Quantifying sunspot group nesting with density-based unsupervised clustering

This paper introduces an automated density-based clustering method to quantify sunspot group nesting across 151 years of observations, revealing that approximately 60% of groups emerge in spatial-temporal nests, with nesting frequency and inter-nest spacing strongly correlated to solar activity levels.

The Gist

Here is an explanation of the paper, translated into everyday language with some creative analogies.

The Big Picture: The Sun's "Neighborhoods"

Imagine the Sun's surface as a giant, bustling city. On this city, sunspots are like temporary construction sites or pop-up markets. They appear, do some business (releasing magnetic energy), and then disappear.

For a long time, scientists knew these construction sites didn't just pop up randomly all over the city. Instead, they tended to cluster together in specific neighborhoods. If a construction site appeared in a certain spot, another one was likely to show up nearby soon after. Scientists call these clusters "nests" or "complexes of activity."

This paper is about finally putting a ruler to this phenomenon. Instead of just saying, "Hey, they seem to hang out together," the authors built a smart computer program to count exactly how often this happens and where it happens.

The Detective Work: How They Did It

The authors used two massive historical archives of sunspot data:

1. The Royal Greenwich Observatory (RGO): Like a meticulous diary kept from 1874 to 1976.
2. The Kislovodsk Mountain Astronomical Station (KMAS): A modern diary from 1955 to 2025.

Together, these cover 151 years of solar history. That's a lot of data!

To find the "nests," they didn't just look at the data with their eyes. They used a technique called DBSCAN (Density-Based Spatial Clustering).

• The Analogy: Imagine you are looking at a map of a city at night. Some areas are packed with bright lights (people hanging out), while others are dark and empty.
• The Method: The computer algorithm acts like a smart flashlight. It shines on the map and says, "Okay, if there are at least 3 sunspots close enough to each other in time and space, that's a 'Nest.' If a sunspot is all alone in the dark, it's just a 'loner.'"

They also used a "sparsity correction." Think of this like adjusting the focus of a camera. If you are looking at a very quiet time of day (low solar activity), the camera zooms out a bit so it doesn't accidentally split a real group of people into two separate groups just because they are a little far apart.

What They Found: The "Sun's Rules"

After crunching the numbers, here are the main discoveries:

1. The Sun is a Social Butterfly (60% Rule)
The most important number in the paper is 0.61. This means that about 61% of all sunspots are born inside these "nests." They don't like to be alone; they prefer to show up with friends. Only about 40% of sunspots are true loners.

2. The "Goldilocks Zone" (Latitude)
Sunspots have a favorite place to hang out.

• Too close to the equator? Not many nests.
• Too close to the poles? Very few nests.
• The Sweet Spot: Between 10° and 20° latitude (roughly the middle of the Sun's northern and southern hemispheres).
• The Analogy: Think of the Sun like a beach. The equator is the hot, crowded water (too chaotic), and the poles are the cold, empty dunes. The "nests" are the perfect spot on the sand where the waves are just right and people gather in groups. This is where the Sun's magnetic "toroidal" (doughnut-shaped) fields are strongest.

3. Stronger Cycles = Tighter Crowds
When the Sun is having a "busy" year (Solar Maximum), the nests get smaller and tighter.

• Low Activity: The nests are spread out, like people standing 200 meters apart in a field.
• High Activity: The nests pack together, like people at a concert standing only 60 meters apart.
• The Twist: The more active the Sun is, the more organized the clustering becomes. It's not just that there are more sunspots; it's that they are better organized into groups.

4. The "Drifting" Neighborhoods
You might think these nests stay in one fixed spot on the Sun. They don't.

• The Analogy: Imagine a group of friends walking down a moving walkway at an airport. They stay together as a group (the nest), but the walkway (the Sun's rotation) moves them along. Sometimes, the group even walks slightly faster or slower than the walkway itself.
• Because these groups drift and move around over the course of a year, if you look at the Sun over a whole 11-year cycle, the "neighborhoods" seem to disappear. They smear out across the entire surface. This explains why we don't see permanent "Active Longitudes" (fixed spots where activity always happens) when looking at long-term averages.

