Low-dimensional chaos and fractal properties of long-term sunspot activity

Two primary solar-activity indicators sunspot numbers(SNs)and sunspot areas(SAs)in the time interval from November 1874 to December 2012 are used to determine the chaotic and fractal properties of solar activity.The results show that(1)the long-term solar activity is governed by a low-dimensional chaotic strange attractor,and its fractal motion shows a long-term persistence on large scales;(2)both the fractal dimension and maximal Lyapunov exponent of SAs are larger than those of SNs,implying that the dynamical system of SAs is more chaotic and complex than SNs;(3)the predictions of solar activity should only be done for short-to mid-term behaviors due to its intrinsic complexity;moreover,the predictability time of SAs is obviously smaller than that of SNs and previous results.     

Analysis of Sunspot Activity Cycles

A nonlinear analysis of the daily sunspot number for each of cycles 10 to 23 is used to indicate whether the convective turbulence is stochastic or chaotic. There is a short review of recent papers considering sunspot statistics and solar activity cycles. The differences in the three possible regimes – deterministic laminar flow, chaotic flow, and stochastic flow – are discussed. The length of data sets necessary to analyze the regimes is investigated. Chaos is described and a chronology of recent results that utilize chaos and fractals to analyze sunspot numbers follows. The parameters necessary to describe chaos – time lag, phase space, embedding dimension, local dimension, correlation dimension, and the Lyapunov exponents – are determined for the attractor for each cycle. Assuming the laminar regime is unlikely if chaos is not indicated in a cycle by the calculations, the regime must be stochastic. The sunspot numbers in each of cycles 10 to 19 indicate stochastic behavior. There is a transition from stochastic to chaotic behavior of the sunspot numbers in cycles 20, 21, 22, and 23. These changes in cycles 20 – 23 may indicate a change in the scale of turbulence in the convection zone that could result in a change in the convective heat transfer and a change in the size of the convection region for these four cycles.     

Identifying the quasi-regular and stochastic components of solar cyclicity and their properties

We have averaged over every Carrington semi-rotation (C.s.-r.), the daily Wolf numbers (RW), total areas of sunspot groups (SA), the 10.7-cm radio flux (F _10.7), and the modulus of the mean magnetic field (|SMMF|). The fractal method of scaling the variance of time series was used to separate the regular and stochastic components. The manifestation of chaotic and stochastic properties of these components was investigated by testing with the methods of chaotic dynamics, as well as with two new methods: (1) close return maps; and (2) multivariate scaling analysis. Results: (1) by separating time series of global indices of solar activity, it is possible to identify the quasi-regular (the quasi-regularity is caused not by the absolute smoothness of the function) component on time scales longer than two years, and the irregular component on time scales shorter than two years; (2) the regular component has the properties of a nonlinear quasi-periodic oscillator; (3) the irregular component is a random one and has the properties of chromatic noise; and (4) by investigating the nonlinear connection of the solar activity indices under consideration it was found that such a connection is strong between F _10.7and RW. A nonlinear correlation between the attractors RW–|SMMF| and F _10.7–|SMMF| was also revealed.     

Nonlinear Time Series Analysis of Sunspot Data

This article deals with the analysis of sunspot number time series using the Hurst exponent. We use the rescaled range (R/S) analysis to estimate the Hurst exponent for 259-year and 11 360-year sunspot data. The results show a varying degree of persistence over shorter and longer time scales corresponding to distinct values of the Hurst exponent. We explain the presence of these multiple Hurst exponents by their resemblance to the deterministic chaotic attractors having multiple centers of rotation.     

