New surface composition and temperature maps of Pluto from New Horizons LEISA data

Abstract

New Horizons RALPH/LEISA near-infrared spectra allow for regional mapping of Pluto's surface ices and their physical state; however, scan-to-scan artifacts and variable spatial resolution complicate quantitative interpretation. We extend previous LEISA compositional studies (Protopapa et al., 2017(arXiv:2110.00662); Schmitt et al., 2016; Emran et al., 2023(arXiv:2301.06027)) by combining five close-approach observations into co-registered equirectangular maps at 7 km/pixel and by jointly retrieving surface temperature, ice abundances, and grain sizes using a Hapke-based mixing model. We mitigate bad pixels and edge overexposure linked to flat-field uncertainties and correct for residual scan-to-scan spectral discrepancies using per-observation scale and offset terms. The resulting maps provide distributions of CH4-rich ice, N2-rich ice, H2O ice, and Titan tholins, alongside a corresponding temperature map.

Figures (10)

• Figure 1: Example of preprocessing and flat-field correction for the closest LEISA scan (lsb_0299176809_0x53c_sci). Panel A shows the original calibrated frame (reflectance level still affected by edge overexposure), while panel B shows the same frame after bad-pixel interpolation, offset field subtraction, and the empirical flat-field correction described in Section 2.3. Panels C and D show the original flat field $f(x,y)$ and the corrected flat field $f^*(x,y)=f(x,y)e^{\beta(x,y)}$, respectively; the strongest adjustments occur near the left detector edge, where the overexposure artifact is most pronounced.
• Figure 2: Frame-averaged moving standard deviation along the x-axis using a sliding window of 7 pixels for the LEISA scan (lsb_0299176809_0x53c_sci). Panel A shows frame-averaged moving standard deviation ($\sigma_{shifted}$) for the scan after bad-pixel correction and offset field subtraction, while panel B shows the same quantity ($\sigma_{corrected}$) after the flat-field correction. Panel C shows the ratio $\frac{\sigma_{shifted}}{\sigma_{corrected}}$ averaged along the FITS y-axis, indicating that the flat-field correction strongly reduces large values of $\sigma$ in overexposed regions (e.g., by 5--35% near the left edge at $x<20$, and by 5--15% near $x\approx150$ and at the right edge), while slightly reducing $\sigma$ everywhere by $\sim$2.5% on average due to smoothing of the flat field.
• Figure 3: Example of map processing from LEISA scan. Panel A shows FITS file (lsb_0299172014_0x53c_sci) after preprocessions described in section 2.3, panel B shows equirectangular map projection, obtained from FITS file by applying ISIS USGS utility and map corrections, described in section 2.4. Panel C shows a set of measurements for one pixel in projection map. From 5 FITS files (Table \ref{['tbl2']}) we have 5 maps and so 5 spectra for one pixel. However, due to the fact that the maps cover different parts of Pluto, the count of spectra for different pixels vary from 0 to 5.
• Figure 4: Illustration of scan-to-scan spectral level mismatch and the role of per-observation scale/offset terms. Blue: measured reflectance spectrum for an example surface element; red: best-fit model spectrum. (A) Fit without allowing per-scan scale and offset parameters ($S_j,O_j$), showing a systematic overestimation of the measured level for this scan. (B) Fit after solving for ($S_j,O_j$), which removes the bulk of the level mismatch while preserving absorption-band structure. Similar level offsets are observed across many pixels in the closest scan, motivating the inclusion of ($S_j,O_j$) as nuisance parameters to isolate compositional/temperature information from calibration artifacts.
• Figure 5: Goodness-of-fit (RMS) maps highlighting where additional parameters improve the spectral FITS and where residual artifacts remain. (A) RMS map for the full model, including fractional area, effective particle diameter, temperature, and per-scan scale/offset parameters ($F_i,D_i,f,T,S_j,O_j$). (B) Same as (A) but with temperature fixed (40K, no $T$ retrieval), showing a marked RMS increase (3.5%) in the methane-rich northern terrains, consistent with temperature-sensitive band shapes contributing to the fit quality there. (C) Same as (A) but without allowing per-scan scale/offset ($S_j,O_j$), demonstrating that scan-to-scan level mismatches dominate the RMS in many regions and especially near the edges of the closest scan (lsb_0299176809_0x53c_sci) and in scan overlap boundaries. $\phi$ and $\lambda$ are North latitude and East longitude, respectively.