Space Climate 7 Meeting Abstract
New reconstruction of the group sunspot number without explicit need for intercalibration or use of backbones
Thierry Dudok de Wit (University of Orléans)
Ed Cliver (National Solar Observatory, Boulder) Greg Kopp (LASP, University of Colorado)
The group sunspot number, which is a unique record of past solar activity, is based on a vast ensemble of different and often quite heterogeneous data sets. The recent reevaluation of the group sunspot number record has revealed the need for intercalibrating the different observers before combining their observations. In addition, a suitable method is required for stitching together data sets that only partly overlap in time.Here we present a new approach that bypasses the need for intercalibration and in addition avoids the artificial introduction of backbone observers for stitching records together. The first novelty is the combination of partly overlapping records in a natural way by means of a statistical method named expectation maximization. Thanks to this method, no specification of backbones or daisy-chaining is required. The second novelty is the use of order statistics to avoid intercalibration. More precisely, we combine ranked group sunspot numbers and not the original numbers.
Tests with synthetic data simulating observers that have different nonlinear responses confirm the robustness of the method. We illustrate the method with real data and propose a new reconstruction of the group sunspot number whose exact spot-to-observer scaling can now be freely determined.
Mode of presentation: poster