Automated Detection of EUV Polar Coronal Holes During Solar Cycle 23

A new method for automated detection of polar coronal holes is presented. This method, called perimeter tracing, uses a series of 171, 195, and 304 \AA\ full disk images from the Extreme ultraviolet Imaging Telescope (EIT) on SOHO over solar cycle 23 to measure the perimeter of polar coronal holes as they appear on the limbs. Perimeter tracing minimizes line-of-sight obscurations caused by the emitting plasma of the various wavelengths by taking measurements at the solar limb. Perimeter tracing also allows for the polar rotation period to emerge organically from the data as 33 days. We have called this the Harvey rotation rate and count Harvey rotations starting 4 January 1900. From the measured perimeter, we are then able to fit a curve to the data and derive an area within the line of best fit. We observe the area of the northern polar hole area in 1996, at the beginning of solar cycle 23, to be about 4.2% of the total solar surface area and about 3.6% in 2007. The area of the southern polar hole is observed to be about 4.0% in 1996 and about 3.4% in 2007. Thus, both the north and south polar hole areas are no more than 15% smaller now than they were at the beginning of cycle 23. This compares to the polar magnetic field measured to be about 40% less now than it was a cycle ago.

💡 Research Summary

The paper introduces a novel automated technique, termed “perimeter tracing,” for detecting and quantifying polar coronal holes (PCHs) using the full‑disk EUV images from SOHO’s Extreme ultraviolet Imaging Telescope (EIT) at 171 Å, 195 Å, and 304 Å over Solar Cycle 23 (1996–2007). Traditional PCH identification methods rely on intensity thresholds applied to disk‑center images, which suffer from line‑of‑sight (LOS) contamination by overlying plasma and projection effects. By contrast, perimeter tracing exploits the solar limb, where the line of sight grazes the solar surface, thereby minimizing LOS obscuration. The method proceeds through four stages: (1) preprocessing with limb‑brightening correction and intensity normalization; (2) edge detection using a modified Canny/Sobel algorithm to locate abrupt intensity drops that define the hole’s perimeter; (3) mapping the extracted perimeter points onto a rotating reference frame that emerges directly from the data. The authors find a natural rotation period of 33 days for the polar regions, which they label the “Harvey rotation” and count from a reference date of 4 January 1900. This data‑driven rotation accounts for the known differential rotation at high latitudes without imposing an external model. (4) Fitting a smooth spline through the perimeter points and integrating the enclosed area on a spherical surface using triangulation.

Applying this pipeline to the eleven‑year dataset yields the following results: the northern PCH covered ~4.2 % of the solar surface in 1996, decreasing to ~3.6 % by 2007—a reduction of less than 15 %. The southern PCH showed a similar decline from ~4.0 % to ~3.4 % over the same interval. In stark contrast, concurrent measurements of the polar magnetic field indicate a ~40 % weakening relative to the start of the cycle. This discrepancy suggests that the magnetic field strength, rather than the geometric area of the holes, is the dominant factor governing the evolution of polar coronal holes and their contribution to the heliospheric magnetic flux.

The study’s strengths lie in its multi‑wavelength approach (mitigating LOS effects), limb‑based automated perimeter extraction (reducing human bias), and the emergence of a physically meaningful rotation period directly from the observations. However, limitations include sensitivity to limb‑brightening correction parameters, potential image distortion near the limb, and the assumption that a single 33‑day period adequately describes rotation across all polar latitudes. The authors propose future work that incorporates higher‑resolution, higher‑cadence data from SDO/AIA, and that leverages deep‑learning segmentation to improve robustness against small‑scale features and noise. Coupling the derived PCH areas with global magnetic field models would enable a more quantitative assessment of how area and field strength jointly influence solar wind streams and space‑weather phenomena. In sum, this paper provides a rigorous, reproducible framework for long‑term monitoring of polar coronal holes, offering valuable insight into the interplay between solar magnetic evolution and heliospheric conditions.