Ground magnetic survey on Mars from the Zhurong rover

Measurements of crustal magnetic fields of terrestrial planets provide key constraints on dynamo history and interior thermal evolution. Mars presently lacks a global field but exhibits heterogeneous crustal magnetization, implying an ancient dynamo. Prior orbital surveys (MGS, MAVEN) mapped Mars’ magnetic field at altitudes of ~150–400 km, resolving wavelengths >100 km and showing localized strong anomalies up to ~1,500 nT. At the other end of the spectrum, palaeomagnetic analyses of Martian meteorites (e.g., ALH84001) suggest a dynamo before ~4.1–3.9 Ga. However, the origin, distribution and timing of crustal magnetization remain open questions, and high-resolution regional (metre–kilometre scale) measurements are needed. NASA’s InSight lander measured a very strong surface field (~2,000 nT) at a single site, but there has been no kilometre-scale ground magnetic survey on Mars. The Zhurong rover, which landed in southern Utopia Basin on 15 May 2021, carried the RoMAG instrument and conducted a regional ground magnetic survey to characterize crustal fields at metre–kilometre scales, constrain subsurface magnetization and implications for the Martian dynamo.

Global and regional orbital magnetic mapping by MGS and MAVEN has provided maps at 90–400 km altitudes, revealing crustal anomalies up to ~1,500 nT but limited in spatial resolution (wavelengths comparable to altitude). Downward continuation of orbital models (e.g., Langlais et al. 2019, L19) is commonly used to estimate surface fields but is known to be unreliable at a point due to amplification of high-degree errors and the omission of shorter-wavelength contributions that dominate near the surface. Palaeomagnetic measurements of Martian meteorites (ALH84001) support an early dynamo (>3.9–4.1 Ga). InSight’s ground magnetometer observed strong surface fields (~2,000 nT) at Elysium Planitia, interpreted as strong small-scale crustal magnetization not resolved from orbit. Impact demagnetization at large basins (e.g., Utopia, Isidis) is inferred from weak orbital signatures; models indicate large demagnetized radii and potential remagnetization or subsequent volcanism. These studies collectively motivate and contextualize the need for in situ, kilometre-scale ground surveys to link orbital-scale anomalies to local geology.

Instrument and survey: Zhurong’s RoMAG payload comprises two identical tri-axial fluxgate magnetometers (±65,000 nT range; noise ~0.01 nT/√Hz at 1 Hz): one rotatable sensor at the mast top and another fixed at the mast base. During 90 sols (4 June–3 September 2021), RoMAG made vector measurements at 16 locations along a 1,089 m traverse within a ~5 km radius ghost crater in southern Utopia Basin (centre ~25.12° N, 109.97° E). Only horizontal components are analyzed because the vertical component cannot be calibrated via the rotation scheme. Calibration and separation of fields: Measured fields comprise ambient Martian field (internal crustal plus external ionospheric/magnetospheric), rover mast field, and rover body field (including solar array currents). A two-step calibration was used. Step 1: Mast yaw rotations at each site isolate the combined horizontal field (ambient plus body) by fitting the circular hodograph in the rotating sensor frame (least-squares solution for B_cx, B_cy and mast offsets). Step 2: Rover yaw rotations (performed during a dedicated along-track calibration on 5 Jan 2022) decouple the ambient field from the body field by comparing measurements at different rover headings using the Mars-to-rover rotation matrix. The along-track calibration yielded a reference rover body field and ambient field at that time. Body field and thermal/current corrections: Body field variations across sites due to differing temperatures and solar array currents were modeled via multiple linear regression of each component against sensor temperature (Ts), electronics temperature (Te), and solar array current (I1). The strong linear dependence (correlation >0.99; RMS ~2.2–2.4 nT for horizontal components) allowed correction of body field differences between the along-track calibration and each of the 16 site measurements using ΔI1 and fitted coefficients. Temperature effects during mast rotations are largely absorbed in Step 1, as they are orientation independent over the short rotation intervals. External (ionospheric) field estimation: External contributions were estimated using InSight surface magnetometer observations and an ionospheric dynamo model constrained by MAVEN background fields and Mars Climate Database parameters. Modeled mean horizontal field at Zhurong is ~83% of that at InSight for Ls = 54–95°. Mean InSight diurnal horizontal components between 10:00–16:00 LTST (the Zhurong observation window) at matching Ls were scaled by 0.83 and subtracted from the calibrated Martian field to estimate the crustal field at each site. Uncertainties were propagated from fitting errors, current-correction uncertainties, and external field variability (standard deviations from InSight-derived estimates). Data processing and uncertainties: The combined field B_c at each site was obtained from mast rotations (circular fit errors quantified as RMS deviations). The rover body field B_body was obtained from the along-track calibration and corrected for ΔI1 at each site. The ambient field in Mars geographic NEC frame was calculated via coordinate transforms. Final crustal fields B_crust were obtained by subtracting the external field estimates. Error propagation accounted for uncertainties in B_c, B_body (including ΔI1 corrections), covariance terms, and external field estimates. Modelling and orbital comparison: Predictions from the L19 orbital crustal field model were downward continued to altitudes 150, 100, 50, and 0 km to compare with Zhurong’s site. The area contributing significantly to the orbital model’s surface estimate has radius ~altitude (~150 km), far exceeding Zhurong’s local sampling scale. A forward model representing a weakly magnetized uniform Hesperian lava layer (1,000 × 1,000 × 1 km with M = [0, 0.1, 0.1] A m−1) containing a cylindrical, unmagnetized hole (ghost crater, 10 km diameter, 1 km thickness, top at 200 m depth) was constructed using polygonal prism integration to evaluate expected surface fields. Geologic context: Stratigraphic constraints from crater statistics, rover radar, and regional mapping indicate Amazonian materials tens of meters thick overlying VBF (≥244 m) atop 1–2 km Hesperian lavas and Noachian basement at several km depth. The landing is within a double-ring ghost crater inferred to have penetrated Hesperian lavas.

