North-south (N-S) asymmetry of solar activity is a known statistical phenomenon but its significance is difficult to prove or theoretically explain. Here we consider each solar hemisphere as a separate dynamical system connected with the other hemisphere via an unknown coupling parameter. We use a non-linear dynamics approach to calculate the scale-dependent conditional dispersion (CD) of sunspots between hemispheres. Using daily Greenwich sunspot areas, we calculated the Neumann and Pearson chi-squared distances between CDs as indices showing the direction of coupling. We introduce an additional index of synchronization which shows the strength of coupling and allows us to discriminate between complete synchronization and independency of hemispheres. All indices are evaluated in a four-year moving window showing the evolution of coupling between hemispheres. We find that the driver-response interrelation changes between hemispheres have a few pulses during 130 years of Greenwich data with an at least 40 years-long period of unidirectional coupling. These sharp nearly simultaneous pulses of all causality indices are found at the decay of some 11-year cycles. The pulse rate of this new phenomenon of dynamic coupling is irregular: although the first two pulses repeate after 22-year Hale cycles, the last two pulses repeat after three and four 11-year cycles respectively. The last pulse occurs at the decay phase of Cycle 23 so the next pulse will likely appear during the decay of future Cycle 25 or later. This new phenomenon of dynamic coupling reveals additional constraints for understanding and modeling of the long-term solar activity cycles.
Antarctic 'Vostok' station works most closely to the center of the icecap among permanent year-around stations. Climate conditions areexclusively stable: low precipitation level, cloudiness and windvelocity. These conditions can be considered as an ideal modellaboratory to study the surface temperature response on solarirradiance variability during 11-year cycle of solar activity. Here wesolve an inverse heat conductivity problem: calculate the boundaryheat flux density (HFD) from known evolution of temperature. Usingmeteorological temperature record during (1958-2011) we calculated theHFD variation about 0.2-0.3 W/m2 in phase with solaractivity cycle. This HFD variation is derived from 0.5 to 1oC/0.1 %. (Gal-Chen and Schneider in Tellus 28:108-1211975). The solar forcing (TSI) is disturbed by volcanic forcing (VF),so that their linear combination TSI + 0.5VF empirically provideshigher correlation with HFD (r = 0.63 ± 0.22) than TSI (r =0.50 ± 0.24) and VF (r = 0.4 ± 0.25) separately. TSIshows higher wavelet coherence and phase agreement with HFD than VF.
Authors: Volobuev D.M.
Publication Status: accepted to Climate Dynamics
Last Modified: 2013-09-11 07:22
Strong polarity-inversion line preceeding an X-class flare.
The forecast of the decadal average sunspot number (SN) becomes possible with an extension of telescopic observations based on proxy reconstructions using the tree ring radiocarbon data during the Holocene. These decadal numbers (SNRC) provide a powerful statistic to verify the forecasting methods. Complicated dynamics of long-term solar activity and noise of proxy-based reconstruction make the one-step-ahead forecast to be challenging for any forecasting method. Here we construct a continuous dataset of SNRC which extends the group sunspot number and the international sunspot number. The known technique of non-linear forecast, the local linear approximation, is adapted to estimate the coming SN. Both the method and the continuous dataset were tested and tuned to obtain the minimum of a normalized average prediction error (E) during the last millennium using several past millennia as a training dataset. E = 0.58SD is achieved to forecast the SN successive differences whose standard deviation is SD =7.39 for the period of training. This corresponds to the correlation (r=0.97) between true and forecasted SN. This error is significantly smaller than the prediction error when the surrogate data were used for the training dataset, and proves the non-linearity in the decadal SN. The estimated coming SN is smaller than the previous one.