Hurst analysis of Mt. Wilson rotation measurements

I study the temporal variation of the solar rotation on time scales shorter than the 11-year cycle by analyzing the daily Mt. Wilson Doppler measurements from 1967 to 1992. The differential rotation is represented by the three coefficients,A, B, andC, of the following expansion: =A +B sin²() +C sin⁴(). TheA, B, andC time series show clearly the 11-year solar cycle and they also show high-frequency fluctuations. The Hurst analysis of these time series shows that a Gaussian random process such as observational noise can only account for fluctuations on time scales shorter than 20 days. For time scales from 20 days to 11 years, the variations of A give rise to a Hurst exponent ofH = 0.83, i.e., the variations ofA are persistent. The temporal variations ofB show the same behavior asC, which is different fromA. From one to 11 years, theB andC variations are dominated by the 11-year cycle, while for time lags shorter than about 250 days, theB andC fluctuations give rise to a Hurst exponent ofH = 0.66, which lies betweenH = 1/2, of a Gaussian random process, and the exponent of the persistent process shown byA. An analysis of the equivalent coefficients of the first three even Legendre polynomials, computed usingA, B, andC, provides additional information. For time scales between 100 and 1000 days, the ranges,R/S, of Legendre polynomial coefficients decrease with increasing order of the polynomials which suggests that the persistent process operates mainly on large spatial scales. The Hurst exponent ofH = 0.83 for variations inA is the same asH for monthly sunspot numbers with time scales between 6 months and 200 years and for¹⁴C radiocarbon data with time scales between 120 years and 3000 years, previously analyzed by other authors. The combined results imply that the underlying solar process shows the same persistent behavior for time scales as short as about 20 days up to time scales of a few thousand years.Operated by the Association of Universities for Research in Astronomy, Inc. under cooperative agreement with the National Science Foundation.     

Periodicities in the solar differential rotation,surface magnetic field and planetary configurations

Using Greenwich data on sunspot groups during 1874–1976, we have studied the temporal variations in the differential rotation parametersA andB by determining their values during moving time intervals of lengths 1–5 yr successively displaced by 1 yr. FFT analysis of the temporal variations ofB (orB/A) shows periodicities 18.3 ± 3 yr, 8.5 ± 1 yr, 3.9 ± 0.5 yr, 3.1 ± 0.2 yr, and 2.6 ± 0.2 yr at levels 2. This analysis also shows five more periodicities at levels 1–2. The maximum entropy method is used to set narrower limits on the values of these periods. The reality of the existence of all these periodicities ofB (orB/A ) except the one at 2.8 yr is confirmed by analyzing the simulated time series ofB andB/A with values ofA andB randomly distributed within the limits of their respective uncertainties. Four of the prominent periods ofB agree, within their uncertainties, with the known periods in the the large-scale photospheric magnetic field. The deviations from the average differential rotation are larger near the sunspot minima. On longer time scales, the variations in the amount of sunspot activity per unit time are well correlated to the variations in the amplitudes of the torsional oscillation represented by the 22-yr periodicity inB. All the periods inB found here are in good agreement with the synodic periods of two or more consecutive planets. The possibility of planetary configurations providing perturbations needed for the Sun's MHD torsional oscillations is speculated upon and briefly discussed.     

大行星轨道运动与太阳黑子数的中长周期变化

本文对不同序列的太阳黑子数资料作了分析研究,计算得到了可能的太阳黑子活动的中长周期变化,并分别与由大行星轨道运动引起的日心轨道角动量变化的周期进行比较,发现二者具有比较一致的谱结构。基于本文的讨论和文[17]的结论,我们进一步认为大行星轨道运动是太阳黑子数周期性变化的可能的外部因素。     

Bimodality and the Hale cycle

Because of the bimodal distribution of sunspot cycle periods, the Hale cycle (or double sunspot cycle) should show evidence of modulation between 20 and 24 yr, with the Hale cycle having an average length of about 22 yr. Indeed, such a modulation is observed. Comparison of consecutive pairs of cycles strongly suggests that even-numbered cycles are preferentially paired with odd-numbered following cycles. Systematic variations are hinted in both the Hale cycle period and R _sum (the sum of monthly mean sunspot numbers over consecutively paired sunspot cycles). The preferred even-odd cycle pairing suggests that cycles 22 and 23 form a new Hale cycle pair (Hale cycle 12), that cycle 23 will be larger than cycle 22 (in terms of R _M, the maximum smoothed sunspot number, and of the individual cycle value of R _sum), and that the length of Hale cycle 12 will be longer than 22 yr. Because of the strong correlation (r = 0.95) between individual sunspot cycle values of R _sum and R _M, having a good estimate of R _Mfor the present sunspot cycle (22) allows one to predict its R _sum, which further allows an estimation of both R _Mand R _sum for cycle 23 and an estimation of R _sum for Hale cycle 12. Based on Wilson's bivariate fit (r = 0.98), sunspot cycle 22 should have an R _Mequal to 144.4 ± 27.3 (at the 3- level), implying that its R _sum should be about 8600 ± 2200; such values imply that sunspot cycle 23 should have an R _sum of about 10500 ± 2000 and an R _Mof about 175 ± 40, and that Hale cycle 12 should have an R _sum of about 19100 ± 3000.     

