Pioneer, Voyager, and Ulysses solar cycle variations

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 102, NO. A5, PAGES , MAY 1, 1997 Three-dimensional nature of interaction regions- Pioneer, Voyager, and Ulysses solar cycle variations fromlto5au J. Am6rico Gonzglez-Esparza and Edward J. Smith Jet Propulsion Laboratory, California Institute of Technology, Pasadena Abstract. We combined in-ecliptic observations of interaction regions from I to 5 AU by four spacecraft that traveled to Jupiter at different phases of the solar cycle: Pioneer 11 (declining phase of cycle 20), Voyager I and 2 (ascending phase of cycle 21), and Ulysses Oust after solar maximum 22). We used this set of 97 interaction regions to study two different aspects of these events. (1) We analyzed the geometry of 38 stream interfaces, and found that the interaction regions detected by the three missions have different orientations. Some of these results might be associated with solar cycle variations such as the low-latitudinal inclination of the interaction regions detected by Ulysses and the high-latitudinal tilts of some interaction regions detected by Pioneer 11. Other results, such as the deviations with respect to the Parker spiral orientation, seem to be related to the persistent variations in the characteristics of the solar winds streams. (2) We studied the bulk speeds and the dynamic, magnetic, and thermal pressures of the fast and slow streams associated with 75 of these interaction regions. As a consequence of solar cycle variations, the bulk speed distributions of the fast and slow streams were different in the three missions. We found that the pressure ratios (dynamic, thermal, and magnetic) between fast and slow streams vary continuously. In about half of the interaction regions the dynamic pressure of the slow stream was higher than the dynamic pressure of the fast stream. We found similar variations in magnetic and thermal pressures. This implies that in many interaction regions the slow stream transfers momentum to the fast stream. The results presented here show that the solar wind streams associated with interaction regions vary at all the phases of the solar cycle, leading to interaction regions with different characteristics. 1. Introduction The aim of this work is to study two different aspects of the in-ecliptic interaction regions detected by four spacecrafthat traveled to Jupiter at different phases of the solar cycle. This combined set of 97 "trans-jovian" interaction regions provides useful information on the heliocentric evolution of these events and the variations with the solar cycle. Gonzdlez-Esparzand Smith eled during the ascending phase cycle 21; and Ulysses traveled during the post maximum of cycle 22 (see Figure I in paper 1). There were differences and similarities between the shock population, magnetic sectors, and interaction regions detected by the three missions. Several interesting questions arose from the previous study: How different are the interaction regions detected by Pioneer, Voyager, and Ulysses? How do the continual changes of solar wind streams affect the properties of [1996] (hereafter paper 1) unified and compared in- the interaction regions? How do the characteristics and ecliptic (non simultaneous) observations of large-scale orientation of the interaction regions change with the solar wind dynamics by Pioneers 10 and 11, Voyager 1 solar cycle? In this paper we report (1) the orientations and 2, and Ulysses from I to 5 AU. The three missions of the interaction regions and (2) variations in the proptraveled to Jupiter at different times: Pioneer traveled erties of the fast and slow solar wind streams associated during the descending phase of cycle 20; Voyager trav- with these interaction regions. Now at Instimto de Geofisica, Universidad Nacional Autbnoma de M6xico, Mexico City, Mexico. Copyright 1997 by the American Geophysical Union. Paper number 97JA /97/97JA Stream Interfaces A stream interface is a distinct boundary that separates fast and slow solar wind flows. This boundary is characterized by an abrupt drop in density and a similar increase in temperature and is associated with a high pressure region [Burlaga, 1974]. A stream interface in

2 9782 GONZ/ LEZ-ESPARZA AND SMITH: SOLAR CYCLE AND INTERACTION REGIONS the inner heliospher eventually evolves surrounded by 1974, fast solar wind (Vr _ 700 km/s) overtakeslow an interaction region, which gradually expands in the wind (V _ 400 km/s) causing an interaction region outer hellosphere [see, e.g., Siscoe 1976, Figure 2]. Sis- (shaded region). The stream interface separates the two coe et al. [1969] predicted east-west and north-south compressed flows and is recognized by the solar wind deflections of solar wind velocity close to stream inter- shear flow (Figure la), the abrupt drop in density, infaces. These deflections result from the high thermal crease in proton temperature, and the high magnetic and magnetic pressures associated with the interface field magnitude (Figures lb-lc). Note, however, that [see, e.g., $iscoe et al., 1969, Figure 1; Pizzo, 1991, not all the interaction regions contain a well-defined Figure 2]. The fast and slow streams carry different stream interface. In the combined set of 97 interaction magnetic fields, and cannot be mixed. There is no net regions about half of them did not contain a well-defined flow across the interface. Gosling et al. [1978] stud- stream interface (section 2.1). ied 23 interfaces detected by IMP 6, 7, and 8 at I AU between 1971 and They produced a superposed 1.2. Ulysses Observations of Interaction Regions at Midlatitudes epoch analysis to emphasize the overall structure of the stream interfaces and reported that (1) the behavior of Ulysses observations have increased interest in the protons, electrons and alpha particles across the inter- structure of the three-dimensional heliosphere. One of face suggests that it separates plasma of distinctly dif- the main results obtained by Ulysses after the Jupiter ferent origins; (2) there is a strong shear flow at the in- flyby was the absence, from about 28 ø to 38 ø south terface; (3) the magnetic field strength maximizes near latitude, of interplanetary shocks leading interaction the interface and is roughly constant across it; and (4) regions and the continuous presence of reverse shocks although their observations were made during the de- trailing these interaction regions [Gosling et al., 1993; scending phase of solar cycle 20, the interfaces did not Pizzo and Gosling, 1994; Balogh et al., 1995; Gonzdlezgenerally recur from one solar rotation to the next. Esparza et al., 1996]. On the basis of the strong lati- Figure I shows an example of an interaction region tudinal shear flows of solar wind velocity inside these detected by Pioneer 11 at 4.3 AU containing a well- interaction regions and the predictions by the threedefined stream interface. Figure l d shows the bulk dimensional model of corotating flows by Pizzo [1982, speed profile of this event. Between day 192 and 196 of 1991, 1994], Gosling et al. [1993] suggested that this a )... : , i 'iii!!iii :::::::':i'.. : :!:.:": i!i :::::i i ::' ':ii!! : : i ii i!i!i!!:!:' ":: iii i! ii:ii!!:iii! i!!i!i! i i i ii! iiiii!!11 '":i!i E 2 210s :191 74:192 74:193 74:194 74:195 74:196 Figure 1. Example of an interaction region (shadow region) detected by Pioneer 11 containing a welldefined stream interface. The interaction region is led by a forward shock Fs and is trailed by a reverse shock Rs. (a) The two polar angles of solar wind velocity in heliospheric coordinates ( b longitudinal and latitudinal) denoting a strong shear flow at the interface. (b) The discontinuity in density and proton temperature. (c) The magnetic field magnitude. (d) The solar wind radial velocity.

