the Atmospheric Ultraviolet Radiance Integrated Code (AURIC; Strickland et al., 1999) and the Meier et al. (2007) photoionization model.AURIC requires that model inputs such as the photoionization cross sections and solar spectrum be on the same wavelength grid.Therefore, we discuss the creation of updated high-resolution cross sections (HR, Table 1) to match the wavelength grid of the high-resolution solar spectrum (hereafter the fine grid) provided by H. Warren and colleagues (Warren, 2005(Warren, , 2006) ) at the Naval Research Laboratory (discussed in further detail in Section 3).The fine wavelength grid has a 0.001 nm bin size from 0.1 to 105 nm and captures the high spectral resolution of the new cross sections in comparison to current cross sections with resolutions typically ranging from 0.05 to 0.1 nm.The HR cross sections are compared to the low-resolution compilation by Conway (1988) (hereafter LR), which are currently used in AURIC.To do so, we use the R-matrix cross sections and LR data to create binned cross sections on a coarse grid comprised of 0.05 nm bins from 0.1 to 10 and 0.1 nm bins from 10 to 105 nm (hereafter referred to as BHR, Table 1).By examining permutations of the Conway (1988) and high-resolution cross sections and solar spectrum provided by this work on the fine and coarse wavelength grids as model inputs for AURIC we can separate the effects of the new cross section data and resolution.To reiterate, figures corresponding to the cross sections in the following passages are labeled as the following: the high-resolution R-matrix cross sections interpolated onto the fine grid are represented by HR and those binned onto the coarse grid are designated as BHR while the Conway (1988) cross sections on the coarse grid are labeled as LR and the Conway (1988) data interpolated onto the fine grid are represented by ILR (Table 1).
The new partial cross sections (the cross section for a given state in the species), excluding the O K-shell, require a linear shift in energy to match the NIST measured ionization potential (IP).The shift in energy is required as a result of the differences in the calculated and experimental IPs that arise due to the fixed nuclei approximation used in the R-matrix calculation.The approximation does not account for the difference of equilibrium bond distance in neutral and ionized molecules causing the calculated and experimental IPs to differ to some degree.In addition, the calculations are not accurate enough to exactly reproduce the experimental IPs completely.In order to account for these differences a uniform shift is commonly applied to the partial cross sections based on accepted values from the literature for each species as the states and resonance peaks may have different accuracy.Small deviations in the IP from experimental values exist and thus the uniform shift should pose negligible differences to the individual states.For example, the NIST Chemistry WebBook lists N 2 ionization energies of 15.5802-15.5812eV from optical spectroscopy (Ogawa & Tanaka, 1962;Trickl et al., 1989;Worley, 1943;Worley & Jenkins, 1938), 15.58-15.61eV from photoelectron spectroscopy (Hotop & Niehaus, 1970;Katrib et al., 1973;Kimura et al., 1981;Lee & Rabalais, 1974;Natalis, 1973;Potts & Williams, 1974), and 15.58-15.7 eV from electron impact techniques (Armentrout et al., 1981;Grade et al., 1983;Sahini et al., 1978;Stephan et al., 1984).Likewise, NIST also lists O IP values ranging from 13.0 to 14.0 eV from various experimental methods (Banon et al., 1982;Kelly, 1987;Lide, 1992;Paule, 1976).The specific shift in the photon energy of the partial cross sections is discussed in the Supporting Information S1.
As discussed in Sections 2.1-3, the high-resolution cross sections and solar spectra are processed and placed on the corresponding fine and coarse wavelength grids via interpolation and binning.We utilized an interpolation procedure that employs piecewise cubic polynomial interpolation to guarantee monotonic behavior and prevent unwanted oscillations (Steffen, 1990).Additionally, the high-resolution R-matrix cross sections are supplemented by ancillary data to extend the wavelength range and preserve autoionization features at transitional wavelengths near the limits of the new data.These ancillary data are discussed in Sections 2.1-2.3 and the Supporting Information S1.Following Kirby et al. (1979), it is necessary to preserve the integrated cross section when data are obtained from multiple sources of varying resolution and have been mapped to a different resolution (discussed in detail in the Supporting Information S1).We impose the same requirement to conserve the integrated cross section by implementing flux-preserving interpolation, or binning, which guarantees that the integrated cross section on the original (finer) grid over a specific wavelength interval is equivalent to the integrated cross section on the coarser grid over the same wavelength interval.The discussion in the following sections and the Supporting Information S1 details these steps in the process while Section 4 presents the impacts the updated cross sections have on atmospheric model calculations.