Periodogram analysis of 240 years of sunspot records

We have analyzed the direct records of sunspot number between 1749 and 1990 with the same technique currently used in the study of stellar activity cycles observed with Mount Wilson Observatory's 60-inch telescope. In order to mimic the stellar time series, which span only two decades, we analyzed twenty- and fifty-year intervals of the sunspot data in comparison to the entire record. We also examined the reliability of the oldest (pre-1850) sunspot records. The mean solar cycle period determined from the entire record (1749–1990) is 11.04 yr with a computed precision of ± 0.01 yr, but an overall accuracy of only ±1.1 yr. The large uncertainty is caused by variation of the cycle period with time and not observational uncertainty.The correct sunspot period is found slightly more often (82%) in 50-year intervals compared to 20-year (74%). The cause is twofold: first, a more precise period results from the longer sample length, and second, other periodicities exist in the sunspot record, so that a more accurate determination of the dominant 11.0-year period results from the longer time series. As a guideline for cycle periodicities in other stars, the solar results indicate that the 50-year intervals would produce more precise and accurate periods than the 20-year time series. On the other hand, useful statistics concerning long-term activity could be obtained from a less-frequently sampled group of stars that is substantially larger than the group of 100 lower Main-Sequence stars currently observed at Mount Wilson, although knowledge of short-term variability would be sacrificed.Pre-doctoral fellow, Harvard-Smithsonian Center for Astrophysics.     

Daily relative sunspot numbers 1749–1848: reconstruction of missing observations

A great part of missing daily relative sunspot numbers in the time interval 1749–1848 was reconstructed by nonlinear two-step method of interpolation. In the first step gaps of missing observations not longer than five days were directly interpolated. In the second step data were sorted to so-called Bartels scheme, i.e., to rows of the length of 27 days subsequently ranged in a matrix. In this step the missing value at any position was interpolated from the data at the same position of preceding and following rows. The interpolation was limited to sequences of no more than four missing data. The procedure enables to interpolate long gaps and simultaneously to respect the 27-day variation of solar activity. Monthly and annual means of relative sunspot numbers are presented. The differences between monthly and annual means of the primary observations and of the data completed by interpolation fluctuate around zero. The amplitude of fluctuations depends inversely on the frequency of observations. Most conspicuous are the deviations in the time interval 1776–1795 where the frequency of observations is very low or almost zero. The average dispersion of monthly differences is ±11.5 R and that of annual differences is ±7.8 R. The two-step method of interpolation was tested on the series of daily data in the time interval 1918–1948. The sequence of missing daily data in the years 1818–1848 represents a masking function. This function was applied to the continuous data series in the time interval 1918–1948 and then the modified series was reconstructed. The differences between the monthly and annual means of primary and reconstructed data are small with fluctuations around zero and with dispersion for monthly differences ±2.7 R a for annual differences ±0.6 R. Corresponding dispersions of the data differences for monthly means in the time interval 1818–1848 are ±4.3 R and ±1.1 R for annual means. The small dispersion values and small differences among them give evidence about the applicability and the effectiveness of the nonlinear two-step method of interpolation and also about high credibility of relative sunspot numbers after reconstruction.     

Relative sunspot numbers in the first half of eighteenth century

We revised relative sunspot numbers in the time interval 1700–1748 for which Wolf derived their annual means. The frequency of daily observations, counting simultaneously the number of sunspots and the number of sunspot groups necessary for determinating Wolf's relative sunspot numbers, is in this time interval very low and covers, on average, 4.8% of the number of all days only. There also exist incomplete observations not convenient to determine relative sunspot numbers. To enlarge the number of daily relative sunspot numbers we used the nonlinear, two-step interpolation method derived earlier by Letfus (1996, 1999). After interpolation, the mean value increased to 13.8%. Waldmeier (1968) found that the scaling factor k can be derived directly from the observed number of spots f and from the number of sunspot groups g. From the observations made at Zürich (Wolf and his assistants, Wolfer), at Peckeloh, and at Moncalieri during the years 1861–1928, we derived a new, more correct empirical relation. The resulting annual relative sunspot numbers are given in Table II. However, only for 26 years (53.0%) from the total number of 49 years was it possible to derive annual relative sunspot numbers. The observations were missing for the other years. This corresponds with results of Wolf, which gives the annual relative sunspot numbers for all 49 years. For the years when the data were missing, he marked these values as interpolated or very uncertain ones. Most of the observations originate from two data series (Kirch, Plantade), for which Wolf derived a higher scaling factor (k=2.0) than followed from the newly derived relation (k=1.40). The investigated time interval covers four solar cycles. After our results, the height of the first cycle (No. –4), given by Wolf, should be lowered by about two-thirds, the following two cycles (Nos. –3 and –2) lowered by one-third, as given by Wolf, and only the height of the fourth one (No. –1) should be unchanged. The activity levels of the cycles, as represented by group sunspot numbers, are lower by about one-fourth and, in the case of the first one (No. –4) even by two-thirds of the levels derived by us. The group sunspot numbers, derived from a much greater number of observations, have also greater credibility than other estimates. The shapes of the cycles, as given by Wolf, can be considered only as their more or less idealized form.     