3 GONZ/i. LEZ-ESPARZA AND SMITH: SOLAR CYCLE AND INTERACTION REGIONS 9783 phenomenon could be explained if the interaction regions were tilted with respect to the solar rotation axis. 2. Orientation of Stream Interfaces This tilt geometry causes the front of the interaction In paper 1 we identified all the interaction regions region to evolve more strongly at low latitudes (close to detected by the five spacecraft from I to 5 AU, and the ecliptic plane) while the trailing edge evolves more plotted them in maps of large-scale features (Figures 3- strongly at higher latitudes [see, e.g., Pizzo, 1994, Fig- 7 in paper 1). As expecteduring the descending phase ure 3; Pizzo and Gosling, 1994, Figures 3 and 4] and of the cycle, Pioneer 11 detected a very regular pattern produces a strong latitudinal shear flow inside the in- of solar wind streams, magnetic sectors, and corotating teraction region. interaction regions. On the other hand, the other space- The three-dimensional (3-D) model of corotating flows craft observed variable patterns of solar wind dynamby Pizzo [1982, 1991, 1994] shows that two opposite ics. By visual inspection we noted that the solar wind tilted interaction regions are formed if (1) the solar plasma and magnetic field profiles of the interaction remagnetic field can be approximated by a dipole, and (2) gions detected by the three missions Were very different. there is a tilt angle between the solar rotation and mag- In this paper we will compare Pioneer 11, Voyager 1 netic axis [see, e.g., Pizzo, 1994, Figure 2]. These con- and 2, and Ulysses observations. We did not include ditions are a good approximation during the descend- Pioneer 10 observations because we consider that the ing phase of the solar cycle [Zhao and Hundhausen, large number (50) and regularity of the interaction re- 1981; Mihalov, 1990], which coincides with Ulysses' tra- gions detected by Pioneer 11 were a very good sample jectory after the Jupiter flyby. In order to corrobo- of thesevents at the descending phase of the cycle. We rate the predictions of the model, the normal of the used solar wind plasma data (1 hour averages) from the corotating shocks detected by Ulysses after the Jupiter National Space Science Data Center (NSSDC) which is flyby in 1992 and 1993 (from 6øto 38øsouth heliolatitudes) have been studied. Burton et al. [1996] and Riley et al. [1996]. used different and independent techniques based on the Rankine-Hugoniot relations to infer the normal direction of these corotating shocks. Both studies found that the distributions of shock normals were qualitatively consistent with the predictions by 3- D model of corotating flows. The results found by Riley et al. [1996] show that, on the whole, the forward cororating shocks were propagating equatorward and westward, while the reverse corotating shocks were stronger and propagating poleward and eastward as expected. However, the variations in the pattern of solar wind streams [Gosling et al., 1976; Gonzdlez-Esparza and Smith, 1996] and the evolution of coronal holes [Hundhausen et al., 1981] and the neutral sheet [Hoeksema, 1986] through the solar cycle suggesthat interaction regions should have a different geometry and characteristics at different phases of the solar cycle. Fast solar wind from equatorial coronal holes produces interaction regions with different geometries than the tilted interaction regions expected during the descending phase of the cycle [see, e.g., $iscoe, 1976, Figure 3; Pizzo, 1982, Figure16a]. The different sets of interaction regions detected by the three missions also represent three different phases of the solar cycle. In section 2 we will describe the analysis of the ori- entation of the interaction regions. To infer the orientation of the interaction regions, we used a different approach than the previous studies by Burton et al. [1996] and Riley et al. [1996]. We did not apply Rankine-Hugoniot relations to infer the normal of the shocks leading and trailing the interaction regions. But we analyzed the shear flow of solar wind velocity at the interface on the basis of a variation of the principal axis analysis technique. The method and the results are discussed in section 2. readily available for the four spacecraft. This temporal resolution is adequate to study the large-scale geometry, not the local struct. ure, if the shear flow is well-defined and the interface can be approximated by a tangential discontinuity (as in the example in Figure 1). Previous studies had used this temporal resolution to study other large-scale properties of stream interfaces [Siscoe, 1972; Gosling et al., 1978] Maximum Variance Analysis Siscoe [1972] analyzed the shear flows of solar wind velocity at five stream interfaces detected by Pioneer 6 at I AU. Under the assumption that the solar wind velocity along the interface normal direction was approximately constant (i.e., there was no net flow through the interface), they applied principal axis analysis to the solar wind velocity data surrounding the interface and identified the minimum variance direction (the eigenvector associated with the minimum eigenvalue) as the interface normal. They neglected the effects associated with the latitudinal flows and reported that the interfaces were closely oriented along the Parker spiral. There is a problem in applying the minimum variance technique to infer the interface normal direction. Although, in many cases, the maximum variance direction is well-defined along the shear flow velocity parallel to the interface plane, the minimum variance direction is ill-defined (any direction orthogonal to the maximum variance direction, in theory, has a zero variance). In order to infer a more accurate result and to solve this "velocity coplanarity problem", it is necessary to modify the technique. Note that we are assuming that the interface can be approximated to a tangential discontinuity. In the interface reference system, two normal vectors are contained in the interface plane: one vector along the maximum variance direction (shear flow