O Photoionization and Photoabsorption Cross Sections
Through the application of the RMPS method, as described in Appendix A, we obtained the valence and K-shell photoionization cross sections of atomic oxygen (Figures 1-9).All of the figures depicting cross sections and solar spectral irradiances are plotted with decreasing wavelengths from left to right which correspond to increasing energies.In Figures 1-7 we show the cross sections for the six dominant states as identified by Conway (1988), the absorption (ABS) cross section (also shown in Figure 8), and combine the remaining 29 calculated states into a pseudo state cross section shown in Figure 9 (further details are provided in the Supporting Information S1).The pseudo state was created to accommodate contributions from the additional 29 states not explicitly included in the Conway (1988) compilation, and thus absent from the AURIC input file.These do not require individual tabulations because the states are not significant contributors to the photoionization rate in the E-region (compare Figure 9 with Figure 1).The HR photoionization cross sections are constructed from the Rmatrix variable grid calculations (defined as the 0.0001-0.02nm wavelength intervals converted from the basic energy grid of the computations), supplemented with LR data at shorter wavelengths where high-resolution Rmatrix data are unavailable.The R-matrix methodology is generally accurate through 20-50 eV above the ionization threshold (∼62.0 24.8 nm) depending on the choice of basis function, R-matrix boundary radius, and so on.Details are provided in Supporting Information S1.
Figures 2-9 illustrate the photoionization and photoabsorption cross sections interpolated onto the fine wavelength (0.001 nm) grid (middle plot in panels) and binned onto the coarse wavelength grid (right plot in panels).The left panels show the new R-matrix cross sections on the native variable grid along with the LR cross sections, which were the default AURIC inputs.Figure 8 shows the total photoabsorption cross section which represents the sum of the R-matrix O photoionization cross sections on the native variable grid.The R-matrix calculations provided wavelength coverage down to about 10 nm.The newly constructed HR cross sections contain a great deal more autoionization structure that is absent in the LR cross sections and which are partially preserved when the high-resolution R-matrix data are binned onto the coarse wavelength grid.While differences in the HR (BHR)        and ILR (LR) cross sections are dominated by these new autoionization features, the new 4 P and 2 P state cross sections are systematically lower than the LR values (Figures 5 and 6).The high-resolution fine and coarse grid pseudo states (Figure 9) also exhibit a "hump" in the cross section values at shorter wavelengths that results from smoothly fitting over the gap in the high-resolution absorption data (Figure 8) from 5.7 nm down to 4.1 nm.Although this side effect is present, incorporating the remaining 29 O + states into the pseudo state allows us to preserve the ionization lines at these shorter wavelengths.

N 2 Photoionization Cross Section
Utilizing the R-matrix method (Burke, 2011;Gillan et al., 1995;Schneider, 1995;Tennyson, 2010) we compute the photoionization cross sections into the N + 2 X, A, and B states (Figures 10-12) which are the dominant contributors to the total N 2 photoionization cross section.Details are provided in the Supporting Information S1.
Figures 10-12 illustrate the photoionization cross sections interpolated onto the fine (0.001 nm) wavelength grid and binned onto the coarse wavelength grid.The newly constructed high-resolution cross sections contain considerably more structure from the inclusion of autoionization lines that are absent in the LR cross sections.As in the case of the atomic O cross sections, some autoionization structure remains after binning onto the coarse wavelength grid.At longer wavelengths, the updated cross sections differ by about an order of magnitude in comparison to the LR cross sections particularly for the X and B states (Figures 10 and 12).

N 2 Photoabsorption Cross Section
To formulate the new N 2 photoabsorption cross section, the states below the ionization threshold of the N + 2 X state are first considered.First, the Conway absorption cross section is interpolated onto the fine wavelength grid and the contribution from the Conway X, A, and B states is removed.The contribution from the new high-resolution HR N 2 X, A, and B states, described in Section 2.2, are then added to the absorption cross section.For the   absorption cross section on the coarse wavelength, the contribution from the former X, A, and B states is removed and the remainder is summed with the 3 partial state BHR cross sections.
The LR photoabsorption cross section lacks the high frequency structure at longer wavelengths which has been reported from measurements and more recent models.Therefore, we incorporate a high-resolution N 2 photoabsorption cross section for the ∼73.0-105.0nm region computed using the method of Bishop et al. (2004Bishop et al. ( , 2007) ) and measurements from Carter (1972) (Figure 13).The Bishop model generates the N 2 photoabsorption spectrum for 6 singlet states (b 1 Π u , b 1 Σ + u , c 1 Σ + u , c 1 Π u , o 1 Π u , and e 1 Σ u ) using a pure rotation approximation.The cross section is generated on a 0.0001 nm grid and then interpolated onto the fine and coarse wavelength grids, respectively, using a flux-preserving interpolation technique as mentioned in Section 2. Comparison of the Bishop cross section with Guertler et al. (1977) (0.003 nm resolution) and Huffman (1969) (0.004 nm resolution) measurements show very good agreement in the alignment of peaks and overall shape of the cross section above ∼86.6 nm with more noticeable discrepancies at shorter wavelengths.There are strong absorption features (>100 Mb) below 86.6 nm (i.e., 83.5 nm (near b ′ 1 Σ + u v′ = 25), 84.2 nm (c′ 1 Σ + u v′ = 7), 85.6 nm (near c′ 1 Σ + u v′ = 6), and 86.5 nm (e′ 1 Π u v′ = 0)) which are not present in the Bishop photoabsorption cross section.To preserve these features and utilize the highest resolution photoabsorption cross sections available, we combine the modified absorption cross section that uses the high-resolution R-matrix calculations on the fine wavelength grid below ∼73.62 nm (and below 73.9 nm for the coarse wavelength grid cross sections) with digitized data from Carter (1972), and the high-resolution Bishop model (Figure 14).The Carter (1972) data are interpolated onto the fine and coarse wavelength grids using a flux-preserving interpolation technique (described in Section 2) and spliced in at ∼73.62 (73.9) nm-86.6 nm, for the HR and BHR cross sections respectively, with the interpolated Bishop data used above 86.6 nm. Figure S1 in Supporting Information S1 shows the photoabsorption cross section from the four sources (Bishop, Carter, Conway, and the modified cross sections) for comparison.As illustrated in Figures 13 and 14, the fine and coarse grid photoabsorption cross sections preserve the absorption features which may have otherwise been smeared out by smoothing (or additional data processing) of the low-resolution photoabsorption cross sections prevalently employed in model calculations.
Incorporation of the Carter (1972) cross section below the ionization limit of the N + 2 X state is necessary to preserve structure at longer wavelengths that is not observed in the new N + 2 X, A, and B cross sections.However,   while the Carter (1972) photoabsorption cross section provides coverage at wavelengths between ∼73 nm and up to the ionization limit (∼79.5 nm), the data in this regime cannot be solely attributed to photoionization.Huffman et al. (1963) stated that the bands below the ionization threshold are diffuse, implying that there is predissociation in the bands both they and Carter measured.Figure 2 of Huffman et al. (1963) shows almost no change in the absorption coefficient at the ionization threshold, implying that the absorption cross section is much larger than the ionization cross section.We infer that the undulations in the Carter N 2 photoabsorption cross section are therefore predissociation into N + N rather than ionization.Furthermore, Conway (1988) states that the ionization yield is 100% for wavelengths below 66 nm and thus below the photoionization threshold, predissociation rapidly decreases toward 66 nm.Samson et al. (1977) found that from 65-80 nm several discrete Rydberg absorption lines exist and can autoionize to an extent; however, the photoionization yield is generally less than 100% in this region.The data of Samson et al. (1977) do not indicate a significant difference between the absorption and ionization cross sections below 70 nm but the data do increase above 70 nm, where Carter measurements becomes available.This again suggests that the peaks in the Carter data are due to N + N predissociation and the ionization cross section is close to the lowest values in the Carter photoabsorption cross section.The much broader oscillations, at ∼75.5 nm exceeding 200 Mb, observed in our computed cross sections between 70-79.5 nm are due to autoionization, as Samson et al. (1977) state.Therefore, while the Carter N 2 absorption cross section can be used to construct the HR N 2 photoabsorption cross section, it is not suitable to apply to the new high-resolution N 2 photoionization cross section.

