🔭 astrophysics

Multi-Thermal CME Detection with ALMANAC

This paper presents a re-engineered, multi-thermal implementation of the ALMANAC algorithm that enhances low-coronal CME detection and early warning capabilities by integrating multi-wavelength EUV observations to improve event discrimination, onset localization, and the identification of pre-eruptive signatures.

Original authors: Thomas Williams, Christopher B. Prior, David MacTaggart, Huw Morgan

Published 2026-06-18✓ Author reviewed ⓘ

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Thomas Williams, Christopher B. Prior, David MacTaggart, Huw Morgan

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the Sun as a giant, restless kitchen. Sometimes, it lets out a massive burp of hot gas and magnetic energy called a Coronal Mass Ejection (CME). If this burp is aimed at Earth, it can mess up our satellites, GPS, and power grids. The big problem for weather forecasters is that these burps often start as tiny, invisible whispers in the Sun's lower atmosphere (the "low corona") before they grow into the huge clouds we can see with special telescopes. By the time we see the big cloud, it's often too late to give a good early warning.

This paper introduces a new, upgraded version of a digital "smoke detector" called ALMANAC. Here is how it works, using simple analogies:

1. The Old vs. The New Detector

The Old Way (Single-Channel): The original ALMANAC was like a security camera that only looked at the kitchen through one specific color filter (like only seeing in red light). It was good at spotting big changes, but it got confused easily. If the sun did something that looked like a CME in red light but wasn't, the detector would sound a false alarm. Also, if the camera got a little shaky or the light changed, it might split one big event into two confusing smaller ones (like seeing a single person as two ghosts).
The New Way (Multi-Thermal): The new ALMANAC is like upgrading that security system to a 7-lens camera that sees the kitchen in seven different colors (wavelengths) at the same time. Each color shows the Sun at a different temperature.
- The Analogy: Imagine trying to identify a person in a crowd. If you only see them in a red shirt, you might mistake a stranger for your friend. But if you see them in a red shirt, blue pants, and a green hat all at once, you know for sure it's your friend. By combining all seven "colors," the new system ignores the confusing noise and false alarms, making it much harder to trick.

2. How It Finds the "Burp"

The system doesn't just look for bright spots; it looks for change.

The "Running Mean" Trick: Imagine watching a video of a calm lake. If a rock is thrown in, the water ripples. The new ALMANAC calculates the "average" look of the Sun over the last hour. Then, it subtracts that average from the current image. This highlights only the new ripples (the eruptions) and ignores the calm water (the static parts of the Sun).
The "Group Hug" (Clustering): Sometimes, the camera might spot a tiny ripple in the red channel and a different tiny ripple in the blue channel. The old system might count these as two separate events. The new system acts like a bouncer at a club: it checks if these ripples are happening in the same place at the same time. If they are, it groups them together into one single event. If they are scattered or don't match up, it ignores them as noise.

3. What They Found

The authors tested this new system against 20 known "Sun burps" (CMEs) that were already recorded by other telescopes.

Better Timing: The new system found the start of these events about 40 minutes earlier on average than the old methods, though sometimes it was a bit off.
Better Location: It pinpointed where on the Sun the burp started with an accuracy of about 12 degrees (roughly the width of your fist held at arm's length).
Fewer Mistakes: It successfully stopped the "splitting" problem where one event was counted as two. It also caught some small eruptions that the old system missed.

4. The "Kurtosis" Crystal Ball

The paper also tested a new mathematical tool called kurtosis.

The Analogy: Think of the Sun's brightness as a crowd of people talking. Usually, everyone is chatting at a normal volume. Kurtosis measures how "spiky" the volume gets. Does the crowd stay calm, or do they suddenly scream in unison?
The Discovery: The researchers found that before a massive solar explosion (like an X-class flare), the "volume" of the Sun's brightness often spikes wildly in a very specific way.
- In active regions (kitchens about to explode), these "spikes" happened right before the big events.
- In quiet regions (calm kitchens), the spikes were tiny or non-existent.
- This suggests that watching these "spikes" could help forecasters predict how big an explosion might be before it even happens.

5. Why This Matters

The main goal of this paper isn't to say "we can now predict the weather perfectly." Instead, it says:

We can see the start of the storm earlier. By looking at the Sun in seven colors instead of one, we catch the "whispers" of an eruption before it becomes a "roar."
We make fewer mistakes. The new system is less likely to cry wolf because it cross-checks its findings across multiple "colors."
It helps humans. The system is fast enough to run automatically, but it produces clear movies and data that human forecasters can look at to make the final call.

In short, this paper presents a smarter, multi-colored, and faster way to watch the Sun, helping us get earlier and more reliable warnings about space weather that could affect life on Earth.

Technical Summary: Multi-Thermal CME Detection with ALMANAC

Problem Statement
Reliable identification of the low-coronal origins of Coronal Mass Ejections (CMEs) remains a critical limitation in space weather forecasting. Traditional coronagraph-based catalogs (e.g., CDAW, CACTus) typically detect CMEs only after they have propagated to 2–6 solar radii, missing low-coronal signatures and precursor activity. Furthermore, single-viewpoint observations suffer from projection effects, and automated detection methods often struggle with instrumental artifacts, projection ambiguities, and the inability to distinguish between eruptive events and non-eruptive coronal variability. The original Automated Detection of CoronaL MAss Ejecta origiNs for Space Weather AppliCations (ALMANAC) algorithm addressed low-coronal detection using a single EUV channel but was limited by projection effects, threshold sensitivity, and spurious detections caused by instrumental artifacts.