4 9784 GONZ/ LEZ-ESPARZA AND SMITH: SOLAR CYCLE AND INTERACTION REGIONS direction) and the second vector, which has a zero average value of solar wind velocity (the "noncoplanarity" component)., The third normal vector is the interface normal direction fisi. Along this vector the solar wind velocity also has a zero average value because there is no net flow across the stream interface (the interface is a tangential discontinuity). However, with respect to the spacecraft the interface is moving with a constant velocity along the interface normal direction. So in the spacecraft reference system the solar wind velocity along the interface normal direction has an average value different from zero. On the basis of this result we can solve this velocity coplanarity problem, If we find two normal vectors with respect to the maximum variance direction, such as along one of the vectors, the solar wind velocity has an average value about zero, and along the other vector the solar wind velocity has an average value different from zero (fisi ). The method of finding this reference system can be summarized as follows: (1) We select an interval of solar wind plasma data of a few hours before and after the interface of at least 8 hours (eight data points) which by inspection contains a well-defined discontinuity and a strong shear flow (see, e.g., interval B in Figure 1). (2) We apply the principal axis analysis to the selected interval of solar wind velocity to obtain the three eigenvectors ( lmex, flint, and timin). (3) We transform the interval of ve- Figure 2 shows how we defined the orientation of the locity data (V) to the principal axis reference system stream interfaces. A stream interface is tilted with reto obtain averages values along the three eigenvectors spect to the ecliptic plane. The black arrow represents (Vmax, Vint, and Vmin). At this point, we are looking the normal direction of the stream interface fisi. The for two normal directions orthogonal to timex: one di- latitudinal inclination of the stream interface is defined rection should have a zero average value of solar wind by the angle, 98i, between fisi and the ecliptic plane. velocity (the noncoplanarity component), and the other direction should have a nonzero average value of solar stream interf e wind velocity (fisi ). This can be done by a simple rotation. (4) We rotate the principal axis reference system Parker spiral along the maximum variance direction lma x through projection an angle a - arctan(vint/vmin). After this rotation, one of the normal components to timex is going to be Sun along the projection of the average solar wind velocity (fisi ), and the other normal component is going to have a zero average value of solar wind velocity (noncoplanarity component). These three normal vectors define a reference system that we could call "the interface reference system as seen from the spacecraft." As commented on before, not all the interaction re- gions contain a well-defined interface. In fact, we could not analyze most of the events. We only applied the technique to those events with a well-defined discontinuity in plasma parameters and a strong shear flow. We did not analyze interaction regions with data gaps or those Containing large-amplitude temporal variations. To make sure that the maximum variance direction was well-defined, we used as criteria that the maximum eigenvalue was at least 5 times larger than the other two, and we varied the length of the interval to make sure that the result was stable (i.e., that did not change significantly using three different inter- vals around the stream interface). Although we were careful selecting the events, the intervals, and testing the results, there is always the problem of the uncer- tainty associated with the results. We have combined data from four different instruments that have different resolution in their measurements of solar wind veloc- ity. This suggests that depending on the performance of the different instruments and the temporal variations in the solar wind, the uncertainty associated with each event can vary. This should be kept in mind. As a first guess, the uncertainty in the orientation of the stream interfaces should vary between 1øto 5ødepending on the event. However, this does not change the qualitative results presented in the section 2.2. Voyager I detected 17 interaction regions, of which we were able to analyze only 4. Voyager 2 detected 13 interaction regions, of which we were able to analyze 5. Ulysses detected 17 interaction regions, of which were able to analyze 9. Pioneer 11 detected 50 interaction regions, of which we were able to analyze 20. By visual inspection of solar wind plasma and magnetic field profiles the interfaces at Pioneer 11 observations are better defined than most of the interfaces at Voyager and Ulysses observations Results: Orientation of Stream Interfaces ecliptic plane interface longitudinal orientation wilh respect to he Parker spiral Figure 2. Three-dimensional orientation of a stream interface. The stream interface is intersecting the ecliptic plane. The interface latitudinal inclination si is the angle between the interface normal fisi, and the ecliptic plane. When si 0 ø the front is pointing north; si 0 ø implies that the interaction region is orthogonal to the ecliptic plane; and si ( 0 ø means that the front is pointing south. The angle A b denotes the angular difference between the in-ecliptic Parker spiral angle and the in-ecliptic longitudinal orientation of the stream interface. When A b 0 ø, the interface is oriented along the Parker spiral; when A b ( 0 ø, the interface is more azi nuthal than the Parker spiral; and when A b 0 ø, the interface is more radial than the Parker spiral.