5. Big vs. Small Sunspots
The study found a hierarchy:

• Small sunspots are like the "parasites" of the solar world. They love to cluster right next to the big, famous sunspots. If you remove the small ones from the data, the clustering actually looks stronger for the big ones. This proves that the nesting isn't just a statistical trick caused by having too many tiny dots; it's a real physical phenomenon where big magnetic structures spawn smaller ones nearby.

Why Does This Matter?

Understanding how sunspots cluster helps us understand the Sun's internal engine (the Solar Dynamo).

• If 60% of sunspots are in nests, it means the Sun's magnetic field is organizing itself into large, coherent bundles deep underground before it even breaks the surface.
• This helps scientists predict how the Sun will behave. If the Sun is in a "nesting" mode, it might produce more intense solar storms because the magnetic fields are packed tightly together.
• It also helps us understand other stars. If our Sun does this, maybe other stars do too, which helps us understand why some stars seem to have wilder weather patterns than others.

The Bottom Line

The Sun isn't a chaotic mess of random spots. It's a highly organized system where magnetic fields like to gather in "neighborhoods." About 6 out of every 10 sunspots are part of a group, and these groups are most active in the middle latitudes. The Sun's magnetic field is constantly rearranging itself, creating these temporary, drifting complexes that shape our space weather.

Technical Summary

Here is a detailed technical summary of the paper "Quantifying sunspot group nesting with density-based unsupervised clustering" by Karapınar et al.

1. Problem Statement

Sunspot groups (SGs) do not emerge randomly across the solar surface; they often form spatial-temporal clusters known as nests or complexes of activity. While the existence of these nests has been recognized for decades, their quantitative characterization remains challenging due to:

• Subjectivity: Previous studies often relied on subjective criteria or assumed specific cluster shapes.
• Data Limitations: Long-term, homogeneous datasets were previously insufficient for automated, reproducible analysis.
• Complexity: Nesting patterns exhibit arbitrary shapes, varying persistence, and drift, making traditional clustering methods (like $k$-means) or sinusoidal fits inadequate.

The authors aim to develop an automated, data-driven approach to quantify the degree of nesting across solar cycles, analyze its dependence on latitude and activity levels, and validate these findings using independent observational datasets.

2. Methodology

The study employs a hybrid approach combining Kernel Density Estimation (KDE) for visualization and DBSCAN (Density-Based Spatial Clustering of Applications with Noise) for discrete cluster identification.

Data Sources

• RGO (Royal Greenwich Observatory): 1874–1976 (Solar Cycles 12–19).
• KMAS (Kislovodsk Mountain Astronomical Station): 1955–2025 (Solar Cycles 19–25).
• Calibration: KMAS area measurements were scaled to the RGO standard ($b_{RGO-KMAS} = 1.031$) to ensure homogeneity. The datasets overlap during Cycles 19–20, allowing for cross-validation.
• Total Scope: 151 years of data covering 13 complete solar cycles plus Cycle 25.

Algorithmic Framework

1. Preprocessing: Data is analyzed in the Carrington reference frame (co-rotating with the Sun at $\pm 16^\circ$ latitude).
2. Kernel Density Estimation (KDE): Used to visualize emergence density. SGs are weighted by the logarithm of their maximum area ($w_{A,i} = \ln(A_i)/\ln(A_{max})$) to reflect magnetic flux content.
3. DBSCAN Clustering:
   • Parameters: Fixed neighborhood radius $\epsilon = 0.11$ and minimum points $m_p = 3$.
   • Sparsity Correction: To handle low-activity periods where data is sparse, an adaptive effective radius $\epsilon_{eff}$ is applied:
     $$\epsilon_{eff} = \epsilon_0 \times \min\left(1.5, \sqrt{N_{ref}/N}\right)$$
     where $N$ is the number of SGs in the window and $N_{ref}=350$. This prevents artificial fragmentation of nests in sparse data.
   • Boundary Handling: A wraparound correction is applied to handle the $0^\circ/360^\circ$ longitude discontinuity.
4. Nesting Degree ($D$): Defined as the fraction of sunspot groups belonging to statistically significant nests:
   $$D_{ij} = \frac{N_{nest}}{N}$$
   where $N_{nest}$ is the number of SGs in nests and $N$ is the total number of SGs in a specific time-latitude window.
5. Validation:
   • Synthetic Data: Injection-recovery tests using synthetic solar-like cycles confirmed the method recovers true nesting degrees with a mean bias of $\sim 0.06$ and RMS scatter of $\sim 0.10$.
   • Statistical Significance: Identified clusters are tested against a null hypothesis of uniform random emergence using Poisson statistics ($p < 0.05$).

