GROUP SUNSPOT NUMBERS - A NEW SOLAR-ACTIVITY RECONSTRUCTION

In this paper, we construct a time series known as the Group Sunspot N umber. The Group Sunspot Number is designed to be more internally self -consistent (i.e., less dependent upon seeing the tiniest spots) and l ess noisy than the Wolf Sunspot Number. It uses the number of sunspot groups observed, rather than groups and individual sunspots. Daily, mo nthly, and yearly means are derived from 1610 to the present. The Grou p Sunspot Numbers use 65 941 observations from 117 observers active be fore 1874 that were not used by Wolf in constructing his time series. Hence, we have calculated daily values of solar activity on 111358 day s for 1610-1995, compared to 66 168 days for the Wolf Sunspot Numbers. The Group Sunspot Numbers also have estimates of their random and sys tematic errors tabulated. The generation and preliminary analysis of t he Group Sunspot Numbers allow us to make several conclusions: (1) Sol ar activity before 1882 is lower than generally assumed and consequent ly solar activity in the last few decades is higher than it has been f or several centuries. (2) There was a solar activity peak in 1801 and not 1805 so there is no long anomalous cycle of 17 years as reported i n the Wolf Sunspot Numbers. The longest cycle now lasts no more than 1 5 years. (3) The Wolf Sunspot Numbers have many inhomogeneities in the m arising from observer noise and this noise affects the daily, monthl y, and yearly means. The Group Sunspot Numbers also have observer nois e, but it is considerably less than the noise in the Wolf Sunspot Numb ers. The Group Sunspot Number is designed to be similar to the Wolf Su nspot Number, but, even if both indices had perfect inputs, some diffe rences are expected, primarily in the daily values.