• Zhurong’s ground survey measured exceptionally weak horizontal crustal fields at 16 sites along a 1,089 m traverse in Utopia Basin. Horizontal component amplitudes (north–south) typically range 3.4–24.0 nT after external and rover-field removal; total intensity excluding vertical is 5.2–39.8 nT. The average horizontal intensity is 11.2 nT with standard deviation 10.9 nT. Horizontal field directions are predominantly toward the southeast. • Spatial variability: Three traverse segments show: 0–200 m (sols 21–50) extremely weak fields (<10 nT, near zero); 200–600 m (sols 58–79) moderately weak (~30 nT); 600–1,000 m (sols 87–110) again extremely weak (<10 nT). Some or all variation could reflect residual ionospheric contributions; probabilities of no detectable crustal field within −5 to +5 nT are ~21% (N–S) and ~34% (E–W) based on uncertainties. • Comparison to orbital model: L19 downward-continued predictions at the site are much larger (horizontal ~55 nT; intensity ~81 nT) than Zhurong’s ~10 nT observations, while at InSight the surface field (~2,000 nT) is much stronger than orbital predictions (~300 nT), underscoring scale and model limitations. • Forward model and magnetization constraint: A weakly magnetized Hesperian lava layer (|M| ~0.14 A m−1) with an unmagnetized cylindrical “ghost crater” hole yields surface horizontal fields up to ~41 nT at the rim and near-zero (≈ −6 nT) at the centre, consistent with Zhurong’s weak measurements along the traverse. This suggests an upper limit of crustal magnetization beneath the landing region far below Southern Hemisphere values (~10 A m−1). • Geological context: The landing area lies within the Utopia Basin where orbital fields are weak and resurfacing by VBF and Amazonian materials modified older terrains; the ~5 km radius ghost crater may have been demagnetized by impact.

The exceptionally weak surface fields directly constrain local crustal magnetization in Utopia Basin. Two end-member explanations are proposed: (1) Basin-scale demagnetization: The entire Utopia Basin and underlying crust remained unmagnetized since the ~4 Ga basin-forming impact. Zhurong’s site lies ~1,200 km from the basin centre within the estimated ~1,600 km demagnetization radius, implying that post-impact Hesperian lavas were not magnetized (consistent with a dynamo that ceased by or before the Utopia impact). (2) Local impact demagnetization: The ~5 km radius ghost crater at the landing site could have demagnetized the local Hesperian lava units, producing a weak or near-zero field over the crater interior and stronger fields near the rim; forward modelling supports this scenario and matches observed magnitudes. Differences between orbital downward-continued predictions and ground observations highlight methodological limits. Orbital datasets at ~150 km altitude are dominated by wavelengths comparable to altitude; downward continuation amplifies high-degree noise and cannot capture shorter-wavelength anomalies that dominate at the surface. Thus, predicted surface fields of 50–100 nT at Utopia may represent noise amplification rather than true crustal signals, consistent with Zhurong’s weak values. Conversely, strong but small-scale magnetization (e.g., InSight) can produce large surface fields with weak orbital signatures. Implications for Mars’ dynamo and crustal history: Weak magnetization in Utopia suggests either dynamo cessation before emplacement of Hesperian lavas or demagnetization during/after emplacement in a reversing or absent field. Some deeper, weakly magnetized basement may persist, but would be masked by near-surface units. Zhurong’s results inaugurate kilometre-scale ground magnetic surveying on Mars, enabling geologically targeted interpretations that orbital data alone cannot resolve.

This study presents the first metre-to-kilometre scale ground magnetic survey on Mars. Zhurong’s RoMAG measurements along a 1.089 km traverse in southern Utopia Basin reveal exceptionally weak horizontal crustal fields (average ~11 nT), substantially below orbital downward-continued predictions and drastically weaker than InSight’s surface field. Forward modelling indicates an upper-limit crustal magnetization of ~0.14 A m−1 beneath the landing region, consistent with basin-scale or local impact demagnetization and suggesting that Hesperian lavas in Utopia were not magnetized, potentially due to dynamo cessation by Noachian–Hesperian times. The results underscore limitations of downward continuation from orbital altitudes and the necessity of in situ surveys to resolve small-scale magnetization. Future work should extend Zhurong’s traverse southward to cross the ghost crater rim and investigate nearby volcanic features, perform coordinated observations with InSight (or successor assets) to better quantify external fields, and conduct additional ground surveys at varied geologic settings to refine the timeline and processes governing Martian crustal magnetization and dynamo evolution.

• Vertical component could not be calibrated with the rotation method; analysis is limited to horizontal components. • External (ionospheric) fields were estimated indirectly using a model and InSight observations scaled by 83%; mismatches in local time, season, and ionospheric conditions introduce uncertainty and may leave residual external contributions. • Rover-generated magnetic interference required empirical corrections based on solar array currents and temperatures; residual systematics may persist despite high correlation fits. • Spatial coverage is limited to a single 1.089 km traverse with 16 sites; results may not fully represent regional variability. • Downward continuation comparisons suffer from amplified noise and lack of short-wavelength content in orbital models, complicating direct surface–orbit comparisons. • Geological interpretations (e.g., ghost crater size/depth, lava thickness) rely on indirect stratigraphic and crater-statistical inferences with associated uncertainties.