Improvements to the NSO Farside Mapping Pipeline: Noise Reduction Updates

TL;DR

This work tackles noise and reliability challenges in farside solar active-region mapping by NSO/GONG using helioseismic holography. It introduces a suite of updates, including a CNN-based quality filter (FQI), higher-cadence component maps, duty-cycle gating, variable per-pixel averaging, and expanded data products such as magnetic-strength maps and multiple projection schemes. Across 7,757 maps, the updates yield measurable gains in near-limb detection and image quality, with quantified improvements in variance, sharpness, and spatial-frequency metrics, enhancing both operational forecasting and scientific analysis. The authors also outline a roadmap for future ML-driven enhancements and Green's-function revisions to further improve farside mapping fidelity.

Abstract

The National Solar Observatory (NSO)'s Farside Pipeline is a critical tool of the space weather industry. It enables the detection and tracking of solar active regions that have rotated to the farside (invisible surface) of the Sun without relying on direct observational platforms such as satellites. By applying the technique of helioseismic holography to continuous Doppler images of the front side (visible surface), the pipeline infers the size and location of these regions through the acoustic signatures. These farside maps, produced using data from the NSO's GONG Network, allow scientists and solar observers to monitor the behavior of solar active regions. They support efforts to protect vital telecommunications and national interest infrastructure. While the data from this pipeline are widely used to many scientific, industrial, and national security applications, global helioseismic monitoring remains a developing field, with ongoing refinements in methodology and reliability. In this report, we will outline the updates made to the NSO's Farside Pipeline which have resulted in more accurate and consistent helioseismic maps, strengthening its value for both operational forecasting and scientific research.

Figures (15)

• Figure 1: Global distribution of six sites of the GONG network.
• Figure 2: A contaminated "dinner plate"(left) and an uncontaminated (right) fqi Dopplergrams.
• Figure 3: (Top row) Phase-shift maps of the farside for timestamps 20250104t1200 (left) and 20250105t1200 (right) generated by the legacy pipeline, classified as an fqm products. (Bottom row) Regions of higher negative phase shifts extracted from the fqm maps are shown in the top row, an fqj product. The images also show the scaled magnetogram on the Earth-side. The demarcation between the two sides is marked by nearly vertical lines. Since fqj product combines phase shifts with observed magnetic field, it is not entirely consistent over the solar sphere. As such, it is primarily intended for symbolic representation rather than quantitative analysis.
• Figure 4: A legacy pipeline product, fqj, that combines observed magnetic field on the frontside with farside pipeline estimated magnetic field strength. The farside map is created by combining two fqm maps. Since the mean duty cycle is 98%, the sequence of input Dopplergrams was largely uninterrupted and the map is not noisy. Three candidate active regions have been identified, marked with red circles and labeled with the probability for their appearance on the front side based on the estimated strength. These labels are only applied if the duty cycle is high enough for the output map to be considered trustworthy. The edges of the two fqm maps are marked by nearly vertical lines. As the map translates phase shifts into calibrated magnetic field strengths, the resulting product is commonly referred to as the farside calibrated map.
• Figure 5: Violin plot showing distribution of network identified good and anomalous images with their corresponding RMS values. Note the frequent occurrence of overlapping rms distributions below 40.