Fine structure in the sunspot spectrum-2 to 70 years

Application of a new data adaptive approach to power spectrum estimation has yielded evidence for a double solar cycle line in the Zurich sunspot time history. There is significant power from 8 to 15 yr in the spectrum with the primary line at 11.1 yr and three attendant multiplets that may be significant. The first four harmonics of the solar cycle are detected too. Quite marginal evidence for a peak at 65 yr in the spectrum is presented. These results closely correspond to those recently found in the geomagnetic spectrum.     

Series of classical solar activity indices: Kislovodsk data

We present data on the series of solar activity indices, Wolf sunspot numbers W and total sunspot areas S, obtained at the Kislovodsk high-altitude station of the Pulkovo Observatory. The problem of properly extending the 133-year-long Zürich series of W and the 102-year-long Greenwich series of S, which were discontinued in 1980 and 1976, respectively, is emphasized. We stress that the Kislovodsk data have retained mutual homogeneity with the classical series until now and that they are preferred for extension. The question under consideration is of fundamental importance in studying the solar activity variations on long time scales and related processes in the Sun-Earth system.     

Sunspot groups at high latitude

Data of sunspot groups at high latitude (35°), from the year 1874 to the present (2000 January), are collected to show their evolutional behaviour and to investigate features of the yearly number of sunspot groups at high latitude. Subsequently, an evolutional pattern of sunspot group number at high latitude is given in this paper. Results obtained show that the number of sunspot groups of a solar cycle at high latitude rises to a maximum value about 1 yr earlier than the time of the maximum of sunspot relative numbers of the solar cycle, and then falls to zero more rapidly. The results also show that, at the moment, solar activity described by the sunspot relative numbers has not yet reached its minimum. In general, sunspot groups at high latitude have not appeared on the solar disc during the last 3 yr of a Wolf solar cycle. The asymmetry of the high latitude sunspot group number of a Wolf solar cycle can reflect the asymmetry of solar activity in the Wolf solar cycle, and it is suggested that one could further use the high latitude sunspot group number during the rising time of a Wolf solar cycle, maximum year included, to judge the asymmetry of solar activity over the whole solar cycle.     

Detection of the Long-Term Microwave `Darkening' Before the 14 July 2000 Flare

The latitudinal distribution of sunspot groups over a solar cycle is investigated. Although individual sunspot groups of a solar cycle emerge randomly at any middle and low latitude, the whole latitudinal distribution of sunspot groups of the cycle is not stochastic and, in fact, can be represented by a probability density function of the distribution having maximum probability at about 15.5°. The maximum amplitude of a solar cycle is found to be positively correlated against the number of sunspot groups at high latitude (35°) over the cycle, as well as the mean latitude. Also, the relation between the asymmetry of sunspot groups and its latitude is investigated, and a pattern of the N-S asymmetry in solar activity is suggested.     

Synchronization in Sunspot Indices in the Two Hemispheres

Historical sunspot records were analyzed by means of nonlinear tools to find synchronization phenomena at different time scales on the Sun. Using cross-recurrence plots it is shown that the north – south sunspot synchronization demonstrates a set of distinct periodic oscillations – 43.29, 18.52, and 7.63 years. Also we have traced the sunspot synchronization on shorter time scales. Very rare and episodic synchronization within half of the Carrington rotation rate was detected. By using the empirical mode decomposition technique the north – south sunspot time series were decomposed into intrinsic oscillatory modes. To determine which modes of the signal are responsible for synchronization we separated them into high- and low-frequency parts. It is shown that phase synchronization is detected only in the low-frequency modes. The high-frequency component demonstrates noisy behavior with amplitude synchronization and strong phase mixing.     

