Lunar Trailblazer: Improving Brightness Temperature Estimation Methods and Applications of Temperature Retrieval for Future Missions
(2025)
Authors:
Fiona Henderson, Namrah Habib, Katherine Shirley, Neil Bowles
Abstract:
Introduction: The Lunar Thermal Mapper (LTM) is a multispectral infrared radiometer, built by the Oxford Physics Instrumentation Group for the Lunar Trailblazer mission; a small satellite launched in February 2025 under NASA’s Small Innovative Missions for Planetary Exploration (SIMPLEx). Trailblazer aims to advance our understanding of the lunar water cycle by mapping surface temperature, water abundance, distribution and form (OH, H2O, ice) and silicate lithology (i.e., Si-O Christiansen spectral feature). LTM was developed to improve upon existing infrared instrumentation in lunar orbit (e.g., Diviner Lunar Radiometer Experiment, hereafter referred to as Diviner) to provide higher resolution temperature estimations and refine interpretations of thermophysical properties at the surface [2, 3]. Accurately determining surface temperatures on airless bodies is essential for deriving emissivity spectral features (such as the Christiansen Feature and Restrahlen bands, which are diagnostic of silicate lithologies) that are representative of the surface. Temperature errors can affect spectral shape, resulting in the misidentification of surface composition [5, 8]. Our team compared six methods for estimating LTM’s brightness temperature (BT), including the temperature retrieval approach used by Diviner, to (1) determine which method provides the most representative surface temperature and (2) assess how variations in BT estimation affect derived emissivity spectral shape. Despite challenges facing the Trailblazer mission, refining methods for BT estimation remains relevant to the planetary community, as future missions continue to depend on infrared instrumentation and accurate BT retrievals for remote compositional interpretation (e.g., LEAP, L-CIRiS, Europa Clipper). Instrumentation: LTM is a 15-channel infrared imager that covers a range between 6 to 100 µm [2,3]. LTM advances infrared compositional analysis by incorporating eleven narrowband compositional filters across the 6.25 to 10 µm range. This expanded spectral coverage enables more precise characterization of key features, such as the Christiansen Feature, Reststrahlen bands, and transparency features, which are essential for identifying spectral endmembers (Table 1) [2,3]. LTM builds upon Diviner, a nine-channel instrument that has a broad spectral range from 0.3 to 400 µm (Table 1) [1]. Diviner’s three narrowband compositional channels, 7.55–8.05 µm (Channel 3), 8.10–8.40 µm (Channel 4), and 8.38–8.60 µm (Channel 5), are specifically tuned to capture the Christiansen Feature (CF), an emissivity peak that is diagnostic of broad silicate mineralogy and sensitive to variations in silica content [1,4]. Table 1: LTM and Diviner observational parameters. Methodology: To assess BT and emissivity retrieval techniques for LTM, we measured four lunar analog samples under controlled laboratory conditions to retrieve high-resolution emission spectra. These laboratory spectra were down-sampled to match LTM’s narrowband spectral resolution. Six BT estimation methods were tested to determine how effectively each method preserved laboratory spectral shape and temperature. The following section describes the laboratory setup and the BT estimation methods examined in this study. Laboratory: Using the PASCALE (Planetary Analogue Surface Chamber for Asteroids and Lunar Environments) in conjunction with a Bruker 70V Fourier Transform Infrared (FTIR) spectrometer, we conducted thermal infrared measurements of four volcanic lunar analogue samples; dunite (Twin Sisters -1 and -2), basalt (BIR-1) and rhyolite (RGM-1) under controlled ambient conditions (350 K, 1000 mbar, N2 atmosphere) [4]. The integration of PASCALE with FTIR allows for the acquisition of thermal emission spectra (as opposed to typical laboratory reflectance), offering a more representative analog of data collected by orbiting infrared instrumentation. Spectra were measured across ~6000 to 350 cm⁻¹ at a resolution of 4 cm⁻¹. Quality assurance and calibration procedures followed established protocols outlined in [6,7,8]. BT Estimations: To evaluate BT performance at LTM’s spectral resolution, each sample’s measured radiance was convolved with LTM’s filter response to simulate instrument-resolution radiance. The resulting spectra were converted to BT using the Planck function. Seven distinct methods were applied to the LTM-resolution BT data to determine the maximum BT values for each sample (Table 2). Emissivity was subsequently derived as the ratio between the observed LTM-resolution radiance and an ideal blackbody at the retrieved maximum BT for each method across all samples. The accuracy of the BT estimation methods was assessed by comparing the resulting emissivity spectra and maximum BT values to the full laboratory reference data (350K and full resolution emissivity). Additionally, a focused comparison with Diviner’s BT retrieval method was conducted to identify method-specific discrepancies and evaluate cross-instrument consistency.Table 2: BT estimation methods Results & Discussion: Six BT estimation methods were applied to laboratory emissivity spectra of four lunar analogue samples (dunite, basalt, and rhyolite), as shown in Figure 2. The associated standard errors (SE) for each method are reported in Table 3. Among the tested approaches, four methods (3rd degree polynomial, quadratic, spline and narrowband maximum) showed close agreement with high-resolution laboratory spectra (Figure 2). Temperature variations across compositions were minor, with low SE values (Table 3). Since the spline fit did not significantly outperform the simpler polynomial or narrowband methods, lower complexity approaches are preferred for LTM temperature retrievals, with a maximum SE of 3.42%.In contrast, due to limited spectral sampling, the Diviner method underestimates surface temperatures by up to 19 K (SE max: 5.55%) in the dunite (TS-2) sample. Expanding this analysis to include a broader range of lithologies or impacted processed samples would help assess whether the Diviner approach (and potential other methods with sparse spectral sampling) introduce systematic shifts in the Christiansen Feature (CF) position or affect the spectral shape relative to more spectrally resolved techniques. Table 3: BT estimations and associated SE of temperature for dunite (TS-1, TS-2), basalt (BIR-1) and rhyolite (RGM-1). Fig 2: Six BT methods are fitted to laboratory emissivity spectra of four lunar analogues. Conclusion:Comparisons between BT estimation methods indicate the 3rd-degree polynomial, quadratic, and narrowband maximum methods offer the best agreement with laboratory data (SE max: 3.42%). Although Diviner’s method tends to underestimate surface temperatures (up to 19 K), it still preserves spectral shape and wavelength range, supporting the reliability of compositional interpretations. Expanding the dataset to include a broader range of compositions could confirm whether different approaches result in systematic shifts in the Christiansen Feature across different lithologies. This work enhances the accuracy of remote compositional interpretation and supports future exploration on airless bodies.
