Backtracking Bipolar Magnetic Regions to their emergence: Two groups and their implication in the tilt measurements

Bipolar Magnetic Regions (BMRs) that appear on the solar photosphere are surface manifestations of the Suns internal magnetic field. With modern observations and continuous data streams, the study of BMRs has moved from manual sunspot catalogs to automated detection and tracking methods. In this work, we present an additional module to the existing BMR tracking algorithm, AutoTAB, that focuses on identifying emerging signatures of BMRs. Specifically, for regions newly detected on the solar disk, this module backtracks the BMRs to their point of emergence. From a total of about 12,000 BMRs identified by AutoTAB, we successfully backtracked 3,080 cases. Within this backtracked sample, we find two distinct populations. One group shows the expected behavior of emerging regions, in which the magnetic flux increases significantly during the emerging phase. The other group consists of BMRs whose flux, however, does not exhibit substantial growth during their evolution, the instances where our algorithm fails to capture the initial emergence of the BMRs. We classify these as discarded BMRs and examine their statistical properties separately. Our analysis shows that these discarded BMRs do not display any preferred tilt angle distribution and do not show systematic latitudinal tilt dependence, in contrast to the trends typically associated with emerging BMRs. This indicates that including such regions in statistical studies of BMR properties can distort or mask the underlying physical characteristics. We therefore emphasise the importance of excluding the discarded population from the whole dataset when analysing the statistical behavior of BMRs.

💡 Research Summary

The paper presents a new “back‑tracking” module for the existing automated Bipolar Magnetic Region (BMR) detection and tracking system AutoTAB, with the goal of identifying the true emergence moment of each BMR on the solar photosphere. Using line‑of‑sight magnetograms from SOHO/MDI (1996‑2011) and SDO/HMI (2010‑2024), the authors first homogenize the data by rebining HMI to MDI’s 4‑arcsec resolution and applying a 1.4‑factor flux scaling. After correcting for projection (division by the cosine of the heliocentric angle), they smooth each magnetogram with an 11‑pixel boxcar and apply an adaptive threshold based on the image’s mean field strength. A flux‑balance criterion (|Φ⁺‑Φ⁻|/(Φ⁺+Φ⁻) ≤ 0) ensures that only regions with reasonably balanced opposite polarities are accepted as BMRs.

AutoTAB then tracks each detected BMR forward in time using a differential rotation profile appropriate for the region’s latitude. The tracking algorithm predicts the future longitude of the region, matches it to a candidate in the next frame, and requires the area change to stay within 80‑300 % (tighter when data gaps occur). This forward tracking yields a catalog of roughly 12 000 unique BMRs.

The new back‑tracking procedure starts from the first appearance time (T₀) recorded by AutoTAB and works backward, frame by frame, to locate the earliest detectable signature (the emergence time, Tₑ). For each backward step the algorithm extracts a rectangular box extending 9 pixels beyond the original mask, computes (a) the ratio R(t) of pixels above 100 G to the total box pixels, (b) the unsigned magnetic flux Φ(t), and forms two diagnostic ratios: rₚ = R(t)/R(T₀) and r_f = Φ(t)/Φ(T₀). If both r_f > 1 and rₚ > 1 the step is deemed unrealistic and skipped; if r_f ≤ 0.4 and rₚ ≤ 0.5 the region is considered to be decaying or lost in the noise, and after five consecutive skips the back‑tracking terminates. These thresholds were tuned empirically and proved robust against modest variations. The algorithm also discards regions whose bounding boxes exceed 100 pixels in either dimension or whose flux density B(T₀) is below 15 Mx cm⁻², as these are likely weak, incoherent fields or remnants of decayed activity.

Out of the 11 987 BMRs in the AutoTAB catalog, 3 080 (≈25 %) could be successfully back‑tracked to an emergence time. Examination of these back‑tracked cases reveals two distinct populations:

1. Emerging BMRs – show a clear increase in both unsigned flux Φ and strong‑field pixel ratio R during the backward evolution, indicating that the algorithm captured the genuine emergence phase. Their tilt angles follow Joy’s law, i.e., a systematic increase of tilt with latitude.

2. Discarded BMRs – exhibit little or no flux growth between T₀ and Tₑ; many have essentially the same flux at both times, suggesting that AutoTAB originally detected them after they had already reached a mature state or that the back‑tracking failed to locate the true emergence. Statistically, these regions display a uniform tilt‑angle distribution with no latitudinal dependence, effectively washing out Joy’s law when mixed with the emerging population.

The authors argue that including the discarded population in statistical studies of BMR properties (tilt angle, flux‑latitude relations, emergence rates) can significantly bias results, masking the physical trends that are only present in truly emerging regions. Consequently, they recommend an explicit pre‑selection step—using back‑tracking or similar emergence verification—to filter out non‑emergent BMRs before any scientific analysis.

The paper also discusses limitations: the 96‑minute cadence of MDI (and the rebinned HMI) restricts the temporal resolution of emergence detection; projection effects become severe near the east limb, preventing reliable back‑tracking for regions first seen at longitudes < –35°. Moreover, BMRs that emerge on the far side of the Sun or very close to the limb are inevitably missed. The authors suggest that higher‑cadence data (e.g., native 45‑second HMI), far‑side helioseismic imaging, and machine‑learning classifiers could improve detection of the earliest emergence signatures and reduce the fraction of discarded BMRs.

In summary, this work adds a practical, automated method for pinpointing the emergence time of BMRs, demonstrates that a substantial fraction of cataloged BMRs are not true emergences, and shows that tilt‑angle statistics are highly sensitive to this contamination. By providing a clear protocol for separating emerging from discarded BMRs, the study offers a pathway toward more reliable measurements of Joy’s law, flux emergence rates, and ultimately better constraints on solar dynamo models.