Milankovitch cycles as a driver of climate variation and fluvial erosion on early Mars before 3.8-3.6 Ga

Abstract

Geomorphological evidence suggests that early Mars had oceans and valley networks, which implies that it had a dense atmosphere and an active hydrological cycle. However, the effects of orbital obliquity cycles remain largely unexplored. We performed fully coupled GCM simulations with a 2 bar CO₂ atmosphere and an initial 500 m global ocean, varying obliquity (40°±10°) and H₂ concentration (from 0 to 6%) for 1.2×10⁵ years. The results show that obliquity and H₂ significantly influence the climate by controlling surface temperatures and precipitation patterns. Ice sheet growth and melt driven by these variations supplied water for runoff. Our findings suggest that river formation was closely linked to subglacial melting and rain-fed flows, which were modulated by Milankovitch-scale cycles.

Introduction

Geomorphological evidence suggests that early Mars (~3.8-3.6 billion years ago) had liquid water and valley networks (VNs) in a wetter climate [1-4]. Two main hypotheses have been proposed: a warm, wet Mars characterized by rivers fed by rainfall [1-5], and a cold, icy Mars with meltwater from ice sheets [6-8]. Recent climate modelling shows that greenhouse warming by CO₂-H₂ CIA could explain the warming that occurred under high H₂ concentrations [9-16]. However, the role of Milankovitch cycles, especially obliquity variations, in shaping the early Martian climate and VN formation remains poorly understood. In this study, we examine the impact of obliquity-driven climate changes using a fully coupled model of the atmosphere, hydrosphere and cryosphere. This study will be the first attempt to provide a detailed, quantitative assessment of how variations in obliquity may have formed the Martian landscape over hundreds of thousands of years.

Methods

We used three coupled numerical models to simulate early Mars climate evolution: the Paleo-Mars Global Climate Model (PMGCM) [12], the Catchment-based River Simulator (CRIS) [13], and the Accumulation and Ablation of Large-scale ICE-sheets (ALICE) [17]. PMGCM simulates atmospheric and surface processes with ~5.625° resolution and 15 vertical layers up to 60 km. CRIS, coupled to PMGCM, resolves fluvial and sediment transport at ~1.125° resolution. ALICE simulates large-scale ice sheet dynamics and thermodynamics at the same resolution, using a 1 Mars year time step for mass continuity and 10 Mars years for thermal evolution. The early atmosphere was assumed to contain 2 bar of CO₂ with H₂ mixing ratios from 0–6%, and a solar constant of 441 W/m² representing 75% of present-day Mars [18]. Ancient topography prior to true polar wander was applied [19], with geothermal heat flux set to 55 mW/m² [20]. To simulate Milankovitch-scale climate change, we iteratively ran PMGCM–CRIS for 30 million years, followed by ALICE for 15,000 years. This sequence was repeated across one obliquity cycle (1.2×10⁵ years), using sinusoidal variation around a mean obliquity of 40°±10°, with eccentricity neglected to isolate obliquity effects.

Results

With 0% H₂, surface temperatures remained below freezing globally, resulting in significant snow and ice accumulation, particularly in the southern highlands. Increasing the H₂ concentration to 3% produced greenhouse warming, enabling seasonal melting in the low to mid latitudes, where temperatures approached 273 K during low obliquity phases. At 6% H₂, the global mean temperature exceeded freezing, enhancing the potential for rainfall and runoff.

Obliquity variations significantly influenced the spatial distribution of temperature and precipitation. High obliquity increased polar insolation and promoted equatorward water transport, whereas low obliquity resulted in equatorial warming and polar ice melt. These changes, combined with higher H₂ concentrations, increased global precipitation. Ice sheets formed more extensively under variable obliquity due to intensified snowfall and melt cycles. Their melting contributed to surface runoff, particularly in the 3% and 6% H₂ scenarios.

River discharge patterns reflected these climatic shifts. At 3% H₂, simulated rivers aligned well with observed valley networks (VNs), particularly under varying obliquity. At 6% H₂, the highest VN coverage (up to 70.6%) occurred under time-varying obliquity. These results suggest that variations in obliquity driven by Milankovitch cycles enhance climate variability, promoting both ice melt and rainfall-fed runoff. Martian rivers probably formed from a mixture of subglacial meltwater and precipitation, influenced by orbital forcing and atmospheric composition.

Summary and Conclusions

We investigated the long-term climate evolution of early Mars, from the late Noachian to the early Hesperian, using a coupled framework comprising global climate (PMGCM), river (CRIS) and ice sheet (ALICE) models. We performed simulations over a period of 1.2×10⁵ years, considering obliquity cycles (30°–50°), a 2 bar CO₂ atmosphere with H₂ mixing ratios of 0–6%, and ancient topography incorporating ocean and lake reservoirs. Our results demonstrate that variations in obliquity significantly impacted surface temperature, precipitation, and ice sheet dynamics. High obliquity increased polar insolation, while low obliquity favored equatorial warming. H₂-driven greenhouse warming increased global temperatures and precipitation, facilitating ice sheet melting and river formation. At 3% H₂, rivers aligned well with observed valley networks (VNs), whereas at 6%, extensive rainfall and meltwater enhanced runoff and fluvial activity. These findings suggest that rainfall and subglacial melt both contributed to VN formation under dynamically evolving climates.

References

[1] Carr (1995), [2] Hynek et al. (2010), [3] Di Achille and Hynek (2010), [4] Citron et al. (2018), [5] Craddock and Howard (2002), [6] Wordsworth et al. (2013), [7] Fastook et al. (2015), [8] Galofre et al. (2020), [9] Ramirez et al. (2014), [10] Wordsworth et al. (2017), [11] Wordsworth et al. (2021), [12] Kamada et al. (2020), [13] Kamada et al. (2021), [14] Jorge et al., 2024, [15] Steakley et al. (2023), [16] Adams et al. (2025), [17] Kamada et al. (2022), [18] Gough (1981), [19] Bouley et al. (2016), [20] Solomon et al. (2005)