Edwin J. Reichmann

A synthesis of solar cycle prediction techniques

A number of techniques currently in use for predicting solar activity on a solar cycle timescale are tested with historical data. Some techniques, e.g., regression and curve fitting, work well as solar activity approaches maximum and provide a month-by-month description of future activity, while others, e.g., geomagnetic precursors, work well near solar minimum but only provide an estimate of the amplitude of the cycle. A synthesis of different techniques is shown to provide a more accurate and useful forecast of solar cycle activity levels. A combination of two uncorrelated geomagnetic precursor techniques provides a more accurate prediction for the amplitude of a solar activity cycle at a time well before activity minimum. This combined precursor method gives a smoothed sunspot number maximum of 154 ± 21 at the 95% level of confidence for the next cycle maximum. A mathematical function dependent on the time of cycle initiation and the cycle amplitude is used to describe the level of solar activity month by month for the next cycle. As the time of cycle maximum approaches a better estimate of the cycle activity is obtained by including the fit between previous activity levels and this function. This Combined Solar Cycle Activity Forecast gives, as of January 1999, a smoothed sunspot maximum of 146 ± 20 at the 95% level of confidence for the next cycle maximum.

The shape of the sunspot cycle

The temporal behavior of a sunspot cycle, as described by the International sunspot numbers, can be represented by a simple function with four parameters: starting time, amplitude, rise time, and asymmetry. Of these, the parameter that governs the asymmetry between the rise to maximum and the fall to minimum is found to vary little from cycle to cycle and can be fixed at a single value for all cycles. A close relationship is found between rise time and amplitude which allows for a representation of each cycle by a function containing only two parameters: the starting time and the amplitude. These parameters are determined for the previous 22 sunspot cycles and examined for any predictable behavior. A weak correlation is found between the amplitude of a cycle and the length of the previous cycle. This allows for an estimate of the amplitude accurate to within about 30% right at the start of the cycle. As the cycle progresses, the amplitude can be better determined to within 20% at 30 months and to within 10% at 42 months into the cycle, thereby providing a good prediction both for the timing and size of sunspot maximum and for the behavior of the remaining 7–12 years of the cycle.

Evidence That a Deep Meridional Flow Sets the Sunspot Cycle Period

Sunspots appear on the Sun in two bands on either side of the equator that drift toward lower latitudes as each sunspot cycle progresses. We examine the drift of the centroid of the sunspot area toward the equator in each hemisphere from 1874 to 2002 and find that the drift rate slows as the centroid approaches the equator. We compare the drift rate at sunspot cycle maximum with the period of each cycle for each hemisphere and find a highly significant anticorrelation: hemispheres with faster drift rates have shorter periods. These observations are consistent with a meridional counterflow deep within the Sun as the primary driver of the migration toward the equator and the period associated with the sunspot cycle. We also find that the drift rate at maximum is significantly correlated with the amplitude of the following cycle, a prediction of dynamo models that employ a deep meridional flow toward the equator. Our results indicate an amplitude of about 1.2 m s 1 for the meridional flow velocity at the base of the solar convection zone.

An estimate for the size of cycle 23 based on near minimum conditions

The first occurrence of a high-latitude, new cycle spot group for cycle 23 was in May 1996, in conjunction with a minimum in the smoothed monthly mean sunspot number. Since then, new cycle spot groups have become more predominant, and the smoothed monthly mean sunspot number has slowly risen. Such behavior indicates that new cycle 23 probably had its minimum annual average sunspot number, R(min), equal to 8.7, in 1996. Because this value is larger than the average for R(min), cycle 23 is expected to have a maximum amplitude, R(max), that, likewise, will be larger than average, suggesting further that it probably will be both fast rising (i.e., peaking before May 2000) and of shorter than average length (i.e., ending before May 2007). Another parameter well correlated with R(max) is the minimum amplitude of the aa geomagnetic index, aa(mm), which usually occurs either in the year of R(min) occurrence or, more often, in the following year. For 1996 the annual average of aa measured 18.6. Presuming this value to be aa(min) for cycle 23, we calculate cycle 23s R(max) to be about 171.0 ± 17.6 (i.e., the 90% prediction interval), based on the stronger (r = 0.98) bivariate fit of R(max) versus both R(min) and aa(min). Comparison of this estimate with others, using various combinations of parameters, yields an overlap in the prediction intervals for R(max) of about 168 ± 15, a range that is within the consensus recently reported by Joselyn et al. [1997] (= 160 ± 30). Thus this study supports the view that cycle 23 will have an R(max) that will be larger than average but smaller than was seen for cycle 19, the largest cycle on record with R(max) = 190.2.

On the behavior of the sunspot cycle near minimum

The decline of cycle 22 is shown to be consistent with the notion that it will have a period <11 years. On the basis of the modern era of sunspot cycles, the average length of short-period cycles has been 123±3 months, suggesting that onset for cycle 23 will be about December 1996 (±3 months). As yet, no high-latitude (25° or more) new cycle spots have been reported. Because the occurrence of a high-latitude new cycle spot group has always preceded conventional cycle onset by at least 3 months, one infers that its occurrence is most imminent.

Evidence for systematic flows in the transition region around prominences

The solar transition region in the neighbourhood of prominences has been studied from observations with the Ultraviolet Spectrometer and Polarimeter of NASAs Solar Maximum Mission satellite. Dopplergrams from observations of the transition-region lines Civ λ 1548 Å and Siiv λ 1393 Å, which are formed at about 105 K, give velocity amplitudes typically in the range ± 15 km s-1. Prominences are found to be located very close to dividing lines between areas of up- and down-draughts in the transition-region. The observed pattern suggests that the 105 K gas flows take place within arcades of magnetic loops, which most likely are part of the supporting magnetic structure for the prominence matter. An additional band of blue-ward Doppler shifts is frequently seen close to quiescent prominences. This may be the source of outward flowing matter along the helmet streamers above filament channels.

