GEOPHYSICAL RESEARCH LETTERS, VOL. 23, NO. 4, PAGES 359-362, FEBRUARY 15, 1996 Comparison of proxy records of climate change and forcing solar Thomas J. Crowley Department of Oceanography, Texas A&M University,College Station, Texas Kwang-Yul Kim ClimateSystem Research Program, Department of Meteorology, Texas A&M University, CollegeStation, Texas Abstract. The role of solar variability as a causeof decadal- centennial scale climate change continues to be a topic of debate. Herein we compare two estimatesof solar variability since 1600 with two estimates of northern hemispheric temperature change. Sun-climate correlations vary between 0.57-0.74 and are significant at the 95% level. However, the small number of oscillations in the time seriessuggests to us that longer records are needed for a more convincing demonstration of a significant sun-climatelinkage. Forcing of an energy balance model with the solar time series indicates that modest amounts of solar variability and climate sensitivity yield temperature variations that represent a significant fraction of the inferred changes during the Little Ice Age. This latter result further supports the hypothesis that solar variability may be significantly contributing to climate changeon decadal-centennial time scales. Introduction The role of solar variability as a cause of decadal-centennial scale climate change continues to be a topic of debate. Although previous work indicates that significant spectral peaks of presumedsolar origin occur in many climate records [e.g., Crowley and Kim, 1993], other records do not show much indication of a solar influence. This "flickering" relationship is very different from Pleistocene studies, in which nearly every record shows the influence of orbital forcing. One explanation for the patchiness of sun-climate correlations involves the possibility of significant regional overprint of the atmosphere and ocean circulation on any perturbation due to solar variability. In order to better address this effect it is necessary to examine hemispheric-scale estimatesof climate change, for to a first approximationthe integrated effects of circulation are minimized in large-scale averages. Using the above approach, Friis-Christensen and Lassen [1991] and Reid [1991] found significant correlations over the last century betweenenvelopeindicesof solar variability with northern hemisphere temperatures and global sea surface temperatures, respectively. Further progress occurred when Hoyt and Schatten [1993] correlated their index of solar variability with a hemispheric time series developed by Groveman and Landsberg [1979] and estimated that as muchas 50% of the variance on decadal-centennial time scales could be attributed to solar forcing. Lean et al. [1995] conducted a Copyright1996 by the American Geophysical Union. Paper number 96GL00243 0094-8534/96/96GL-00243503.00 similar exercise with different sets of indices and found evidence for a sun-climate relationship comparable to or larger thanthat foundby Hoyt and Schatten [ 1993]. Because the above studies employed different solar and climate indices and reached different conclusions about the magnitudes of the sun-climate correlation, some clarification of the problem is required in terms of testing sensitivity of conclusions to choice of index utilized. In this study we build uponthe above-cited efforts by intercomparing correlations between different climate and solar indices and conducting an energy balancemodel simulation in order to evaluatethe sensitivity of the system response to solarforcing. Sun-climate Correlations Figure 1 illustrates theproxy solar andclimate time series. The solar indices were constructed differently. The Hoyt- Schatten [1993] index represents the estimatedirradiance changes due to changes in factors such as the length of the solar cycle,the normalized decay rate of the solar cycle,and the meanlevel of solar activity.The magnitude of the Hoyt- Schattensolar constantchange is slightly larger (0.30% vs. 0.24%) thanthat estimated by Lean et al. [1995]. The Lean et al. [1995] index is based on two terms. The first is an ~11-year periodvariation that estimates solarirradiance based on the sunspot/irradiance relationship observedfrom satellites during the last fifteen years. The background variations estimated by Leanet al. [1992]are based on comparison of the present range of solar activity with the observed range of activity in solar-type stars thatis inferred from emission lines in Ca-II, which correlate with solar luminosity variations. Because the present level of solaractivity is typical of the high end of variability observed in solar-typestars, the potential reductionin solar activity during the Maunder Minimum (about 1650-1710; cf. Eddy, 1976)can be estimated. The time variations of the background variations were then fitted to the revised estimate of the envelope of group sunspot numbers determined by Hoyt et al. [1994]. The Lean et al [1995] index is ~20 years out-of-phase with the Hoyt-Schatten indexbecause the primary building blocks for the two time series (cycle amplitude andcyclelength)are also ~20 years out-of-phase. Because observations over the last fifteen years indicate only about a 0.1% change in the solar constant, it is useful to address an oft-raisedobjection to the above two indices that their estimated range is larger than the observed range. However, there are several converging lines of evidence supporting at leastsome higher level of pastsolarirradiance changes: (1) the observations in solar-type stars for such behavior [Baliunas and Jastrow, 1990]; (2) spot sounding measurementsfrom the 1960s for larger variability than observed over the last fifteen years [Fr6hlich, 1987]; (3) the 359