phanerozoic/jovian-lightning

Juno MWR Jovian Lightning Catalog

11,617 lightning discharges in Jupiter's atmosphere, detected across 69 Juno perijoves (2016–2025) by the Microwave Radiometer. The catalog carries three statistically striking observational findings that are not yet widely reported in the literature.

Findings

1. System-III longitude clustering

Lightning is not uniform in System-III longitude after accounting for Juno's sampling geometry. A chi-squared test on twelve thirty-degree bins, weighted by per-bin on-planet exposure, gives χ² = 254 on 11 d.o.f., p = 5 × 10⁻⁴⁸ against the uniform-rate null. Peak rates at 30°–60°E and 150°–180°E; trough at 210°–240°E. Raw (unweighted) test gives 215; the signal strengthens under exposure correction, confirming it is atmospheric rather than sampling-driven.

2. Concentration near the northern magnetic pole

In the Connerney tilted-dipole frame, the bin within 15° of Jupiter's northern magnetic pole holds 486 strikes against 131 expected on a uniform sphere (3.7×); the conjugate southern bin holds 120 against 1,129 expected (10× deficit). The full chi-squared against uniform magnetic colatitude is 5,561 on 11 d.o.f. Lightning is organized by the magnetic field, not simply by latitude.

3. Seventy-nine-fold N/S polar asymmetry

The northern polar region between 60° and 75° latitude produces 1,121 pulses; the conjugate southern region produces 9. Coverage ratio is only 1.57× (Juno samples the north more often), leaving a 79.5× coverage-corrected ratio — a physical hemispheric asymmetry, driven by Jupiter's non-dipolar magnetic field geometry.

Supporting observations

Depth varies with latitude

The ratio of R1 peak amplitude (0.6 GHz, probing ~1,000 bar) to R3 peak amplitude (2.6 GHz, ~10 bar) is a rough depth proxy. Deep lightning (ratio > 5) has mean |latitude| 36°; shallow lightning (ratio < 2) has mean |latitude| 70°. Polar lightning is systematically shallower than mid-latitude lightning — consistent with a colder polar atmosphere and a shallower water cloud base. The reproduction notebook plots this.

Great Red Spot R3 halo excess

Averaging the R3-channel residual brightness within 8° of the Great Red Spot (22°S, 317°E) across all 69 perijoves gives +1.88 K (1,533 samples). Known phenomenon, cleanly recovered here.

Quick start

from datasets import load_dataset

lightning = load_dataset("phanerozoic/jovian-lightning",
                         "lightning_catalog", split="train")

candidates = load_dataset("phanerozoic/jovian-lightning",
                          "transient_candidates", split="train")

stats = load_dataset("phanerozoic/jovian-lightning",
                     "per_perijove_stats", split="train")

Configs

lightning_catalog — 11,617 rows. One row per lightning discharge group (consecutive MWR samples exceeding 5σ at R1 or R2 high-gain). Carries UTC time, planetographic and planetocentric latitude, System-III longitude, emission angle, range to 1-bar surface, magnetic colatitude, zone/belt assignment, and peak residual brightness in each MWR channel.

transient_candidates — 1,409 rows. Output of a matched-filter search for multi-channel anomalies not fit by lightning, synchrotron, or thermal atmospheric structure. Each candidate has spectral classification, multi-channel SNR, template fit parameters, and multi-label taxonomy (e.g., CONVECTIVE_STORM, POLAR_AURORAL, LIMB_ARTIFACT, NONTHERMAL_CANDIDATE). Useful for anomaly-detection benchmarks.

per_perijove_stats — 69 rows. Summary statistics per Juno perijove (1 through 73, with PJ2, 47, 55, 58, 68, 72 missing or partial). Total and on-planet sample counts, latitude coverage, residual MAD per channel, and per-perijove lightning rate.