Estimating the size and timing of maximum amplitude for cycle 23 from its early cycle behavior

On the basis of the lowest observed smoothed monthly mean sunspot number, cycle 23 appears to have conventionally begun in May 1996, in conjunction with the first appearance of a new cycle, high-latitude spot group. Such behavior, however, is considered rather unusual, since, previously (based on the data-available cycles 12–22), the first appearance of a new cycle, high-latitude spot group has always preceded conventional onset by at least 3 months. Furthermore, accepting May 1996 as the official start for cycle 23 poses a dilemma regarding its projected size and timing of maximum amplitude. Specifically, from the maximum-minimum and amplitude-period relationships we infer that cycle 23 should be above average in size and a fast riser, with maximum amplitude occurring before May 2000 (being in agreement with projections for cycle 23 based on precursor information), yet from its initial languid rate of rise (during the first 6 months of the cycle) we infer that it should be below average in size and a slow riser, with maximum amplitude occurring after May 2000. The dilemma vanishes, however, when we use a slightly later-occurring onset. For example, using August 1996, a date associated with a local secondary minimum prior to the rapid rise that began shortly thereafter (in early 1997), we infer that the cycle 23 rate of rise is above that for the mean of cycles 1–22, the mean of cycles 10–22 (the modern era cycles), the mean of the modern era “fast risers,” and the largest of the modern era “slow risers” (i.e., cycle 20), thereby suggesting that cycle 23 will be both fast rising and above average in size, peaking before August 2000. Additionally, presuming cycle 23 to be a well-behaved fast-rising cycle (regardless of whichever onset date is used), we also infer that its maximum amplitude likely will measure about 144.0 ± 28.8 (from the general behavior found for the bulk of modern era fast risers; i.e., 5 of 7 have had their maximum amplitude to lie within 20% of the mean curve for modern era fast risers). It is apparent, then, that sunspot number growth during 1998 will prove crucial for correctly establishing the size and shape of cycle 23.

Partial analysis of the flare-prominence of 30 April 1974

A portion of an east limb flare-prominence observed in Hα by NOAA/Boulder and NASA/ MSFC patrol facilities on 30 April 1974 is analyzed. Following a rapid (∼2 min) achievement of a maximum mass ejection velocity of about 375 km s−1, the ascending prominence reached a height of, at least, 2 × 105 km. We use a one-dimensional, time-dependent hydrodynamic theory (Nakagawa et al., 1975) to compute the total mass (∼2 × 1011 g) and energy (∼4 × 1026erg) ejected during this part of this event. Theoretical aspects of the coronal response are discussed. We conclude that a moderate temperature and density pulse (factors of ten and two, respectively), for a duration of only 3 min, is sufficient for an acceptable simulation of the Hα observations and the likely coronal response to the ascending prominence and flare-related ejections. No attempt was made to simulate the additionally-important spray and surge features which probably contributed a higher level of mass and energy efflux.

Observation of the impulsive phase of a simple flare

We present a broad range of complementary observations of the onset and impulsive phase of a fairly large (1B, M1.2) but simple two-ribbon flare. The observations consist of hard X-ray flux measured by the SMM HXRBS, high-sensitivity measurements of microwave flux at 22 GHz from Itapetinga Radio Observatory, sequences of spectroheliograms in UV emission lines from Ov (T ≈ 2 × 105 K) and Fexxi (T ≈ 1 × 107 K) from the SMM UVSP, Hα and Hei D3 cine-filtergrams from Big Bear Solar Observatory, and a magnetogram of the flare region from the MSFC Solar Observatory. From these data we conclude:(1)The overall magnetic field configuration in which the flare occurred was a fairly simple, closed arch containing nonpotential substructure.(2)The flare occurred spontaneously within the arch; it was not triggered by emerging magnetic flux.(3)The impulsive energy release occurred in two major spikes. The second spike took place within the flare arch heated in the first spike, but was concentrated on a different subset of field lines. The ratio of Ov emission to hard X-ray emission decreased by at least a factor of 2 from the first spike to the second, probably because the plasma density in the flare arch had increased by chromospheric evaporation.(4)The impulsive energy release most likely occurred in the upper part of the arch; it had three immediate products:(a)An increase in the plasma pressure throughout the flare arch of at least a factor of 10. This is required because the Fexxi emission was confined to the feet of the flare arch for at least the first minute of the impulsive phase.(b)Nonthermal energetic (∼ 25 keV) electrons which impacted the feet of the arch to produce the hard X-ray burst and impulsive brightening in Ov and D3. The evidence for this is the simultaneity, within ± 2 s, of the peak Ov and hard X-ray emissions.(c)Another population of high-energy (∼100keV) electrons (decoupled from the population that produced the hard X-rays) that produced the impulsive microwave emission at 22 GHz. This conclusion is drawn because the microwave peak was 6 ± 3 s later than the hard X-ray peak.

SMM/UVSP observations of the distribution of transition region oscillations and other properties in a sunspot

Observations of a sunspot in the Civ line at 1548 Å formed in the transition region have been analyzed to obtain the time variations and/or mean values of the velocity, intensity, longitudinal magnetic field, and line width. Oscillations with periods between approximately 110 and 200 s are observed only over the umbra where the transition region magnetic field is highest and the line width is smallest. When periodic intensity variations occur at the same frequency as the velocity oscillations, the peak intensities occur slightly before the maximum upward motions. No periodic variations in the transition region magnetic field have been detected. Scatter diagrams are presented which show possible relationships between the flow velocity, emission line intensity, line width, and transition region magnetic field.