5 ._ GONZ,iLEZ-ESPARZA AND SMITH: SOLAR CYCLE AND INTERACTION REGIONS AU 2 AU 3 AU 4 AU 5 AU 6 AU a, 30 iiiii!!il.voyager 1 3 ::ii::::ii:: 4 imences o -=. '0.:..:!::!: '... : : : :.: : : :,: : : : ', i i C ) 30 u ysses interfaces 20 ø ' 0 71 ' ' '-' i, i, i, i, F d ) 30 ' ' Pioneer i::i::::!i:::: 20 interfaces 20 -o 0.iiii:i... i ::,,,... :ii!ii;i?, : ;.! ;::... i i i i! i... :: : : =' -20 ::: ::: ::': : i I i, i, ' 1 AU 2 AU 3 AU 4 AU 5 AU 6 AU, o Heliocentric Distance Figure 3. Latitudinal orientation of the stream interfaces 98i against heliocentric distance: (a) Voyager 1, (b) Voyager 2, (c) Ulysses, and (d) Pioneer 11 (6 8i is the angle defined in Figure 2). We did not find a clear tendency of 6 si to evolve with the heliocentric distance from 1 to 5 AU. The fluctuations in 6 si are more likely associated with temporal variations. The longitudinal orientation of the stream interface was defined by an angle A b = bp - b i. Where b i is the in-ecliptic longitudinal orientation of the stream interface and es is the in-ecliptic Parker spiral angle at the site of the interface: Figure 4 shows the esi distribution histograms as detected by the three missions. There is a clear difference in the latitudinal tilts of the interaction regions detected by Ulysses and the ones detected by Voyager i and 2 and Pioneer 11. Ninety percent of the interfaces de- tected by Ulysses had an absolute latitudinal tilt lesil q ps(si) -- arctan (-ft R(si) / Vr(si)) less than 5ø; while 55% of Voyager and Pioneer 11 interfaces had a I t sil larger than 10 ø. This result might be (where fto is the Sun's equatorial angular speed, R(si) associated with the solar cycle. The in-ecliptic Ulysses is the heliocentric distance, and Vr(si) is the average observations were made just after the maximum of cysolar wind radial velocity at the interface). cle 22, when we do not expect polar coronal holes but Figure 3 shows interface latitudinal inclinations esi small holes at midlatitudes IHundhausen et al., 1981]. against heliocentric distance as deduced from the four Interaction regions caused by fast streams from equaspacecraft. The Voyager (Figures 3a and 3b) and Pio- torial coronal holes are not expected to have significant neer 11 (Figure 3d) resultshow variablesi with many tilts close to the ecliptic [SisCoe, 1976; Pizzo, 1982]. On interfaces with significant latitudinal tilts. However, the the other hand, Pioneer 11 interaction regions are ex- Ulysses result show small e si, suggesting that he inter- pected to have inclinations produced by the tilted solar action regions were nearly perpendicular to the ecliptic geometry commented on in section 1.2. plane. We should keep in mind that the four Spacecraft Do the interfaces follow a Parker spiral? Figure 5 detected unstable patterns of solar wind streams (Fig- shows the distribution histograms of the angle differures 2 and 8 in paper 1). The variations in 6 si detected ence A b. Statistically, Voyager and Ulysses results are by Voyager I and 2 and Pioneer 11 are more likely as- very similar: in both cases about 67% (6 out of 9) of sociated with these changes of solar wind streams than their interfaces were more azimuthal than the Parker with heliocentric effects (Ulysses' out-of-ecliptic results spiral (A b_<-10ø), and about 33% (3 out of 9) of their show similar fluctuations in Osi [Riley et al., 1996, Fig- interfaces were closely oriented along the Parker spiral ure 5]). (IA bl < 10ø). On the other hand, Pioneer 11 results

6 9786 GONZ LEZ-ESPARZA AND SMITH: SOLAR CYCLE AND INTERACTION REGIONS -40 ø -30 ø -20 ø - 10 ø 0 ø 1 0 ø 2 0 ø 3 0 ø a)50.,.,.,.,.,.,., 9 interfaces 40 Voyagers 1 and lo 4 0 ø 50 4o 3o[.5' lo b)5o. 4 ß 3 Ulysses 9 interfaces 50 4oo TM 30_xm. 20õ 10 o I... 4 Pioneer Interfaces o o 1 '40ø': 0ø ' '20 ø -10øTheta0øsi ø 20 ø 3'0 ø 5O 30_=. 2o lo 4 o Figure 4. Histograms of the latitudinal orientations, 8si, of the stream interfaces detected by the three missions. The three distribution show clear differences that might be associated with the solar cycle. The low si at Ulysses might be related to interaction regions caused by equatorial coronal holes. The high si at Pioneer 11 might be associated with the tilt solar geometry at the descending phase of the cycle. show a more diverse distribution: 50% (10 out of 20) of deviations from the Parker spiral. Some of these reits interfaces were closely oriented along the Parker spi- sults can be associated with variations of the physical ral (IA bl < 10ø); 20% (4 out of 20) of its interfaces were properties of the solar wind streams causing interaction more azimuthal than the Parker spiral (A½ <_ -10ø); regions that we will discuss in section 3.1. We did not and 30% (6 out of 20) of its interfaces were more radial find a clear tendency in A½ as a function of heliocentric than the Parker spiral (A½ >_ 10ø). The results from distance in any of the four spacecraft. As pointed out, the three missions show some interfaces with significant in connection with Figure 3, the longitudinal variations.30 ø a) ø - 10 ø 0 ø i -, ß, ß more warped spiral oriented 1 0 ø 2 0 ø,., more radial Voyager 1/2 9 Inter]aces 3 0 ø ' 3o 2o = lo b)so ( 1,, J!. more warped spiral odented more radial UlySSeS interfaces 5O 30 _ 2o. o c) 50. _q 2.30 ø Pioneer 11 more warped splrel odented more radial 20 interfaces t 34 0 =_. '20ø ' 10ø Deltaø hi si ß -...:.:.:...,..: : i! 20 ø 20 ø 30 ø Figure 5. Distribution histograms of the in-ecliptic longitudinal orientation of the stream interfaces with respect to the in-ecliptic Parker spiral angle, A½, as detected by the three missions (A½ is the angle defined in Figure 2). Voyager and Ulysses results are statistically similar. Pioneer 11 shows a more diverse distribution. Some stream interfaces had significant deflections with respect to the Parker spiral orientation. '"'