Methodology
The authors present a re-engineered, multi-thermal implementation of the ALMANAC algorithm, fully re-implemented in Python to leverage parallel computing and modern scientific libraries. The workflow operates on SDO/AIA EUV observations across seven wavelengths (94, 131, 171, 193, 211, 304, and 335 Å) and proceeds through five primary stages:

Data Acquisition and Preprocessing: Synoptic or near-real-time AIA data are retrieved, exposure-normalized, and rebinned. A running temporal mean is computed for each pixel using a reflective padding scheme. A normalized ratio cube ( $R$ ) is generated by dividing the absolute deviation from the running mean by the temporal median of that deviation, effectively suppressing static coronal structures while enhancing transient signatures.
Single-Channel Detection: Candidate eruptive regions are identified in each wavelength independently via spatio-temporal thresholding. Regions are filtered by size (minimum 1000 pixels exceeding $R=2$ ) and smoothed spatially and temporally to enforce continuity. Connected-component labeling identifies candidate regions, which are discarded if they fall below a minimum voxel volume (6000) or duration (18 minutes).
Spatio-Temporal Clustering: To associate detections across wavelengths, all single-channel centroids are aggregated into a 3D spatio-temporal data cube. Spatial and temporal binning groups detections that are closely spaced in time and space. Three-dimensional connected-component labeling (6-connected) identifies contiguous clusters of occupied voxels. Clusters must meet a minimum voxel count ( $N_{vox,min} = 4$ ) to be retained, filtering out isolated noise or channel-specific artifacts.
Origin Determination: For each validated cluster, heliographic source coordinates are derived by collecting all associated centroids. Robust statistical filtering (using the Interquartile Range) removes outliers, and the mean of the remaining coordinates defines the final source location. The onset time is defined as the first frame containing a valid voxel in the clustering cube.
Synergy with Topology (ARTop): The framework integrates with the Active Region Topology (ARTop) code to analyze magnetic winding and helicity flux densities derived from HMI vector magnetograms. The study specifically examines "current-carrying" components of these topological quantities to identify pre-eruptive signatures.

Key Contributions

Multi-Thermal Architecture: The transition from a single-channel to a multi-wavelength system improves robustness against projection effects and instrumental artifacts by requiring multi-channel coincidence for event validation.
Spatio-Temporal Clustering: A novel clustering scheme merges detections across channels, significantly reducing event bifurcation (where a single event is split into multiple detections) and improving the coherence of source region identification.
Kurtosis Time-Series Analysis: The authors introduce the calculation of kurtosis for intensity and ratio distributions across all AIA channels. They demonstrate that these statistical metrics exhibit recurrent pre-eruptive spikes that often align with enhancements in magnetic winding and helicity injection.
Operational Scalability: The Python re-implementation enables parallel processing, reducing the total processing time for an 8-hour window (including clustering and movie generation) to approximately 132 seconds on a standard desktop, facilitating near-real-time monitoring.

Results

Benchmarking: When tested against 20 halo CMEs from the CDAW catalogue, the multi-thermal ALMANAC achieved an effective event association accuracy of 65% (43 detections), a marginal improvement over the 63% accuracy of the single-channel version. The mean temporal discrepancy with CDAW was $40.4 \pm 30.2$ minutes, and the mean spatial discrepancy was $11.9 \pm 11.6^\circ$ . When excluding outliers (events with sympathetic eruptions or filament eruptions lacking X-ray signatures), these discrepancies reduced to $27.9 \pm 14.9$ minutes and $8.7 \pm 4.8^\circ$ .
Event Discrimination: The multi-thermal approach successfully distinguished between distinct eruptions that appeared as bifurcated detections in the single-channel analysis (e.g., separating two distinct CMEs associated with CDAW Index 5).
Pre-Eruptive Signatures: In active regions AR 11158 and AR 12673, kurtosis spikes frequently preceded major solar activity, including X-class flares and CMEs. These spikes often aligned with increases in magnetic winding and helicity injection. In contrast, the magnetically quiet region AR 12699 exhibited kurtosis values an order of magnitude lower, with no significant pre-eruptive spikes.
Detection of Missed Events: The updated algorithm successfully detected a CME associated with AR 11875 that the original single-channel ALMANAC failed to identify, detecting micro-flare precursors prior to the photospheric magnetic winding signatures reported in previous studies.

Significance and Claims
The paper claims that the multi-thermal ALMANAC framework significantly enhances the interpretability and operational applicability of automated CME detection. By integrating complementary temperature responses, the method provides clearer separation of eruptions from non-eruptive coronal variability and reduces fragmentation of events. The authors posit that the integration of coronal statistical diagnostics (kurtosis) with photospheric magnetic topology offers a pathway toward improved eruption forecasting. Specifically, the recurrent pre-eruptive kurtosis spikes suggest a potential for early warning capabilities that may precede events not yet captured in white-light coronagraph catalogs. The study concludes that while the method is not perfect and still suffers from some sensitivity to instrumental artifacts and data gaps, it provides a scalable, physically grounded tool for both retrospective analysis and near-real-time space weather monitoring, particularly when coupled with magnetic topology diagnostics.

1. The Old vs. The New Detector

2. How It Finds the "Burp"

3. What They Found

4. The "Kurtosis" Crystal Ball

5. Why This Matters

More like this