A Generalized Linear Mixed Model for Enumerated Sunspots (Abstract)

Volume 40 number 1 (2012)

Jamie Riggs
Department of Applied Science and Research Methods, McKee 518, Campus Box 124, University of Northern Colorado, Greeley, CO 80639; Jamie.Riggs@unco.edu

Abstract

(Abstract only) Monthly sunspot counts data from consistently submitting AAVSO observers were provided to determine monthly average sunspot numbers and the individual observer parameters that correct each observer’s counts to the monthly average. The data span a fourteen-month period from May 2010 through June 2011. The parameters are determined from a mixed-effects, loglinear model constructed specifically from the fourteen months of Poisson-distributed sunspot numbers. This model differs in the treatment of the data distribution assumptions of the existing linear regression model developed by Shapley (1949). The loglinear model methodology exceeds the correction coefficient performance criteria set by Shapley, and provides a method for determining the relative sunspot number reported monthly by the American Association of Variable Star Observers Solar Section. Model improvements are discussed.