Automated Coronal Hole Detection Using Local Intensity Thresholding Techniques

Abstract

We identify coronal holes using a histogram-based intensity thresholding technique and compare their properties to fast solar wind streams at three different points in the heliosphere. The thresholding technique was tested on EUV and X-ray images obtained using instruments onboard STEREO, SOHO and Hinode. The full-disk images were transformed into Lambert equal-area projection maps and partitioned into a series of overlapping sub-images from which local histograms were extracted. The histograms were used to determine the threshold for the low intensity regions, which were then classified as coronal holes or filaments using magnetograms from the SOHO/MDI. For all three instruments, the local thresholding algorithm was found to successfully determine coronal hole boundaries in a consistent manner. Coronal hole properties extracted using the segmentation algorithm were then compared with in situ measurements of the solar wind at ∼ 1 AU from ACE and STEREO. Our results indicate that flux tubes rooted in coronal holes expand super-radially within 1 AU and that larger (smaller) coronal holes result in longer (shorter) duration highspeed solar wind streams.

Evolution of Coronal Holes and Implications for High-Speed Solar Wind During the Minimum Between Cycles 23 and 24

“…Ideally, detection of blue-shifted regions would be a better means to identify coronal holes, but velocity estimates in the corona and chromosphere are difficult and not routinely available, so it is customary to use intensity images to detect coronal holes. Historically, the best determinations of coronal holes were drawings made by careful and experienced observers (see, e.g., Harvey and Recely, 2002;McIntosh, 2003) but recently several methods have been proposed to detect coronal holes automatically, which vary from simple brightness thresholding at one single wavelength (see, e.g., Abramenko, Yurchyshyn, and Watanabe, 2009;Obridko et al, 2009) to more sophisticated approaches, which attempt to identify coronal holes objectively and to separate them from other dark regions on the Sun (see, e.g., de Toma and Arge, 2005;Henney and Harvey, 2005;Scholl and Habbal, 2008;Kirk et al, 2009;Krista and Gallagher, 2009).…”

Real-Time Solar Wind Prediction Based on SDO/AIA Coronal Hole Data

“…In the early phases of coronal holes identification, CHs were visually tracked by experienced observers (Harvey and Recely, 2002;McIntosh, 2003). Recently, several groups tried to automate the process for the identification and detection of coronal holes using different approaches such as perimeter tracing (Kirk et al, 2009), fuzzy clustering (Barra et al, 2009), multichannel segmentation (Delouille, Barra, and Hochedez, 2007), edge-based segmentation (Scholl and Habbal, 2008), intensity thresholding (Krista and Gallagher, 2009;de Toma, 2011;Rotter et al, 2012) and magnetic track-boundaries (Lowder et al, 2014). One way to model the physical parameters of the solar wind are MHD models of the corona and heliosphere, using synoptic solar magnetic field maps as input, such as ENLIL (Odstrcil and Pizzo, 2009).…”

Decay of Activity Complexes, Formation of Unipolar Magnetic Regions, and Coronal Holes in Their Causal Relation

“…After an appropriate smoothing and refinement, the resultant image detects CHs and low-intensity features in solar corona. Similar approaches are also used for automatic tracking of CHs (e.g., Barra et al, 2009;Krista and Gallagher, 2009).…”