Solar Flux Model and Spectra
The Naval Research Laboratory's Extreme Ultraviolet (NRLEUV) spectral irradiance model of Warren (2005) is used to generate the high-resolution solar spectra.The model uses a differential emission measure distribution calculated from spatially and spectrally resolved solar data versus solar activity.To improve the limited absolute accuracy afforded by the emission measure method, the NRLEUV spectra are calibrated using EVE solar measurements (Woods et al., 2005(Woods et al., , 2012)).EVE measures the solar EUV irradiance from 0.1 to 105.0 nm at 0.1 nm resolution with 20% absolute accuracy.The NRLEUV 0.001 nm spectrum is normalized to EVE by partitioning the EVE irradiance at each 0.1 nm EVE interval into 0.001 nm subintervals.
The resulting high-resolution solar spectrum is triple box-smoothed (close approximation to a Gaussian line profile) above 46.8nm to simulate realistic solar spectral line widths with FWHM of ∼0.01 nm (Figure 15, top).To compare with the LR cross sections, the solar flux is binned onto the coarse wavelength grid (Figure 15, bottom).The coarse grid spectrum may then be scaled with data tuned for specific solar conditions such as the F10.7 cm radio flux, a reliable indicator of solar activity that is measured daily and used in forecasting space weather.The empirical solar EUV irradiance variability model of Lean et al. (Lean et al., 2003, 2011, 2020;Woods et al., 2012) is used to simulate the solar activity for a spectrum with an F10.7 cm radio flux = 70.The forward model incorporates the NRLSSI-EUV solar EUV spectral irradiance (Lean et al., 2003), updated from Lean et al. (2011) using more recent SEE (Solar EUV Experiment) data (Woods et al., 2018).The NRLEUV spectra, provided by H. Warren of NRL, are then scaled in increments of 0.1 nm (the lowest resolution of the two grids) to the integrated flux of the Lean et al. (2020) model spectra with an F10.7 = 70 (Figure 16).Using this method we can scale the high-resolution solar spectrum from NRL to specific F10.7 values that can fluctuate on daily time scales.The coarse grid solar spectrum in Figure 16 appears shifted due to the ∼100 factor increase in bin size from 0.001 to 0.1 nm resolution when binning the high-resolution solar spectra onto the coarse wavelength grid.The flux from 0 to 105 nm in the scaled spectra are conserved with differences in the integrated fluxes from 0 to 10 nm and 10-105 nm of less than ∼0.6%.The total integrated flux for the new solar minimum spectrum is 2.93 erg cm2 s 1 .