Forecasting the Time Series of Sunspot Numbers

Forecasting the solar cycle is of great importance for weather prediction and environmental monitoring, and also constitutes a difficult scientific benchmark in nonlinear dynamical modeling. This paper describes the identification of a model and its use in the forecasting the time series comprised of Wolf’s sunspot numbers. A key feature of this procedure is that the original time series is first transformed into a symmetrical space where the dynamics of the solar dynamo are unfolded in a better way, thus improving the model. The nonlinear model obtained is parsimonious and has both deterministic and stochastic parts. Monte Carlo simulation of the whole model produces very consistent results with the deterministic part of the model but allows for the determination of confidence bands. The obtained model was used to predict cycles 24 and 25, although the forecast of the latter is seen as a crude approximation, given the long prediction horizon required. As for the 24th cycle, two estimates were obtained with peaks of 65±16 and of 87±13 units of sunspot numbers. The simulated results suggest that the 24th cycle will be shorter and less active than the preceding one.     

Modeling of Sunspot Numbers by a Modified Binary Mixture of Laplace Distribution Functions

This paper presents a new approach for describing the shape of 11-year sunspot cycles by considering the monthly averaged values. This paper also brings out a prediction model based on the analysis of 22 sunspot cycles from the year 1749 onward. It is found that the shape of the sunspot cycles with monthly averaged values can be described by a functional form of modified binary mixture of Laplace density functions, modified suitably by introducing two additional parameters in the standard functional form. The six parameters, namely two locations, two scales, and two area parameters, characterize this model. The nature of the estimated parameters for the sunspot cycles from 1749 onward has been analyzed and finally we arrived at a sufficient set of the parameters for the proposed model. It is seen that this model picks up the sunspot peaks more closely than any other model without losing the match at other places at the same time. The goodness of fit for the proposed model is also computed with the Hathaway – Wilson – Reichmann measure, which shows, on average, that the fitted model passes within 0.47 standard deviations of the actual averaged monthly sunspot numbers.     

New Evidence for Long-Term Persistence in the Sun's Activity

Possible persistence of sunspot activity was studied using rescaled range and detrended fluctuation analyses. In addition to actual Wolf numbers (1700–2000 A.D.), two solar proxies were used in this research, viz., an annual sunspot proxy obtained for 1090–1700 A.D. and sunspot numbers reconstructed from the decadal radiocarbon series (8005 B.C. – 1895 A.D). The reconstruction was made using a five-box carbon exchange model. Analyses showed that in all cases the scaling exponent is significantly higher than 0.5 in the range of scales from 25 yr up to 3000 yr. This indicates the existence of a long-term memory in solar activity, in agreement with results obtained for other solar indices.     

Duration of Polar Activity Cycles and Their Relation to Sunspot Activity

We have defined the duration of polar magnetic activity as the time interval between two successive polar reversals. The epochs of the polarity reversals of the magnetic field at the poles of the Sun have been determined (1) by the time of the final disappearance of the polar crown filaments and (2) by the time between the two neighbouring reversals of the magnetic dipole configuration (l=1) from the H synoptic charts covering the period 1870–2001. It is shown that the reversals for the magnetic dipole configuration (l=1) occur on an average 3.3±0.5 years after the sunspot minimum according to the H synoptic charts (Table I) and the Stanford magnetograms (Table III). If we set the time of the final disappearance of the polar crown filaments (determined from the latitude migration of filaments) as the criterion for deciding the epoch of the polarity reversal of the polar fields, then the reversal occurs on an average 5.8±0.6 years from sunspot minimum (last column of Table I). We consider this as the most reliable diagnostic for fixing the epoch of reversals, as the final disappearance of the polar crown filaments can be observed without ambiguity. We show that shorter the duration of the polar activity cycle (i.e., the shorter the duration between two neighbouring reversals), the more intense is the next sunspot cycle. We also notice that the duration of polar activity is always more in even solar cycles than in odd cycles whereas the maximum Wolf numbers W _\max is always higher for odd solar cycles than for even cycles. Furthermore, we assume there is a secular change in the duration of the polar cycle. It has decreased by 1.2 times during the last 120 years.