STORMY : A Real-time Triggering Framework using Yamagawa Solar Spectrograph for Active Solar Emission Observations with the MWA

Technical Summary: STORMY – A Real-time Triggering Framework for MWA Solar Observations

Problem Statement
Solar radio bursts, which provide critical insights into solar physics and space weather, are inherently unpredictable, transient, and infrequent. Modern radio interferometers like the Murchison Widefield Array (MWA), precursors to the Square Kilometre Array (SKAO), offer superior imaging capabilities but are highly oversubscribed. Relying on blind, scheduled solar observations is inefficient, leading to a high probability of missing significant solar events. While automated detection algorithms exist for archival data, they are often unsuitable for real-time triggering because they require the full event morphology to be present before generating a trigger. A robust, low-latency framework is needed to utilize limited observing time effectively by capturing solar activity only when it is detected in near-real-time.

Methodology
The authors developed STORMY (Solar Triggered Observations of Radio bursts using MWA and Yamagawa), a real-time triggering framework designed for the MWA. The system operates on the following workflow:

1. Data Source and Latency Management:

   • The trigger is derived from near-real-time data from the Yamagawa Solar Spectrograph (Japan), which operates in the 70 MHz–9 GHz band. Due to similar longitudes, Yamagawa's observing window overlaps completely with the MWA (80–300 MHz).
   • The system utilizes a custom datastream from Yamagawa with a 2-minute latency.
   • To mitigate data loss caused by this latency, the MWA maintains a voltage ring buffer (storing up to 160 seconds of raw voltage data). When a trigger is received, the buffer is frozen and dumped to disk, allowing the recovery of the initial phase of the solar event that occurred prior to the trigger arrival.

2. Pre-processing Pipeline:

   • Bandpass Correction: The raw dynamic spectrum is normalized using a median bandpass calculated from previous full-day data (split into 180-second chunks) to remove instrumental spectral responses.
   • RFI Mitigation: The system employs specific algorithms to remove Radio Frequency Interference (RFI):
     • Periodic RFI: Identified via peak finding with a 300-second periodicity and removed.
     • Narrow-band Persistent RFI: Filtered using a modified z-score based on the Median Absolute Deviation (MAD) to avoid flagging genuine solar signals.
     • Transient RFI: Addressed through morphological operations.
   • Background Subtraction: A moving median filter (window width of 1800 seconds) is applied to subtract slowly varying background noise (instrumental gain, system temperature, quiet Sun).

3. Event Detection:

   • A Signal-to-Noise Ratio (SNR) dynamic spectrum is generated.
   • A binary dynamic spectrum is created using a threshold of 13σ (determined to optimize the balance between precision and recall).
   • Morphological Closing: A 2-step morphological closing operation (dilation followed by erosion using a 2×2 kernel) fills missing pixels caused by RFI flagging.
   • Contour Detection: Closed contours are identified using OpenCV. Events are validated if they span at least 15 MHz in frequency and 3 seconds in time.

4. Triggering and Observation:

   • Upon detection, a web-based mechanism triggers the MWA.
   • The system estimates the spectral span of the burst and configures the MWA's 30.72 MHz bandwidth to cover the detected frequency range, adding guard bands (30 MHz below, 10 MHz above) to account for spectral drift and weaker emissions.
   • The MWA switches to a "picket fence" mode (log-spaced frequency chunks) for the triggered observation.

Key Contributions

• Development of STORMY: The first implementation of a near-real-time triggering framework specifically for solar radio bursts using the MWA and Yamagawa data.
• Algorithmic Innovation: A robust pipeline combining median-based denoising, modified z-score RFI removal, and morphological closing to detect bursts in noisy, real-time data streams.
• Latency Compensation: A practical demonstration of using the MWA's voltage ring buffer to recover event data lost during the trigger generation latency (approx. 70–80 seconds).
• Operational Deployment: The system has been made operational at the MWA for routine solar observations.

Results

• Performance Metrics: Testing on 8,400 two-minute data chunks (containing 226 true events) showed that a threshold of 13σ yields the optimal F1-score, balancing precision and recall.
• Operational Success: Between July 31, 2024, and January 31, 2025, STORMY successfully triggered and observed approximately 110 solar bursts.
• Case Studies: The paper presents two successful captures:
  1. An M-class X-ray flare on August 1, 2024, where the system triggered an observation at 126 MHz, capturing radio images consistent with SDO/AIA data.
  2. A Group Type-III event on November 4, 2024, where buffer data was successfully correlated to produce a radio image at 103 MHz overlaid on LASCO C2 data.

Significance and Claims
The paper claims that STORMY provides a robust and efficient means to capture interferometric data of transient solar activity, addressing the inefficiency of blind scheduling. The authors emphasize that this framework:

• Enables the efficient capturing of event-rich solar data with existing instruments like the MWA, LOFAR, and OVRO-LWA.
• Paves the way for triggered observing with the SKAO (specifically SKA-Low). The authors propose that the "solar subarray" concept for SKA-Low could utilize similar voltage buffering and real-time detection to trigger high-bandwidth observations, minimizing the impact on other science observations while maximizing the capture of unpredictable solar events.
• Demonstrates that solar triggering is feasible with current technology, offering a pathway to optimize future solar radio astronomy during the ongoing solar maximum.