7 GONZ/kLEZ-ESPARZA AND SMITH: SOLAR CYCLE AND INTERACTION REGIONS 9787 A b seem to be a temporal effect rather than reflecting this case), the solar wind dynamic pressure is about heliocentric evolution x 10-9 Pa. With the same density number and a We could not analyze many interaction regions with proton temperature of Tp = 1.2 x 105 K the solar wind ill-defined stream interfaces. In some cases, these in- thermal pressure is about 1.66 x 10 -ll Pa. Finally, takteraction regions contained large temporal fluctuations. ing the magnetic field magnitude as B = 5 nt, the solar These temporal fluctuations might be related to differ- wind magnetic pressure is about 9.95 x Pa. The ent causes: (1) the solar wind streams causing some of magnetic and thermal pressure have similar orders of these interaction region might had had unstable proper- magnitude. However, the dynamic pressure is about 2 ties (e.g., variations in density and/or large amplitude orders of magnitude larger than the thermal and mag- MHD waves), (2) the expansion of the reverse corotat- netic pressures for typical parameters of the solar wind. ing shock in the rarefaction region trailing the interac- Thus the dynamic pressures of the fast and slow winds tion region; or (3) the "collision" of a transient event play a predominant role in the formation and evolution (ejecta) with the interaction region. of interaction regions. Section 3 studies a different aspect of the interac- How significant are the variations of solar streams tion regions detected by the three missions. The di- causing interaction regions? We compared the bulk verse characteristics of the stream interfaces suggest speeds and pressures (dynamic, thermal, and magnetic) that there are significant variations in the properties of of the fast and slow streams associated with every interthe solar wind streams causing the interaction regions. action region. Note that the plasma inside the interaction region has already been modified by the corotating 3. Variability of Solar Wind Streams shocks bounding the interaction region. So in order to study the streams and the pressure ratios associated with the interaction regions, we averaged the "uncompressed" slow and fast solar wind just before and after each interaction region using the following procedure: (1) we selected an interval of solar wind data of a few hours (4-7) just before and after the interaction region (see, e.g., intervals A and C in Figure 1); In this section we study a different aspect of the complementary set of in-ecliptic interaction regions detected by the three missions from I to 5 AU. We compare some properties of the fast and the slow solar wind streams associated with the interaction regions. If, as we have shown, interaction regions can have very different orientations, different widths at the same heliographic location, different plasma and magnetic field profiles, etc., then the streams causing the interaction regions should also have different characteristics. Different characteristics of fast and slow streams produce different dynamics. Interaction regions arise when fast solar wind overtakes slow wind. The collision between a fast and slow stream implies an interchange of momentum. In principle, there should be a point where an equilibrium between the two streams is reached and the sum of the dynamic pressure, plus the thermal pressure plus the magnetic pressure should be constant across the normal direction of the stream interface. In this approximation, assuming that we can apply the ideal gas law to the solar wind and that protons and electrons have broadly the same temperature (which might not be valid in some cases) and that we can neglecthe contribution by alpha particles and pick-up ions, [mpnv n + 2NKTp + B2/(2/ o)] =0 (1) (where mp is the proton mass, N is the proton density number, Va is the solar wind velocity normal to the interface (as observed from the spacecraft), K is the Boltzman's constant (= 1.38 x J/K), Tp is the proton temperature, and B is the magnetic field magnitude). It is interesting to compare the orders of magnitude of the three parameters, given typical solar wind values at 1 AU. Taking the proton density number as N = 5 cm -s and the solar wind bulk velocity as Vr km/s (which we will use as Va in (2) we averaged the bulk speed and pressures of the two streams; (3) we obtained the three pressure ratios, between fast and slow wind, associated with every interaction region: dynamic rdyn. -- NfV /NfV?, thermal rther. = NfTpf/NsTp., and magnetic rmag. = B /Bs (where the subscripts f and s refer to parameters of the uncompressed fast and slow solar wind, respectively). As we will show, there are continual variations in the characteristics of the fast and slow streams associated with interaction regions. It is important to point out that the duration of the interaction regions in the data series is between I and 5 days. Then, in many events, the uncompressed slow and fast streams that we used in the study do not necessarily represent the characteristics of the slow and fast streams that actually caused the interaction region. The study of an event, in particular, should take into account this point. However, on the whole, the statistical results presented in this section, we believe, are qualitatively correct. 3.1 Results: Pressure Ratios and Bulk Speeds Table 1 presents the averages of bulk speeds and pressure ratios associated with interaction regions as detected by the four spacecraft. The second column shows the number of interaction regions analyzed. Unlike the study of stream interfaces, where we were only able to determine the geometry of a minority of events, in this analysis we were able to study most of the events (except the ones with data gaps). The third column shows that the bulk speed average of the fast streams behind the interaction regions was much higher at Pi-