AURIC
The need to incorporate both high-resolution cross sections and solar spectra into photochemistry models prompted the generalization of the Atmospheric Ultraviolet Radiance Integrated Code (Strickland et al., 1999) to ingest data files with over 100,000 wavelengths.Following the update to the model as part of this work (Evans et al., 2023), we replaced the current 0.05-0.1 nm wavelength grid solar spectra used in AURIC with the highresolution fine wavelength (0.001 nm) grid solar spectrum at solar minimum conditions, scaled to F10.7 = 70 as discussed in Section 3 (Figure 16, top).AURIC comparisons presented herein represent model outputs using the Conway (1988) tabulation of cross sections, the newly computed high-resolution O and N 2 photoionization and photoabsorption cross sections, and an MSIS 2.0 model atmosphere (Emmert et al., 2021).AURIC also requires molecular oxygen cross sections.Lacking computations of the O 2 cross sections at high-resolution, we use instead the Conway (1988) O 2 cross sections on the coarse wavelength grid (LR) and then interpolate them onto the fine wavelength grid (ILR).The O 2 cross sections include states that dissociate into O eV, 2,4 Σ g , and K-shell; Conway, 1988) and we account for the ionization arising from these states independent of the total production for O + 2 .Even though photoelectron ionization at ∼110 km is dominated by N + on photoionization only.Therefore, the photoionization rate and volume production rate values reported refer to the contributions from photoionization only and exclude photoelectron ionization contributions.Lyman alpha photoionization of NO is not included in the analysis as the cross sections only extend up to 105 nm.We also note that AURIC rigorously includes ionization from solar photons and photoelectrons (Strickland et al., 1999).The spectrum is then scaled using Lean's model (Lean et al., 2003(Lean et al., , 2011(Lean et al., , 2020) ) for the specific F 10.7 radio flux noted in the panel (see Section 3).The final fine (red) and coarse (blue) wavelength grid solar spectra are used in model calculations.Figures 17-19 illustrate AURIC model outputs for a combination of inputs.By comparing model results for the different spectral resolution cross sections and wavelength grids, we can asses the impact of increasing the resolution versus the change resulting from differences in the cross section value.Comparison of the model volume production rates using the coarse grid cross sections (BHR and LR) and the new solar spectrum on the coarse wavelength grid as inputs highlights the impact of the new cross section data while comparison of the model ionization rates using the Conway (1988) cross sections and new solar spectrum on the coarse and fine wavelength grids (LR and ILR) as inputs highlights the impact of using new high-resolution solar spectrum data alone.We find that using new high-resolution solar spectrum data binned onto the coarse wavelength grid (to match the grid of Conway, 1988) is not sufficient to resolve the discrepancies in both magnitude and shape in model and observational EDPs (Sakib et al., 2023).Similarly, utilizing high-resolution cross section data on a coarse wavelength grid (BHR) with a coarse grid solar spectrum is not sufficient.Using updated high-resolution solar irradiances with the ILR cross sections also does not adequately resolve the data-model discrepancy.Rather, the combination of high-resolution cross sections and high-resolution solar spectra on a fine wavelength grid begins to resolve the discrepancy between the model and observational EDPs (Sakib et al., 2023).The calculations presented in the paper by Sakib et al. (2023) utilize these updated high-resolution cross sections (HR cross sections) and high-resolution solar spectrum on the fine wavelength grid.
Figure 17 illustrates a significant divergence, over ∼150 cm 3 s 1 at and below about 150 km, in the HR results in comparison to LR results for the O + 2s2p 3 4 S 0 state.Below approximately 175 km (E-region), the HR case has much greater volume ion production rates reaching a ratio of about 0.05 for the LR case at approximately 110 km.From ∼100-160 km, the HR case results in a volume production rate ratio of 0.50 or less than previously calculated for LR.These equate to increases of over 100% in the Volume Production Rate (VPR) when using the HR cross sections in comparison to the LR case.Both O + 2s2p 3 2 D 0 and O + 2s2p 3 2 P 0 also show increased production rates when using the HR cross sections from 100 to 135 km with production rate ratios of ∼0.8 (increase of 25%) and ratios of ∼0.95 (increase of 5%) upwards of 210 km. Figure 18, indicates that HR photoionization cross sections also result in much larger volume production rates (an increase of over ∼190 cm 3 s 1 at about 180 km for N + 2 X), specifically for the N + 2 X and B states above approximately 150 and 195 km, respectively, with VPR ratios reaching 0.70 (∼43% increase in the VPR).The high-resolution cross sections on the fine grid result in lower production rates at altitudes lower than approximately 130 km.This may be due to the competing effects of absorption that allows the penetration of more solar EUV radiation into the Eregion and photoionization (around 120 km) which is modulated by the passage of solar radiation through N 2 and O 2 longward of 80 nm and the low-resolution O 2 photoionization cross sections.Lower HR cross section magnitudes, such as for the N + 2 A state, may also contribute to the decrease in the VPR.
The total photoionization production rates each species also demonstrate similar trends.In Figure 19, the top left panel illustrates the sum of the O + VPR that originates from photoionization of atomic oxygen.The contribution from the dissociated states of O + 2 , listed above, are shown as dotted lines in the O + VPR panel; however, are not included in the totals or ratios.We find an overall increase in the O + VPR utilizing the newly calculated partial cross sections with a maximum of 515 cm 3 s 1 at 189 km for the HR case and an increase of 172 cm 3 s 1 over the LR total VPR at 138 km.The total O + photoionization production rate ratios show increased volume production rates in the E-region below from 110 to 140 km with ratios less than ∼0.3 for all the cases reaching a ratio of 0.14 at 115 km for the LR case.The total N + 2 photoionization production rate demonstrates a large increase in magnitude for the HR case, reaching a peak of 800 cm 3 s 1 at 180 km with ratios less than 0.75 for the LR and ILR cases above 207 km (33% increase).It is important to note that like O 2 , several states in the N 2 cross section also dissociate or predissociate (N g , H, and K-shell).The contribution from the N + 2 dissociated states are likewise shown as dotted lines in the middle left panel of Figure 19 and are included in neither the total N + 2 VPR or ratio.Everything else being the same (same solar spectrum and wavelength grid), the updated high-resolution N 2 photoabsorption cross section also impacts the VPR profile with deficits and enhancements at various altitudes.It is clear that the newly improved highresolution cross sections result in larger production rates at higher altitudes in the F-region for N 2 .
The large increase in the ionization rates produced by using the new high-resolution cross sections and solar spectrum may account for the magnitude discrepancy in modeled electron density profiles discussed in Section 1 (Buonsanto et al., 1995;Solomon, 2006;Solomon & Qian, 2005).Although these calculations use the lowresolution Conway (1988) LR and ILR O 2 cross sections, we can infer the impact of the O + 2 volume production rates.As illustrated by the bottom panels of Figure 19, the total O + 2 VPR (which excludes the dissociated states) when utilizing the high-resolution data (O & N 2 HR and BHR cases) produces a large increase in the total O + 2 production rate with a ratio of less than 0.7 at ∼115-175 km, corresponding to an increase of over 40%.The LR case has a ratio of less than 0.4 at 135 km which correlates to an increase of over 150%.
Figure 20 shows the total photoionization production rate (left panel) and ratios (right panel) which do include the contributions from the dissociated N + 2 and O + 2 states.The total VPR for the HR and BHR cases are both larger in magnitude in comparison the LR and ILR cases above 110 km.The HR and LR both peak at 180 km with cross section values of 1529 and 1313 cm 3 s 1 , respectively.The largest difference in magnitude is seen in the Eregion at 126 km where the ratio drops to ∼0.43, corresponding to an increase of 133% in the HR VPR in relation to the LR case.While the BHR, ILR, and LR cases show two distinct peaks in the VPR altitude profile, the HR case exhibits a small tertiary peak at 126 km.This demonstrates that the increase in the VPR of the HR case is due not only to the new partial cross sections but also as a result of the finer wavelength grid.The large difference in magnitude between the HR and LR cases at 126 km (515 cm 3 s 1 ) may produce an increase in the electron density profile when utilizing the HR cross sections.Meier et al. (2007) show that above 140 km the total photoionization rates for atomic oxygen when using highresolution theoretical cross sections, regardless of the resolution of the solar flux grid, are relatively equal when the same cross sections are evaluated at bin centers (Figure 3 from Meier et al., 2007).Their ionization rate ratio increases to 1.4 at ∼110 km and then drops to 1.0 at higher altitudes.Thus, according to Meier et al. (2007) the cross section resolution does impact the calculated photoionization rates between 100 and 140 km.In the analysis presented here the BHR cross section can be compared to results from Figure 4 of Meier et al. (2007), with grid increments of 0.1 nm, although with the caveat that the resolution of the wavelength grid is not identical below 10 nm.Photoionization rate ratio results from Meier et al. (2007) Figure 4 for atomic oxygen show a ratio of ∼0.7 at about 115 km, indicating that the ionization rates for the 0.001 nm case are much greater than for the 0.1 nm case (by about 40%-50%), and even more so for a bin size of 1 nm (ratio of ∼0.3 at 115 km).