Full column schemas

lightning_catalog.parquet: perijove, t_utc_start, t_utc_end, duration_samples, duration_seconds, pg_latitude_deg, pc_latitude_deg, system_iii_longitude_deg, emission_angle_deg, range_to_surface_km, magnetic_colatitude_deg, zone_belt, peak_R{1..6}_K (with _low_gain / _high_gain variants for R1, R2), depth_ratio_R1R3.

transient_candidates.parquet: all of the above plus template_kind, template_param, t_utc_peak, max_emission_angle_in_window_deg, spectral_class, peak_channel, combined_snr, max_channel_snr, n_channels_coincident, per-channel peak_{R*}_K / sigma_{R*}_K / snr_{R*}, taxonomy_labels.

per_perijove_stats.parquet: perijove, perijove_closest_approach_utc, n_total_samples, n_on_planet_samples, n_good_geometry_samples, coverage_lat_min_deg, coverage_lat_max_deg, mad_R{1..6}_K, n_lightning_pulses, lightning_rate_per_on_planet_hour.

Methodology

Level-2 calibrated IRDR and GRDR files for all high-rate perijove windows were downloaded from the NASA PDS Atmospheres node (bundle jnomwr_1100, DOI 10.17189/1518928) and parsed into per-hour parquet tables. A three-dimensional baseline was constructed by pooling all 69 perijoves and computing the median antenna temperature in each (5° latitude × 20° longitude × 5° emission-angle) bin. Residuals are antenna temperature minus binned median.

Lightning detection flagged samples whose R1 or R2 high-gain residual exceeded 5σ of the local baseline; consecutive flagged samples were grouped into single events. The threshold was calibrated against an independent-permutation shuffle null that produced zero detections. Magnetic colatitude uses the Connerney tilted-dipole approximation (north pole at 80.4°N, 196.4°E). Zone/belt labels follow the standard Jovian banding. Full pipeline source is in the project repository linked in the citation block.

Suggested tasks

• Anomaly detection on the transient_candidates config: given multi-channel residual spectra and taxonomy labels, train a classifier or anomaly-ranker to identify the rare genuinely-unexplained events amid lightning extensions, synchrotron residuals, and limb artifacts.
• Spatial density regression. Predict lightning density as a function of (planetographic latitude, System III longitude, magnetic colatitude) — a challenging low-signal regression with strong non-uniformity.
• Time-series classification. With the three configs joined on perijove number, test whether per-perijove statistics predict lightning-rate variations over the 2016–2025 mission.
• Physics-informed forecasting. The depth–latitude trend and magnetic-colatitude concentration offer two independent physical priors; models that encode them correctly should outperform geometry-naive baselines.
• Cross-dataset correlation. Correlate lightning density maps with published Jupiter magnetic-field model (JRM33) tesseral coefficients, or with JunoCam/JIRAM storm catalogs.

Reproduction notebook

notebooks/reproduce_headline_results.ipynb loads the dataset and reproduces every statistic in this card — the three headline tests, the depth-versus-latitude trend, the PJ65 candidate lookup, and the per-perijove rate plot. Executed inline; all figures render in the HF preview without re-running.

Origin

The catalog was assembled in the course of a search for persistent microwave transients consistent with the resonant-cavity ball-lightning hypothesis (Wu 2016, Scientific Reports 6:28263). No high-significance candidate emerged; the 95% CL upper limit on such events is 0.148 per perijove after efficiency correction. One marginal candidate at 59.5°N, 8°E in PJ65 (2024-09-20T02:28 UTC) survived every confound filter and is preserved in transient_candidates under the peaked_interior spectral class. Cross-channel time-lag analysis mildly favors an atmospheric convective origin over a resonant-cavity one. See the transient_candidates config for the full taxonomy.

License

CC0 1.0. Underlying antenna temperatures are NASA public-domain data; this catalog is a derived product released under the same terms.

Citation

Juno MWR Jovian Lightning Catalog.
https://huggingface.co/datasets/phanerozoic/jovian-lightning

Janssen, M.A. et al. (2017). MWR: Microwave Radiometer for the Juno
  Mission to Jupiter. Space Science Reviews 213, 139.

Juno MWR Bundle, NASA PDS Atmospheres Node. DOI 10.17189/1518928.

Prior work on Jovian lightning against which these findings should be read:

Brown, S.T. et al. (2018). Prevalent lightning sferics at 600 megahertz
  near Jupiter's poles. Nature 558, 87.

Wong, M.H. et al. (2026). Radio Pulse Power Distribution of Lightning
  in Jupiter's 2021–2022 Stealth Superstorms. AGU Advances.
  DOI 10.1029/2025AV002083.