Spatial and temporal co-occurrence of dusty regions and ionospheric plasma depletions at Mars

​

Final Presentation�

Tsai Wei Jung / NTNU�

Prof. Wei-Ling Tseng

dust storm

• Dust is dominant radiative driver of the Martian atmosphere; column opacity varies ~10× within weeks

• Global dust storms are rare and irregular — 2001 (MY25) · 2007 (MY28) · 2018 (MY34);

activity peaks near perihelion, Ls ≈ 180–360°

• Dust itself reaches 85 km at GDS peak　[Luginin et al. 2020]

• Dust heating lifts the whole atmosphere up ~10 km; neutral densities up to +200% at 170–220 km　[Liu et al. 2018]

• Ion densities at 165–190 km differ during storms — regional and global give opposite signs for O₂⁺ (+50% / −11%)　[Farahat et al. 2022]

https://science.nasa.gov/asset/hubble/a-global-dust-storm-on-mars/

How dust storms (event) are defined?

• Manual standard:�Contours drawn by hand; sequences are "seemingly organized / identifiable / coherent"　

[Wang & Richardson 2015; Battalio & Wang 2021; Wang et al. 2023]

• automatic attempt:�“no such sharp limit” — they set an 80th-percentile CDOD threshold, storms are inside a manually chosen time window and must last ≥ 10 Sols, and three global dust events are discarded

[Ramette et al. 2026]

"There is no precise definition of the value of atmospheric dust, is sufficiently large to correspond to a dust storm, which also depends on the wavelength of the observations."

[Leseigneur et al. 2026]

Plasma depletion events

• All ion species drop >1 order of magnitude; electron temperature rises abruptly in 1,125 events (2014–2021) [Basuvaraj et al. 2022]

• 1,570 event (2015-2022) seen in LPW electron density, altitude peaks at ~265 km and ~410 km; inside most of them ion temperature rises and suprathermal electron flux is enhanced [Madanian et al. 2024]

• Preferentially nightside, local winter, over radial crustal magnetic fields [Basuvaraj et al. 2022; Park et al. 2024]

• Mechanism unresolved — proposed: ion frictional heating → recombination; Similar to Venus ionospheric holes and Earth equatorial plasma bubbles [Madanian et al. 2024; Brace et al. 1980; Kil 2015]

• Day-to-day variability shows no strong correlation with thermospheric CO₂ disturbances or solar wind [Park et al. 2024]

​

2015-11-28 (MY33, Ls 74.6°) from MAVEN (Multiple Instrument Ensemble)

Basuvaraj et al. 2022

All five species’ depletion peak within 2.6 km of the same altitude (385.3–387.9 km) — a plasma structure

σ_Z ≈ 28 km (vertical) vs

σ_G ≈ 122 km (horizonal)

Motivation

1. The altitudes overlap.�Dust storm reaches 85 km and its atmospheric response reaches 220 km; PDEs sit at 200–410 km, nightside median 204 km.�→ Could the two be related?
2. storm definitions are manual, qualitative, and exclude the largest events.�

• Define dusty regions objectively and continuously from CDOD　(MY24–36)
• Reconstruct and validate the PDE population　(MY32–36)
• Test their spatial and temporal co-occurrence against four independent null models models

Data1: CDOD (column dust optical depth) �Montabone et al. (2015,2020)

• Data selection:�Daily (MY24-36) global kriged maps of 9.3 μm absorption column dust optical depth (3°×3°, normalised to 610 Pa)

1. Continuous, long term (Same reason as Ramette et al. (2026))

2. Normalised to 610 Pa — removes topographical impact

• Caveats:�Three instruments combine: TES nadir(MY24–27), THEMIS nadir(MY26–33), MCS limb(MY28–35+)

limb dust profiles are often missing near the ground -> extrapolated by pressure� lack of observations in polar-winter -> hard to interpolate so set = 0.01

Data2: MAVEN PDEs�Basuvaraj et al. (2022)

• Data selection:

Mendeley data, 1125 PDEs from 8618 MAVEN orbits (MY32-36) -> Their fitting results

MVN_INSITU_KP-4SEC from NASA MAVEN -> Our fitting data

We use the Basuvaraj et al. (2022) values rather than our own fits (slide 6):�Sometimes time window have 2 peaks, it’s impossible to do the gaussian fitting

Basuvaraj et al. (2022) split each peak(in same time window) into individual PDE

Coverage:�

Building dust climatological map

Two regional dust events (MY34 Ls 324–344 / MY36 Ls 311–330):�From Kass et al. (2016), C-storm = start at Ls = 305-320 / end at Ls = 325-335

Both are similar to C-storm

Ramette et al. (2026)

Our method

detect dusty region

1. ρ > 1.5:

Our setting threshold

2. Τ > 0.20:�The clear-atmosphere ceiling τ(0.9 μm) < 0.5 (Leseigneur et al. 2026), and converted to 9.3 μm absorption by the factor 2.6 (Montabone et al. 2015)

0.5 / 2.6 = 0.192

3. area > 10⁵ km²:�MDAD minimum area

4. Classify local / regional dusty region (Cantor et al. (2001))�< 1.6×10⁶ km²: local

> 1.6×10⁶ km²: regional

-> 24,624 detections

From detections to events: the tracking rule

1. Pair and calculate:�detection overlap (sol t / sol t+1 one by one) as pairs�> calculate their ov/max | ov/min | min/max
2. Filter:�A pair is kept if�(i) ov/max > 0.30, and neither of both detections has more than one contained partner on that sol-pair, or�(ii) ov/min > 0.85 and min/max > 0.10 and it is the only contained pair on both sides.
3. Connect:�Using Hungarian algorithm(cost 1 − ov/max, over the kept pairs) to connect pairs(one to one) into track �→ 11,235 tracks
4. Link:�Any pair with ov/min > 0.85 that “connect” step did not use becomes a link → 2,285 links
5. catagorize:�min/max ≤ 0.10 → shed(recorded, never joined) �one-to-many split �many-to-one merge�one-to-one growth / decay → split 488 · merge 426 · growth 132 · decay 135 · shed 1,104 → 10,106 systems

Pairing PDEs with dusty regions

Pairing PDEs & dusty region with lags Δt = 0 … 7 sols

Conclusions

Future

• PDEs projected vertically onto the grid
• Pair PDEs with neutral / ion density, not only dust presence
• Column τ only; repeat with MCS vertical dust profiles
• Error vs a manual reference; object-by-object against MDAD
• Second PDE catalogue: Park et al. (2024), 1,570 LPW events
• OMEGA ODSC (~5 km) — what CDOD misses at local scale