29. The STARA Sunspot Catalog

Contributed by Fraser Watson. Posted on October 2, 2014

Fraser Watson
National Solar Observatory, Tucson, AZ 85719, USA

With technological advancements, it has become more difficult to manually analyze solar data collected from all of the world’s telescopes. As an example, the Solar Dynamics Observatory (SDO) now returns high resolution data for a variety of wavelength channels, some with cadences as short as 45 seconds. As such, automated ways of processing and analyzing the data have become necessary and this was the reason behind the creation of the SDO Feature Finding team.

To ease the burden of analyzing white light sunspot observations, STARA (the Sunspot Tracking And Recognition Algorithm) was developed. The algorithm allows fast, robust detections of sunspots from full-disk, white light data1. It is regularly used with SOHO/MDI and SDO/HMI data, and ground based observatories such as Kodaikanal2 and Kanzelhöhe have used a modified version for their own white light observations.

figure1Figure 1: An example set of sunspot penumbral and umbral detections from HMI data on February 16th, 2011.

The methodologies of detection are relatively straightforward using techniques from the field of morphological image processing, a nonlinear image processing technique that uses shape and structure in digital images to aid in feature detection. All operations in morphological image processing can be broken down into a combination of two operators, erosion and dilation. These operators work as would be expected based on their names. Erosion reduces the area of a shape by removing pixels from the edges and dilation increases the area by adding pixels along the edges. For STARA, a combination of erosions and dilations known as the top-hat transform is used.

The top-hat transform is defined as T(f) = f – ( (f ⊖ s) ⊕ s) where f is the original data, ⊖ is the erosion operation, ⊕ is the dilation operation, and s is a probe known as the structuring element chosen specifically for the required problem. By applying this to the white light solar data, we find morphological elements that are smaller than the structuring element while still maintaining some similarity to its shape, and are darker than their local surroundings. The local surroundings constraint is important and useful in this context as limb darkening affects the level of solar intensity as a function of location on the disk. Therefore, this technique removes the need for limb darkening to be calculated or modeled for each image due to the local nature of the detection. This increases the speed of the algorithm over some other methods as the limb darkening correction is folded into the feature detection step, which is crucial if used as part of a feature detection pipeline. Additional advantages of this method involve the algorithms dependence on relative intensities as opposed to absolute intensities, making it easily adaptable to different data sets. However, the entire procedure is heavily reliant on the structuring element, or probe shape, used. To ensure the detections were as accurate as possible, a test data set was used containing human sunspot detections and was treated as a ‘ground truth’ for the algorithm allowing the optimal structuring element to be chosen.

Once the parameters of the algorithm were established, the SOHO/MDI and SDO/HMI data was processed and a catalog of sunspot properties from these datasets were recorded. These catalogs are publicly available at http://www.nso.edu/staff/fwatson/STARA.

figure2Figure 2 : The sunspot number determined by STARA for SOHO/MDI and SDO/HMI compared to the International Sunspot Number reported by the SIDC. Due to the International Sunspot Number being a group-weighted count and the STARA number being a raw count, the STARA data has been scaled up to the same magnitude as the International Sunspot Number.

The catalogs have been used for a variety of sunspot studies to date including determining the depth of sunspots (known as the Wilson depression), obtaining distributions of sunspot numbers, areas and magnetic field strengths, as well as the time evolution of these parameters3. In addition to this, the sunspot properties listed in the catalog have been used in comparisons with umbral magnetic field strengths from the McMath-Pierce Solar Telescope4, and sunspot areas measurements made at Debrecen Observatory.

It is hoped that these sunspot catalogs will be of use to the solar community. We welcome people to begin analyzing the data in conjunction with their own datasets. The data currently available in the STARA catalog includes the following.

  • Observation date and time
  • Area
  • Location in solar coordinates
  • Maximum magnetic field strength in umbra (from ME data for HMI and LOS data for MDI)
  • Number of spots


[1] Watson, F.T., L. Fletcher, L., Dalla, S., Marshall, S., 2009, Sol Phys., 260, 5
[2] Ravindra, B. et al., 2013, A&A, 550, A19
[3] Watson, F.T., Fletcher, L, Marshall, S., 2011, A&A, 533, A14
[4] Watson, F.T., Penn, M.J., Livingston, W.C., 2014, ApJ, 787, 22

One comment on “The STARA Sunspot Catalog

Leave a comment

Your email address will not be published. Required fields are marked *