Producing High-Resolution Martian Surface Temperature Maps Using VIR – TIR Relationships

Sep 10, 2026

Upload by Anda

Thermal infrared data (TIR; 8–15 μm) supports Earth and planetary remote sensing, including derivation of Martian thermal inertia (TI), which reflects surface physical properties and underpins geologic interpretation, ISRU environment assessment, and mission planning. TI from THEMIS is limited to 100 m/pixel, while CRISM VIR hyperspectral data (0.5–5 μm) provides composition at 12 m/pixel. A machine-learning regressor is trained to constrain THEMIS TI–CRISM VIR relationships and predict TI from CRISM spectra, achieving R2≈0.90 and RMSE≈23.6 TIU. The model generates a downscaled TI map at 12 m/pixel, refining decametre-scale features unresolved in THEMIS.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages9

Reading Time23 min

AccessFree to download

Abstract

Keywords

Nomenclature

Acronyms/Abbreviations

1. Introduction

1.1. Thermal data

1.2. Thermal inertia

2. Data and Methods

3. Regression Model

4. Results

5. Discussion

76th International Astronautical Congress (IAC 2025), Sydney, Australia, 29 Sep-3 Oct 2025. Copyright ©2025 by the International Astronautical Federation (IAF) . All rights reserved. IAC-25-A3.3B.8.x96743 Producing High-Resolution Martian Surface Temperature Maps Using VIR – TIR Relationships Michael A. Frazera*, Eriita G. Jonesa, Katarina Miljkovica, Gretchen Benedixa a Space Science and Technology Centre, School of Earth and Planetary Science, Curtin University, Kent St, Bentley, Perth, Australia 6102 * Corresponding Author Abstract Thermal infrared data (TIR; 8 – 15 μm) has a wide range of applications in Earth and planetary remote sensing. On Mars, this includes deriving thermal inertia (TI), which describes surface physical characteristics (e.g. particle size, degree of cementation) and is key for understanding geologic processes, assessing in-situ resource utilisation (ISRU) environments, and assisting mission planning. However, TI data from the THEMIS instrument is limited to 100 m/pixel resolution. Hyperspectral visible and near-infrared data (VIR; 0.5 – 5 μm) compliments TIR data by providing information on surface composition and is provided by the CRISM instrument at 12 m/pixel. In this work, we generatea machine learning regressor-based model to constrain relationships between THEMIS TI and CRISM VIR images at THEMIS resolution, and predict TI values from CRISM spectra with high accuracy (R2 ~ 0.90, RMSE ~ 23.6 TIU) . We use the model to produce a downscaled TI map at a spatial resolution of 12 m/pixel, an order of magnitude finer than currently available, revealing decametre-scale features previously unresolved in THEMIS data. Keywords: Mars, remote sensing, LST, downscaling, machine learning Nomenclature VIR – Visible infrared; 0.4 – 1.2 μm SWIR – Shortwave infrared; 1.2 – 4.5 μm TIR – Thermal infrared; 4.5 – 15 μm TIU – Thermal inertia units; J m-2 K-1 s-1/2 Acronyms/Abbreviations CRISM – Compact Reconnaissance Imaging Spectrometer for Mars THEMIS – Thermal Emission Imaging System CTX – Context Camera MRO – Mars Reconnaissance Orbiter MTRDR – Map-projected Targeted Reduced Data Record 1. Introduction Satellite-derived remote sensing observations date back to the mid-1900s, with the Luna 3 spacecraft returning the first images of the dark side of the moon during a flyby in 1959, and TIROS-1 returning the first footage of Earth from space a year later (Fig. 1) [1] . As of 2025, there are over 300 Earth Observation (EO) satellites currently in orbit [2], and dozens more spacecraft throughout the wider the solar system. Since then, instruments on board have improved dramatically, with higher resolution sensors, new techniques and increased temporal coverage [1] . Some trade-offs do remain, however, with different modes of remote sensing (e.g. visible, infrared, hyperspectral, LiDAR, SAR) each providing different spatial and temporal resolutions. This has driven the development of data fusion in remote sensing, in which two unique, separately acquired data sets of the same physical Figure 1: First television picture from space, taken April 1, 1960 by the TIROS-I satellite. Credit: NASA. environment are synthesised into one which is ‘greater than the sum of its parts’. Data fusion has a range of applications in the EOspace [3] . In this work, we employ data fusion to combine high-resolution VIR-SWIR hyperspectral data with medium-resolution TIR-derived data from separate instruments in orbit around Mars. Section 1.1. introduces thermal data, 1.2. details the applications and calculation of thermal inertia, 1.3. describes the limitations of thermal data in terms of spatial resolution, and 1.4. is a summary of data fusion and downscaling in the EO and planetary science space. Section 2. describes the data that we use, and section 3. details the regression model that we train and apply on the data. Sections 4. and 5. detail our results and discussions respectively. IAC-25-A

FAQs

Why is Mars thermal inertia (TI) important for remote sensing applications?

TI characterizes surface physical properties and helps interpret geologic processes, evaluate ISRU environments, and support mission planning.

What limitation exists for THEMIS TI data, and how does CRISM help address it?

THEMIS-derived TI is limited to 100 m/pixel, while CRISM provides hyperspectral visible/near-infrared information at 12 m/pixel that complements TIR data.

How does the proposed machine-learning approach improve TI mapping resolution?

A machine-learning regressor learns relationships between THEMIS TI and CRISM VIR images, predicts TI from CRISM spectra, and produces a downscaled TI map at 12 m/pixel to reveal finer decametre-scale features.

Sep 10, 2026

Upload by Anda