This page will feature both publications arising from the Team meetings and short summaries of published articles on which the Team will build.



Some “essential reading” on the sunspot recalibration effort, courtesy of Laure Lefevre:

Here is a summary of proposed reading for the future ISSI workshops on the Recalibration of the Sunspot Number Series.

First, historical background about the Sunspot Number (Zurich method) can be found in Friedli (2016a) and Svalgaard, Cagnotti and Cortesi (2017).

Friedli (2016a): doi:10.1007/s11207-016-0907-0
Svalgaard, Cagnotti and Cortesi (2017): doi:10.1007/s11207-016-1024-9

Of course the historical reconstruction of the Group Number can be found in Hoyt & Schatten (1998a and b).

Hoyt & Schatten (1998a): doi:10.1023/A:1005007527816
Hoyt & Schatten (1998b): doi:10.1023/A:1005056326158

As the latest workshops were organized in triads the rest is organized in order of the program of the latest sworkshop held in Brussels June 12-15th 2017.

1. Clette & Lefèvre (2016): Clette et al. (2016a 2016b) and Clette et al. (2014) describe corrections applied directly to version one of the Sunspot Number SSN_V1 (published on the SILSO website: to SSN_V2 ( Contrary to the GN series described below, this is not a reconstruction based on the original sources, but rather a set of corrections applied after carefully establishing the reasons behind them, the periods of correction, and defining the level at which they should be applied. On this subject, the “Waldmeier Jump” due to a weighting of the spots in the counting of stations of the Zurich network is also discussed in the series of articles by Mike Lockwood (see point 2).

Clette et al. (2014):, doi:10.1007/s11214-014-0074-2
Clette et al. (2016a):
Clette et al. (2016b):, doi:10.1007/s11207-016-0875-4

2. Lockwood et al. (2014, 2016): The output series described in Lockwood et al. (2014, 2016) concern the SSN and focus on the correction to be applied around 1945. The so-called “Waldmeier Jump” referred to in the last section. The Lockwood method basically compares different series supposed to have similar properties as the SSN (related to the SSN like geomagnetic data, or very close to the SSN_V1, such as the Number of groups and sunspot areas extracted from the Royal Greenwich Observatory data) around the jump assuming that the mean difference before and after the jump should be consistent with zero. They compare the series to SSN_V1 before and after the alleged jump with one side of the jump corrected for different supposed jump levels in SSN_V1.

Lockwood et al. (2014):;jsessionid=81F0BFB70D64EB87585BCC59E3248BD4.f03t01
Lockwood et al. (2016a):,doi:10.1007/s11207-016-0855-8
Lockwood et al. (2016d):, doi:10.1007/s11207-016-0967-1

3. Friedli (2016) proposes a Recalculation of the Wolf series from 1849 to 1893. However, as most of the methods used by the authors are very disputable, it has still not been accepted for publication. Here is a summary of what should appear in the paper if it is published.

Homogeneity is considered as the most important property of the Wolf series of sunspot relative numbers since without a stable scale no valid conclusions about variations in the long-term course of the solar activity can be drawn. However, the homogeneity testing of the Wolf series is a difficult task, since the raw data and the data reduction methods are widely unknown. Based on hitherto unpublished original sources we reconstruct the calibration algorithms and data reduction methods and discuss their impact on the homogeneity of the Wolf series. Furthermore, we provide evidence that the practice of weighting the individual spots according to their extent started already with the assistants of Wolfer and was passed to their successors, including Brunner and Waldmeier. Thus, no additional long-term correction has to be applied to keep the homogeneity of the Wolf series. Based on Alfred Wolfer as standard observer, we recalculate the course of the Wolf series from 1849 to 1893, correcting for the widely disregarded diminishing of Wolf’s eyesight during the late 1870s and the 1880s and for the hitherto unrecognized scale jump from the 83/1320~mm Fraunhofer refractor to the 40/700~mm Parisian instrument. Before 1849 the level of the Wolf series has to be lowered further, since Wolf overestimated the k- factor of Heinrich Schwabe.

4. Svalgaard & Schatten (2016): This article describes a reconstruction of a complete Group Number series using a backbone method. Contrary to the excessive use of “Daisy chaining” used in previous reconstructions, the backbone enables to actually minimize the number of links in the “chain”. Lockwood et al. (2016b and c) discuss the use of Daisy chaining and the propagation of possible errors made on the linear fit between two observers.

Lockwood et al. (2016b):, doi:10.1007/s11207-016-0913-2
Lockwood et al. (2016c):, doi:10.1007/s11207-015-0829-2

5. Usoskin et al. (2016a): the main paper describing the Active Day Fraction (ADF) method is Usoskin et al. (2016a). A slight modification of the method with new observers added is developped in Willamo et al. (2017) and the computation of monthly and yearly values is described in Usoskin, Mursula and Kovaltsov (2003). Usoskin et al. (2016b) analyses in more details the possibility of assuming the difference of reported groups is stronlgy (i.e. strongly enough that other effects are negligible) related to the acuity of the observer, i.e. his capacity at seing/reporting spots of a size above a certain threshold.

Usoskin, Mursula and Kovaltsov (2003):
Usoskin et al. (2016a):, doi:10.1007/s11207-015-0838-1
Usoskin et al. (2016b):, doi:10.1007/s11207-016-0993-z
Willamo et al. (2017):,

Note that, apart from the ADF method, the Usoskin et al. (2016) paper also contains a new method to scale on observer to the other based on “correspondance matrices”. This part of the work is used especially in Chatzistergos et al. (2017).

6. Chatzistergos et al. (2017): This article describes a “backbone” reconstruction of group numbers that is based on the “correspondance matrices” described in Usoskin et al. (2017) instead of using a “simple” linear dependency between observers as in Svalgaard and Schatten (2016). Also note that the data from Svalgaard and Schatten (2016) is based on the scaling of yearly data, while the series from Chatzistergos et al. (2017) is based on daily data that is later interpolated to yearly.

Chatzistergos et al. (2017):,

In that vein, Andres-Munoz Jaramillo discussed a way of doing the “backbone” without any human bias at the Sunspot Workshop in Reading in Feb. 2017, by constructing the backbone with unbiased criteria based on the quality of the relations between observers and choosing who to link to who with an automatic method. You can ask him about his ideas.

Two articles from Ed Cliver are also really interesting to read. Cliver & Ling (2016) investigates how the original Hoyt & Schatten (1998) series was built, including some inconsistencies and anomalies and Cliver (2017) confronts the recent reconstructions with the ground truth that observations cannot degrade with time between the 19th and 20th century.

Cliver & Ling (2016): doi:10.1007/s11207-015-0841-6
Cliver (2017): doi:10.1007/s11207-016-0929-7

Aside from these references to specific time series, the articles from Thierry Dudok de Wit on how to fill gaps and the study of uncertainties in the Sunspot Number might be of interest.

Dudok de Wit (2011):,
Dudok de Wit et al. (2016):, doi:10.1007/s11207-016-0970-6