Meier Photoionization Code
Figure 21 shows increased photoionization production rates for all species when utilizing an updated version of the ionospheric photoionization rate code from Meier et al. (2007), with similar treatment of the high-resolution cross sections.The updates to the Meier et al. code are the new high-resolution cross sections and high-resolution solar irradiances.The magnitude and shape of the VPR profiles are comparable to results obtained from AURIC in Figures 19 and 20.For example, the O + VPR peaks at 511 cm 3 s 1 at 188 km, nearly identical to the AURIC output.However, the Meier et al. code produces greater ionization at 137 km (232 cm 3 s 1 more than the LR case) in comparison to AURIC results and reaches a ratio of ∼0.12 at 117 km.The ratios shown in Figure 21 are of their LR cross sections with respect to the new high-resolution data.Likewise, the N + 2 VPR peaks with a magnitude of 797 cm 3 s 1 at 184 km with ratios less than 0.75 above 184 km.The O + 2 VPR profiles also exhibit similar results, with O + 2 VPR ratios of less than 0.7 at 114-176 km and a minimum ratio of ∼0.36 at 135 km.The shape of the O + 2 VPR profile is also similar to the VPR results from AURIC which exhibit a bump above about 110 km, implying large increases in the ionization for tens of kilometers above this altitude.The total VPR produced from these calculations has a primary peak of 1505 cm 3 s 1 at 182 km with additional peaks of 1012 cm 3 s 1 at 129 km and 937 cm 3 s 1 at 108 km.The ratio of the total VPRs drops to 0.39 at 129 km which corresponds to an increase of 156% in the volume emission rate.These results are either identical or comparable with those obtained from AURIC with the largest difference arising from the discrepancy in the ionization of O at 137-138 km.The Meier et al. photoionization code does not use flux-weighted cross sections, so that may also contribute to the differences seen in Figure 21.This demonstrates that both methods produce fairly consistent results with a difference of ∼1.5% and ∼5.8% in the total ionization rate of the upper and lower peaks in the Eregion when using the updated high-resolution cross sections.