3. Key Results

A. Global Nesting Statistics

• Mean Nesting Degree: Across all cycles and latitude bands, the mean nesting degree is $\langle D \rangle = 0.61 \pm 0.12$. This implies that approximately 60% of all sunspot groups emerge within nests.
• Dataset Consistency: RGO and KMAS data show strong correlation ($r=0.777$) during the overlap period, though RGO yields slightly higher $D$ values due to better detection of small groups.

B. Latitude Dependence

• Mid-Latitude Peak: Nesting is strongest in the **$10^\circ$–$20^\circ$** latitude bands, where$\langle D \rangle \approx 0.69$. This aligns with the region of maximum toroidal flux emergence (the "butterfly diagram" activity belts).
• Decline at Extremes: Nesting decreases toward the equator ($0^\circ$–$5^\circ$,$\langle D \rangle \approx 0.58$) and high latitudes ($30^\circ$–$35^\circ$,$\langle D \rangle \approx 0.42$).
• Statistical Significance: A Kruskal-Wallis test confirms highly significant latitude dependence ($p < 10^{-8}$).

C. Correlation with Solar Activity

• Positive Correlation: There is a clear positive correlation between the nesting degree and the total sunspot area (solar activity level). Stronger cycles exhibit higher nesting degrees ($r \approx 0.60$ for RGO, $0.73$ for KMAS).
• Robustness: The correlation strengthens when small sunspot groups are excluded (thresholds up to 200–300 MSH). This indicates that the correlation is not a statistical artifact of abundant small groups but reflects genuine organization of medium-to-large active regions.

D. Hierarchical Structure

• Small Groups: Small SGs ($<200$ MSH) show high nesting degrees ($\sim 0.5$–$0.6$), suggesting they are "parasitic" flux emerging near large active regions.
• Large Groups: Large SGs ($>400$ MSH) show lower but non-zero nesting ($\sim 0.25$), indicating they also possess a tendency to cluster, albeit with greater spatial independence.

E. Inter-Nest Spacing

• The characteristic inter-nest length scale ($L_{typ}$) contracts as activity increases:
  • Low Activity: $\sim 200$–$500$ Mm.
  • Solar Maximum: $\sim 60$–$100$ Mm (approaching the typical longitudinal size of a single sunspot group).
• This suggests a transition from sparse emergence to dense packing of magnetic structures during solar maximum.

F. Longitude Distribution

• While individual nests can persist for multiple rotations, no persistent "active longitudes" are found when averaged over full solar cycles.
• Differential rotation and intrinsic longitudinal drift disperse nests over decadal timescales, resulting in a homogeneous longitude distribution in cycle-averaged data.

4. Significance and Contributions

• Automated Quantification: This is the first fully automated, cycle-by-cycle quantification of sunspot nesting using unsupervised machine learning on a century-scale dataset.
• Methodological Advance: The use of DBSCAN with sparsity correction allows for the detection of arbitrary-shaped clusters without pre-defining cluster numbers, overcoming limitations of previous statistical methods.
• Physical Insight:
  • Confirms that solar magnetic activity is highly organized, with ~60% of emergence occurring in clusters.
  • Links high nesting degrees to coherent toroidal flux bundles, suggesting that stronger cycles produce more organized magnetic structures.
  • Provides a benchmark ($D \approx 0.6$) for comparing solar activity with stellar activity cycles, aiding in the interpretation of photometric variability in Sun-like stars.
• Implications for Variability: The persistence and clustering of nests significantly influence solar irradiance modulation and rotational variability, offering constraints for "Sun-as-a-star" models.

5. Limitations and Future Work

• The current method uses a single spatial-temporal scale parameter, which may not capture variations in nest size within a window.
• The analysis treats longitude and time symmetrically after standardization, potentially ignoring physical anisotropies.
• Future work should incorporate full 3D clustering (including latitude) and probabilistic models to track the continuous evolution of nests.