[Previous] | [Session 1] | [Next]
S.R. Walton, D.G. Preminger (San Fernando Observatory, Cal State Northridge)
The automatic identification of solar features, such as faculae and plages, is becoming increasingly important as the size of solar data sets increases. These identifications are required for quantitative study of the solar cycle, including but not limited to: locations, lifetimes, contrasts, and other characteristics of sunspots and faculae; modeling of the total solar irradiance (TSI); and variations of sunspot and facular properties with latitude and/or solar cycle phase. We have been experimenting with an approach which, while simple, is more sophisticated than a simple threshold technique. We identify an isolated feature by scanning the disk and finding a pixel whose contrast exceeds a given criterion called the trigger. We then `walk' around the trigger pixel, looking for adjacent pixels which are part of the same spot. If three contiguous pixels meet the trigger criterion, then a feature has been found. We continue to add adjacent pixels to the feature; these must meet a less stringent contrast criterion, the threshold, in order to be considered part of the feature. We find that this technique allows the trigger to be chosen at a level which is barely above the noise level in the image while still not identifying noise as features.
This technique allows the identification of very faint features on these images. We can, for example, reliably identify faculae on continuum images and study their maximum contrasts as a function of \mu. These increase with decreasing \mu, on average, but have a good deal of scatter. By comparison, maximum facular contrasts on the K images show no obvious dependence on \mu. The sunspot deficit, which is the main component of total solar irradiance (TSI) models, becomes somewhat more negative with this identification when compared to a simple threshold identification at -8.5% contrast, normally used for SFO data. Sunspot areas increase substantially, and are roughly 50% larger than those reported in, for example, the Solar Geophysical Data bulletin, due to the inclusion of more low-contrast pixels. We are investigating the effects of these new feature identifications on total solar irradiance models.
This research has been partially supported by NSF Grant ATM95-04374 and NASA Grants NAGW-3017 and NAG5-4973.