Data inconsistency in sunspot detection

Abstract

Availability of integrated high quality information is a prerequisite for many intelligent knowledge based systems. Consistency of data plays an influential role in reassuring the quality of the integrated data. In this paper we discuss issues of data inconsistency in the domain of sunspot detection (i.e. a manifestation of solar activity). Sunspots data are collected from ground based observatories and also from instruments aboard satellites. Due to instrumentation limitations and inevitable subjectivity of human observers, the collected data bear some levels of inconsistency. This paper discusses issues regarding inconsistency in data used for performance validation of an automated sunspot tracking system. For evaluating the results, an integrated probability reference map is created, using knowledge integration, which reinforces data accuracy with higher certainty. Further, we use a weighted matching technique to reduce the impact of some data inconsistency.