Discussion
The incorporation of new high-resolution N 2 and O photoionization cross sections into models such as AURIC and the Meier et al. photoionization code demonstrates that the preservation of structure found in the cross sections and solar spectrum is crucial to properly account for electron production in the E-region ionosphere and perhaps resolve the sometimes worse discrepancies found in the D-region (Siskind et al., 2015(Siskind et al., , 2019)).Our analysis indicates large increases (over 500%) in the total O ionization rate profile in the E-region and increases of over 33% in the total N 2 ionization rate in the F-region (above ∼200 km).Meier et al. (2007) found that the 1 nm binned solar flux and cross section cases were a factor of 3 lower (increase of ∼300%) than the high-resolution case at 120 km.While the new high-resolution solar spectrum may be responsible for an increase in the production rate at high altitudes where the atmospheric EUV opacity is low (typically F-region), the driver of the increase in the ionization rate in the E-region is the updated high-resolution O and N 2 cross sections.Although all of the O photoionization cross section states have been updated with the new theoretical partial cross sections, only a subset of the N 2 cross section states were revised (X, A, and B states), as well as the total photoabsorption cross section.The O 2 cross sections are the Conway compilation and have not been updated.Variations in the N 2 photoabsorption cross section, comprised of the revised high-resolution partial cross sections, experimental measurements, and high-resolution models that resolve the rotational absorption lines, produce the largest impact on the calculated ionization rates in the E-region.Interestingly, the changes to the N 2 photoabsorption cross section had a greater impact on the O ion production rates than for N + 2 .This is undoubtedly due to the enhanced penetration of solar radiation through the dips in the N 2 cross section.
Comparison to photoionization codes such as the upgraded Meier et al. code  the N + 2 ionization rate ratios are typical greater than 1.0 within the same range, indicating that the source of the increase in the E-region ionization is O + and O + 2 ionization while N + 2 rates drop when using the new highresolution in comparison to the low-resolution cross section.Meier et al. (2007) also found that there were no accidental resonance overlaps between cross peaks and strong solar lines at the top of the atmosphere and we confirm that there are no major resonances with the updated cross sections from 300 km and above.Comparison of model outputs with observations is also necessary to verify these findings.However, further validation of the model results with incoherent scatter radar measurements and additional models is addressed in the paper by Sakib et al. (2023).
We find an overall increase in the total E-region ion production rate for the new high-resolution inputs relative to low-resolution inputs, which confirms the expectation that high-resolution cross sections and solar spectra allow photons to penetrate deeper into the atmosphere (Meier et al., 2007).Implementing high-resolution cross sections is important and therefore necessary to properly account for the electron production rate in the E-region.A preliminary test was also conducted at higher resolution (0.0001 nm) and showed differences of less than 1% in the total photoionization rate indicating that 0.001 nm resolution is sufficient to characterize the peaks and valleys of the cross sections.For this test the solar spectrum was Gaussian-smoothed with a resolution of about 0.009 nm for wavelengths longer than 40 nm.Results presented here expose the sensitivity of photoionization rate models to the resolution of cross sections, primarily in how preserved autoionization lines impact model outputs.
We also show that O 2 plays a dominant role in photoionization production rates in the E-region (Figure S2 in Supporting Information S1), moreso than O and N 2 .Long wavelength photons (>80 nm) dominate the direct photoionization of O 2 in the E-region due to the penetration of solar radiation between rotational lines of N 2 .On the other hand, soft X-rays produce photoelectrons with sufficient energy to cause collisional ionization (Solomon & Qian, 2005), thereby promoting the importance of higher energy photons relative to long wavelength photons, which cannot produce ionizing photoelectrons.While the increase in the electron density with the HR model calculations does help address the discrepancy in the magnitude of the EDPs, the discrepancy in the EDP peak altitude is puzzling.A simple test scaling the X-ray solar flux (2-20 nm) upward as suggested by Solomon et al. (2001) and Titheridge (2003); Titheridge (2000Titheridge ( , 1997) ) does not aid in resolving the discrepancy in altitude of the lower E-region ionization (and thus EDP) peak.The scaling only resulted in an upward shift of 1 km in the altitude of the lower photoionization peak.This demonstrates that scaling the solar X-ray flux would simply address differences in EDP magnitude but not altitude.The apparent loss of the lower E-region peak when using high-resolution inputs may also be attributed to broadening and an upward shift in the peak altitude of the O + 2 ion production rate; however, the cause for this phenomenon is not currently understood.Therefore, is it also necessary to incorporate high-resolution O 2 cross sections in model calculations to properly account for the total ionization rate profile, and thus electron density, with respect to altitude.We are currently investigating this issue, which ultimately will be incorporated for completeness into AURIC and be made available for other modeling codes.