8 9788 GONZ/ LEZ-ESPARZA AND SMITH' SOLAR CYCLE AND INTERACTION REGIONS Table 1. Comparison of Physical Parameters of the Uncompressed Fast and Slow Solar Wind Streams Associated With Interaction Regions as Observed by the Four Spacecraf Spacecraf Events VF SS rdyn. rther. rmag. Voyager I km/s 371 km/s 1.6 :E :E :E 2.0 Voyager km/s 342 km/s 0.9 :E :E :E 1.6 Ulysses km/s 412 km/s 1.1 :E :E :E 1.4 Pioneer km/s 386 km/s Events, number of interaction regions analyzed for each spacecraft; rr, average of bulk speed of the fast streams behind the interaction regions; V$, average of bulk speed of the slow streams preceding the interaction regions; rdyn., average and standar deviation of the ratios of dynamic pressures between fast and slow streams; rther., average and standardeviation of the ratios of thermal pressures; rmag., average and standard deviation of the ratios of magnetic pressures. 1 AU 2 AU 3 AU 4 AU 5 AU 6 AU InteractiOn 400 oo oneer 11 and Ulysses than at Voyager I and 2. The Voyagers fourth column shows that the bulk speed average of fast wind the slow streams preceding the interaction regions was higher at Ulysses and lower at Voyager 2. These results t are related to the changes in the distribution of solar,,oww,n wind bulk speeds through the solar cycle that were dishighe; dynamic; pressur;... Voyagers cussed in paper 1. The fifth column presents the avin fast wind 25 interaction regions erage and standard deviation of the ratios of dynamic pressures rdyn.. Note the large standard deviation asso- higher dynamic pressure in slo wind! I, i, I i, higher thermal pressure ' ' ' Voyagers ciated with every average. In many interaction regions the larger proton density of the slow wind caused the dynamic pressure of the slow stream to be higher than the dynamic pressure of the fast stream. Ulysses and fast wind 25 interaction :...'.: : regions Pioneer 11 rdyn. are very similar, but Voyager I and 2 differ significantly. In paper I we pointed out that the observations of large-scale dynamics by the two Voyhigher thermal pressure In slow wind ager spacecraf vary in many aspects (e.g., on the aver- I, i, i, I- age on the whole trajectory (1 to 5 AU) Voyager I dehighe; magnetl; pressure In fast wind Voyagers - tected about 1.2 interaction regions per solar rotation 25 interaction (27 days), while Voyager 2 detected about 0.8 interac-.:' "' ' i i... i?": :; '"... :::::..:::::,. tion regions per solar rotation). Voyager 2 observed the lowest average values of solar wind bulk speeds assohigher magnetic pressure in slow wind i, i, i, I : :, ciated with interaction regions, but the average bulk 2 AU 3 AU 4 AU 5 AU 6 AU speed along the whole trajectory was higher (see Table 4 in paper 1). It is possible that these differences Figure 6. Physical parameters of the uncompressed between Voyager I and 2 were related to the data gap fast and slow streams associated with the interaction in Voyager 1 and/or transient events (e.g., ejecta) that regions detected by Voyager i and 2 (ascending phase cycle 21) plotted versus heliocentric distance. The avwere detected predominantly only by one spacecraft; erages of slow and fast solar wind streams were taken however, this point requires further study. The sixth from intervals just before and after each interaction re- column presents the average and standard deviation of gion (see, e.g., the intervals A and C in the example of the ratios of thermal pressures. In this case the average Figure 1). (a) The bulk speeds of the fast (dark bars) values obtained by Voyager I and 2 were very similar and slow streams (gray bars). The pressure ratios be- and higher than the values obtained by Pioneer 11 and tween the fast and slow streams (b) dynamic rayn., (c) Ulysses. Finally, the seventh column presents the avthermal rther., and (d) magnetic rmag.. When the pressure of the slow stream was higher than the pressure of erage and standard deviation of the ratios of magnetic the fast stream (r < 1), this ratio had been redefined pressures where again the average values obtained by to be r equals negative (pressure slow wind/ pressure Voyager I and 2 were higher than the values obtained fast wind) to facilitate visual comparison. The plots by Ulysses and Pioneer 11. show continual variations in the properties of the fast We examined the variations of the physical paramand slow winds associated with interaction regions. eters associated with the interaction regions in more

9 GONZ/t. LEZ-ESPARZA AND SMITH' SOLAR CYCLE AND INTERACTION REGIONS AU 1 AU 2 AU 3 AU 4 AU 5 AU. Figure 7 presents the parameters of the interaction 700 regions 14 Intelaction ' ' ' ' ' ' ' Ulysses regions detected by Ulysses. The plots show the variations in the patterns of solar wind bulk speeds (see 6001 fast wind.1 I.! i! ilj! also Figure 8c in paper 1) and the pressure ratios. In Ulysses observations we studied 14 interaction regions i: For seven of these events (50%), rdyn. and rther. were b) less than I (always rdyn. 1 and rther. < 1 in a one-to- ' ' ' ' ' ' U ysses ' ' higher dynamic pressure -- In fast wind 14 Interaction ' one correspondence), and for nine of thesevents (64%), regions rmag. WaS less than 1. In this case we did not find a clear,?..:¾-iiii:i... correlation between rmag. and the other two pressure rahigher In slow dynamic wind tios rdyn. and rther.. i i pressure, I i I i :i:l i Figure 8 shows the parameters of the interaction re- ' 10 -:_:J higher In fast wind thermal pressure gions detected by Pioneer 11. Although in many as-,,ss,s 14 Interaction pects Pioneer 11 and Voyager observations were very different, in this case the pressure ratios show similar Figures 7 and 8c in paper 1. From about 2.3 to 3.2 AU, = In slow wind n- '15 1,, Pioneer 11 detected a change in the pattern of solar I higher thermal pressure variations. It is interesting to compare Figure 8 with ß [. 