Appendix A: Computational Methodology
In this work, the R-matrix method is used to treat photoionization of atomic oxygen and molecular nitrogen.Specifically, e O + and e N + 2 collision problems are calculated by the R-matrix method to extract the photoionization cross sections.In the R-matrix method of photoionization we have to treat the problem of an electron collision with an ion in order to extract the initial state wavefunction, where an electron is bound to the ion, and the final state wavefunction, where an electron is moving in the outward direction from the ion.The amplitude of photoionization is then obtained by calculating a dipole transition moment between these initial and final state wavefunctions.The R-matrix theory of electron-atom and electron-molecular collisions (Burke, 2011;Burke & Berrington, 1993;Burke & Robb, 1976) is based on the variational principle, in which the electron configuration space is divided into inner and outer regions.The boundary between the inner and outer regions is known as the R-matrix boundary radius.This radius should be large enough to contain all the electron density of the target atom or molecule.In the outer region, the scattering process is approximated as single electron dynamics, neglecting the exchange of target and scattering electrons.In the inner region, the dynamics of all the electrons in the system, the target electrons plus the scattering electron, are treated as in usual quantum chemistry computations.This means the exchange effect of target and scattering electrons and some degree of correlation effect are properly considered in the inner region.In this method, we first solve the inner region problem which is equivalent to quantum chemistry computation within a restricted space, where the system is confined within a box radius, and its eigenstates form a discrete basis.These eigenstates carry information of the scattering electron.In addition, the wavefunctions of the bound states, that is, neutral O atom or N 2 , are also obtained from these eigenstates.Then, the scattering problem in the outer region is solved to obtain wavefunctions of the outgoing photoelectrons, based on the eigenstates calculated in the inner region.Photoionization cross sections are calculated by evaluating dipole transition moments between the wavefunctions of the bound states obtained in the inner region, and the wavefunctions of the photoelectrons constructed in the outer region calculation.
The target states included in the scattering calculations are represented by configuration-interaction (CI) wavefunctions.The target and continuum orbitals are orthogonalized using Schmidt orthogonalization (Leon et al., 2013).The continuum molecular orbitals are then orthogonalized among themselves using symmetric or Lowdin orthogonalization to remove the linearly dependent functions.
Inside the R-matrix sphere, the inner region, the total wavefunction for a given symmetry is expanded in basis states.The (N + 1) scattering system, where N is the number of electrons in the target atom or molecule, in the inner region is represented by a CI-type basis expansion; where A is an antisymmetrization operator, x p is the coordinate of the pth electron, ϕ N i is the ith target state, ξ j is the continuum orbital basis of the scattering electron, and k represents a particular R-matrix basis function.The continuum orbital basis, unlike a bound orbital basis function, does not vanish on the boundary.X m are (N + 1) electron correlation functions constructed from the bound orbitals to ensure the completeness of the basis function when the continuum orbital basis is orthogonalized to the bound electron basis.It further allows us to include correlation effects that arise from virtual excitations to higher electronic states.The coefficients a ijk and b mk are determined by matrix diagonalization, resulting in the R-matrix eigenstates in the inner region.
The standard R-matrix method has also been extended to include the effects of coupling of the bound states to the target continuum, using the R-matrix pseudo state (RMPS) method (Bartschat et al., 1996;Burke, 2011;Gorczyca & Badnell, 1997;Mitnik et al., 1999).In our version of the RMPS method, used in the photoionization of atomic oxygen, the target continuum states are represented by a set of Laguerre pseudo states.Here pseudo states are nonphysical excited electronic states included in the model to improve a calculation, that is, to describe ionization of the target in this case.Details of the R-matrix calculation for valence photoionization on atomic oxygen are similar to previous work by Meier et al. (2007).We used 34 O + states as target states in the close-coupling expansion, obtained from the configurations 2s 2 2p 3 , 2s2p 4 , and 2s 2 2p 2 3l (l = s, p and d).The K-shell photoionization of atomic oxygen was treated as in the work by McLaughlin et al. (2013), where additional 1s-hole configurations were considered, resulting in 910 levels.The radius of the R-matrix sphere was taken to be 7.3 a 0 , where a 0 is the Bohr radius.We neglect relativistic effects in our work and perform all calculations in LS coupling.All atomic Rmatrix results were obtained using the Belfast suite of codes (Berrington et al., 1987(Berrington et al., , 1995;;Burke & Berrington, 1993).
The R-matrix calculation for the photoionization of the nitrogen molecule was performed based on target molecular orbitals obtained by the Complete Active Space Self Consistent (CASSCF) calculation on N + 2 molecular ion using the Molpro suite of codes (Knowles & Werner, 1985;Werner et al., 2012;Werner & Knowles, 1985).We used the cc-pVTZ basis set (Dunning, 1989) and full-valence active space for the CASSCF calculation.The continuum orbitals up to g (l = 4) partial waves were represented by Gaussian-type molecular orbitals, centered at the center of gravity of the molecule.The radius of the R-matrix sphere was taken to be 10 a 0 .All molecular calculations were performed in the fixed nuclei approximation, using a modified version of the UK Molecular Rmatrix Scattering package (Morgan et al., 1998;Tashiro, 2010).The details of the method and calculations are described in Tashiro (2010).

Figure 2 .
Figure 2. O to O + [2s2p 3 4 S 0 ] Photoionization Cross Sections versus Wavelength.(Left) R-matrix calculations on the native variable grid (R-matrix) and the low-resolution Conway (1988) compilation (LR) are shown.(Middle) The LR data interpolated onto the fine wavelength grid (ILR) and the R-matrix data interpolated onto the fine wavelength grid (HR) are shown.(Right) The LR and the R-matrix data binned onto the coarse wavelength grid (BHR) are shown.Please refer to Table1for more details.