1 0 higher ß magnetic ' pressure ' ' Ulysses wind speeds and magnetic sectors; this period coincides In fast wind 14 Interaction with times in Figure 7 when the ratios of dynamic pres- 5 regions sures were less than 1. It appears that the variations... "':'"' in the ratios of pressures were associated with temporal - 10 higher magnetic pressure in slow wind fluctuations in the pattern of solar wind streams. In the I I i, ii i I, 1 AU 2 AU 3 AU 4 AU 5 AU 6 AU analysis of Pioneer 11 observations we studied 36 inter- l I Figure 7. Physical parameters of the uncompressed fast and slow streams associated with the interaction regions detected by Ulysses (postmaximum phase (cycle 22)). Format is the same as Figure 6. The variations in the pressure ratios is similar as in Figure 6. However, the patterns of bulk speeds are different. i a ) 38,nieraction' ' Pioneer 11 - i 600 II i!11 i:ii-!!l / fast wind ca ! slo wind detail. Figures 6, 7, and 8 show the values of these parameters against heliocentric distance as detected by the three missions. Figures 6a-Sa show the bulk speeds of the fast and slow streams, and Figures 6b-6d, 7b-7d, ure 6a shows that the bulk speeds of the fast streams associated with interaction regions were very variable (see also Figures 8a and 8b in paper 1). Figures 6b - 6d show clearly how the pressure ratios between fast and slow streams also vary continuously. In Voyager 1 observations we studied 15 interaction regions where about 27% had rdyn. < 1, about 47% had rther. < 1, and about 47% had rmag. 1. In Voyager 2 observations we studied 10 interaction regions where about 70% had rdyn. 1, about 30% had rther. < 1, and 47% Figure 8. Physical parameters of the uncompressed fast and slow streams associated with the interaction had rmag. 1. These remarkable differences between regions detected by Pioneer 11 (descending phase (cythe contemporary measurements by Voyager I and 2 at cle 20)). The format is the same as Figures 6 and 7. different hellographic locations suggesthat interaction There are similar variations in the pressure ratios as in regions during the ascending phase of the solar cycle Figures 6 and 7. This suggesthat these variations are have very irregular shapes and characteristics as a con- present at different phases of the solar cycle, not just sequence of inhomogeneousolar wind streams. close to the solar maximum. b) higher dynamic pressure in slowind Pioneer Interaction regions and 8b-8d show the plots of the three pressure ratios associated with every interaction region. For display 15 ' ' ' ' ' ' ' ' ' purposes, when the pressure value of the slow stream C ) higher thermal pressure P Jena _e r in fast wind :.i: 35 nteraction was higher than the pressure value of the fast stream 5 -? :ii regions.. (r < 1) we "redefined" the pressure ratio as r = -r -z. Figure 6 presents Voyager I and 2 observations. Fig- o o l:.i,,. i...*...,..:,:: thermal ::...*.....'... :,:..-...,...i ii pressure i',i' ii higher '= I I in slowind, I, I![[ :.!, I, I I" d) ] highe; magneti;: pressure in fast wind In slow wind higher magnetic pressure 1 AU 2 AU 3 AU 4 AU Pioneer Interaction regions I 5 AU 6 AU

10 GONZ LEZ-ESPARZA AND SMITH: SOLAR CYCLE AND INTERACTION REGIONS a) o% 20% i,,,! ß,, :,,, i,,, i,,,::j i:!:!:i:!: : : : : : iii!iiiiiii!ii!ii!i! :::::::::: ::::::::: Voyager/slow streams 25 interaction regions Voyager/fast streams % mean st.dv. 82 : 25 interaction 20% -o :::::::::::::::::::: 10% % - 20%- i ' i,,, I,,, i,, Ulysses/slow streams mean 414 st.dv intera.tion regions 30% - 20%- i,,, i,,, i,,,!,,, I,, i mean Ulysses/fast 578 streams L st.dv. 70! 14 i. nteration reg,ons I -?, i :'"', i,,, i,, Pioneer 11/slow streams 30% mean 386 st.dr interaction 20% regions 40%.. I ' ' ' I ' ' ' I ' ' ' I,, ß I ', t Pioneer 11/fast 30% mean 591 st.dv interaction 0.o-i ro'o streams e 0 10% ß, i.. i [km/s] [km/s] Figure 9. Bulk speed histograms of the uncompressed (a) slow and (b) fast solar wind streams associated with the interaction regions detected by the three missions. The averages of slow and fast solar wind streams were taken from intervals just before and after each interaction region (see, e.g., the intervals A and C in the example of Figure 1). The distribution of slow winds was very similar in Voyager and Pioneer 11 observations, but Ulysses detected faster slow winds. The distributions of fast winds were very different in the three missions. action regions. For 17 events (47%), rdyn. < 1. For 12 events (34%), rther. 1. For 15 events (47%), rmag. 1 (note that for some of the interaction regions, the data gaps in proton temperature and/or magnetic field did not allow us to estimate the three pressure ratios). Figure 9 shows the histogram distribution of the bulk speeds of the fast and slow streams associated with interaction regions as detected by the three missions. Figure 9a shows that the speed distribution of the slow winds was very similar at Voyager I and 2 and Pioneer 11. At Ulysses the slow winds were faster. Figure 9b shows that the distribution of fast winds during the three missions was very different. Voyager I and 2 detected the slower fast winds. Ulysses detected a predominantly fast stream around 650 km/s, and Pioneer 11 shows a diverse distribution covering from about 400 km/s to about 750 km/s. 4. Summary and Conclusions We have studied two different aspects of the interaction regions detected by three missions that traveled to Jupiter at different phases of the solar cycle. This combined set of 97 in-ecliptic interaction regions detected from I to 5 AU allow us to study heliocentric evolution of these events and solar cycle variations. A summary of the results presented in sections 2 and 3 is given next. From the analysis of the orientation of 38 stream interfaces we found that 1. The interaction regions detected by Voyager and Pioneer 11 had broad distributions of latitudinal incli- nations. In both missions about 55% of their interaction regions had an absolute latitudinal tilt IO il larger than 10 ø, and about 45% of their interaction regions had an absolute latitudinal tilt I sil less than 10 ø. On the other hand, the interaction regions detected by Ulysses had very low latitudinal inclinations; about 90% of them had an absolute latitudinal tilt 10sil 5 ø. 2. The distribution of longitudinal orientations of the interaction regions detected by Voyager and Ulysses were very similar. In both missions about 67% of their interaction regions were more azimuthal than the Parker spiral (A b_ -10ø), and 33% of their interaction regions were closely oriented along the Parker spiral (IA bl 10ø). Note that the two missions did not detect "more radial" (A b _ 10 ø) interaction regions. 3. The interaction regions detected by Pioneer 11 had a broad distribution of longitudinal orientations; about 50% of them were closely oriented along the Parker spiral 10ø), 20% of them were more