Figure 1 .
Figure 1.Atomic Oxygen Photoionization Cross Sections.(Left) The total and partial R-matrix photoionization cross sections interpolated onto the fine wavelength grid (HR cross sections) versus wavelength.(Right) The high-resolution Rmatrix photoionization cross sections binned onto the coarse wavelength grid (BHR cross sections) versus wavelength.The six dominant ionized states are shown in addition to the photoabsorption cross section (blue) and a pseudo state that incorporates the cross section of the remaining 29 partial states (orange).

Figure 9 .
Figure 9. O to O + Pseudo State Photoionization Cross Sections versus Wavelength (Left) Native grids, (Middle) fine wavelength grid, and (Right) coarse wavelength grid.Only cross sections derived from the R-matrix data are shown in red as the 29 partial states comprising the pseudo state were not included in the Conway (1988) compilation.

Figure 13 .
Figure 13.N 2 Photoabsorption Cross Section.N 2 photoabsorption cross section constructed from the Bishop et al. (2007) model and digitized Carter (1972) interpolated onto the coarse wavelength grid (top) and the fine wavelength grid (bottom).

Figure 15 .
Figure 15.NRLEUV Solar Minimum Spectrum.(Top) The original high-resolution NRLEUV solar spectrum delivered by H. Warren is shown in red.The black solid line is the triple smoothed (at wavelengths > 46.8 nm, non-thermal region) highresolution solar spectrum (fine grid with 0.001 nm bins).(Bottom) The NRLEUV high-resolution solar spectrum is shown in red while the binned (coarse grid with 0.05 nm bins below 10 and 0.1 nm bins above 10 nm) solar spectrum is shown in black.

Figure 16 .
Figure 16.Modified Solar Minimum Spectrum.The final solar spectrum for minimum (F 10.7 = 70) solar activity.The solar spectra are derived by reconfiguring the NRLEUV delivered minimum spectrum onto the fine and coarse wavelength grids.The spectrum is then scaled using Lean's model(Lean et al., 2003(Lean et al., , 2011(Lean et al., , 2020) ) for the specific F 10.7 radio flux noted in the panel (see Section 3).The final fine (red) and coarse (blue) wavelength grid solar spectra are used in model calculations.

Figure 17 .
Figure 17.O + Photoionization Production Rates & Ratios.(Top) Photoionization production rates versus altitude for the dominant states in atomic oxygen.AURIC volume production rates using the HR (purple solid line) and BHR (yellow dashed line) cross sections are shown with the model results from the ILR (blue solid lines) and LR (red dashed lines) cross sections.Corresponding ratios of the volume production rates to the HR cross section case versus altitude.For reference, the ratio of the ILR/LR results are shown (solid black line).Ratios of photoionization rates of the ILR (solid blue line), BHR (dashed yellow line), and LR (dashed red line) cases to the HR case are shown.

Figure 18 .
Figure 18.N + 2 Photoionization Production Rates & Ratios for X, A, and B States.(Top) Photoionization production rates versus altitude for the dominant states in molecular nitrogen utilizing the newly calculated N 2 cross sections.AURIC volume production rates using the HR (purple solid line) and BHR (yellow dashed line) cross sections are shown with the model results from the ILR (blue solid lines) and LR (red dashed lines) cross sections.(Bottom) Corresponding ratios of the volume production rates to the HR cross section case versus altitude.For reference, the ratio of the ILR/LR results are shown (solid black line).Ratios of photoionization rates of the ILR (solid blue line), BHR (dashed yellow line), and LR (dashed red line) cases to the HR case are shown.

Figure 19 .
Figure 19.Total Photoionization Production Rates & Ratios for each Species.(Left) Volume production rate results from AURIC for the various permutations of cross section, solar spectrum, and wavelength grid inputs for O + (top), N + 2 (middle), and O + 2 (bottom).(Right) Volume production rate ratios for O + (top), N + 2 (bottom), and O + 2 (bottom).These are the ratios with respect to the HR volume production rates.The ratio of LR and ILR is shown in black.Unity is shown in green.

Figure 20 .
Figure 20.Total Photoionization Rate & Production Rate Ratio.(Left) The total volume production rate for all four input cases for AURIC.(Right).The corresponding total volume production rate ratio from AURIC with respect to the updated high-resolution cross section inputs on the fine wavelength grid.

Figure 21 .
Figure 21.Total Photoionization Production Rates & Ratios using the Model of Meier et al. (2007).(Top) Photoionization production rates as a function of altitude for O, N 2 , O 2 , and the total.The unbinned HR cross section results are shown in black.The LR cross section results, with a bin size of 0.1 nm, are shown in red.(Bottom) Photoionization rate ratios for O, N 2 , O 2 , and the total as a function of altitude are shown in red and unity is shown in black.The ratio represents the total production rate of the species (or total) from the binned LR cross sections/corresponding total production rate from the unbinned HR cross sections.
aid in validating AURIC model results.Results for O + , N + 2 , and O + 2 demonstrate similar VPR magnitudes, overall shape of the profile, and ionization production rate ratios (Figures 19 and 21).As discussed in Sections 4.1 and 4.2, the results between AURIC and the updated Meier et al. photoionization code are in relatively good agreement with the exception of the larger increase in the total O + VPR from the Meier et al. code at 137 km.The wider ionization peak, with larger VPRs extending to lower altitudes may account for the 500% versus 300% increase observed in the Meier et al. results.Both the O + and O + 2 ionization rate ratios drop below 0.4 in the 110-150 km altitude regime while