11 GONZ/ LEZ-ESPARZA AND SMITH: SOLAR CYCLE AND INTERACTION REGIONS 9791 azimuthal than the Parker spiral (A b <_ -10ø), and We found significant variations in the pressure ra- 30% of them were more radial than the Parker spiral tios (dynamic, thermal, and magnetic) between fast (A b _ 10 ø). This somewhat symmetric distribution and slow winds associated with interaction regions. In about half of the cases the slow wind transfers momenmay suggest that these interaction regions (declining: phase of the cycle) were oriented along the Parker spiral. tum to the fast wind. The wide range of pressure ratios and the broad distribution just represents local fluctu- may provide insight into differences in the geometry and ations. In paper 1 we showed that interaction regions can evolution of interaction regions. These differences also place upper limits on the assumption of periodicity that have different thickness at the same heliocentric dis- is implicit in many models of solar wind flow. tance (Figure 10 in paper 1). The results presented here show that interaction regions can also have different geometries (latitudinal and longitudinal orientations). Acknowledgments. We are grateful to V. J. Pizzo for This result is similar to the broad latitudinal orienta- tions of the corotating shocks detected by Ulysses at midlatitudes reported by Burton et al. [1996] and Riley plasma and magnetic field data for this study were obtained et al. [1996]. The diversity in the width and large-scale from the National Space Science Data Center at Goddard orientation of interaction regions seems to be related to the continual variations in the properties of the fast and slow solar wind streams. We did not find a clear tendency in the orientation of the stream interfaces to evolve with heliocentric distance (from I to 5 AU). The reason, we believe, is that the three missions detected several changes in the patterns of solar wind streams affecting the solar wind large-scale dynamics (see paper 1); so the variations in the orientations were more related to these temporal changes than to heliocentric effects. In the second part of the paper we analyzed some physical parameters of the uncompressed fast and slow streams associated with 75 interaction regions. The main results are as follows: 1. The three missions observed very different distributions of bulk speed for the fast streams behind the interaction regions. The fast streams at Voyager I and 2 were much slower than the fast streams observed at Ulysses and Pioneer The bulk speed distribution of the slow streams preceding the interaction regions were very similar at Voyager and Pioneer 11 observations, but at Ulysses observations the slow streams were faster. 3. In approximately half of the interaction regions that we analyzed, the difference in density between slow and fast streams was sufficiently large, so that the dynamic pressure of the uncompressed slow wind was higher than the dynamic pressure of the uncompressed fast wind. 4. The three missions detected variations in the ratios of thermal and magnetic pressures between the fast and slow streams associated with interaction regions. 5. The simultaneous measurements, at different heliographic locations, by Voyager I and 2 during their trans-jovian trajectory at the ascending phase of cycle 21 revealed significant differences in the dynamics of solar wind streams. This suggests that the patterns of solar winds may be highly inhomogenous during the ascending phase of the solar cycle and probably at other times as well. This would produce interaction regions with irregular shapes and characteristics. reading the manuscript. The very useful comments and suggestions by the two referees are highly appreciated. J.A.G.- E. is grateful to the National Research Council for financial support as a Resident Research Associated at JPL. The Space Flight Center. J. Wolfe is the principal investigator of the Pioneer 10 and 11 solar wind plasma experiments. H. Bridge and J. Belcher are the principal investigators of the Voyager I and 2 solar wind plasma experiments. N. Ness is the principal investigator of the Voyager I and 2 magnetic field experiments. D. McComas is the principal investigator of the Ulysses solar wind plasma experiment. A. Balogh is the principal investigator of the Ulysses magnetic field experiment. The research at Jet Propulsion Laboratory, California Institute of Technology, was under contract with the National Aeronautics and Space Administration. The Editor thanks P.R. Gazis and P. Riley for their assistance in evaluating this paper. References Balogh, A., J. A. Gonzdlez-Esparza, R. J. Forsyth, M. E. Burton, B. E. Goldstein, E. J. Smith and S. J. Bame, Interplanetary shock waves: Ulysses observations in and out of the ecliptic plane, Space Sci. Rev., 72, , Burlaga, L. F., Interplanetary stream interfaces, J. Geophys. Res., 79, , Burton, M., E. J. Smith, A. Balogh, R. J. Forsyth, and B. E. Goldstein, Ulysses out-of-ecliptic observations of interplanetary shocks, Astron. Astrophys. Suppl. Set., 316, , Gonzdlez-Esparza, J. A., and E. J. Smith, Solar cycle dependence of the solar wind dynamics: Pioneer, Voyager, and Ulysses from I to 5 au, J. Geophys. Res., 101, , Gonzdlez-Esparza, J. A., A. Balogh, R. J. Forsyth, M. Neugebauer, E. J. Smith, and J. L. Phillips, Interplanetary shock waves and large-scale structures: Ulysses' observations in and out of the ecliptic plane, J. Geophys. Res., 101, , Gosling, J. T., J. R. Asbridge, S. J. Bame, and W. C. Feldman, Solar wind speed variations: J. Geophys. Res., $1, , Gosling, J. T., J. R. Asbridge, S. J. Bame, and W. C. Feldman, Solar wind stream interfaces, J. Geophys. Res., 83, , Gosling, J. T., S. J. Bame, D. J. McComas, J. L. Phillips, V. J. Pizzo, B. E. Goldstein, and M. Neugebauer, Latitudinal variation of solar wind corotating stream interaction regions: Ulysses, Geophys. Res. Lett., 20, , Hoeksema, J. T., The relationship of the large-scale solar field to the interplanetary magnetic field: What will ulysses find?, in The Sun and the Heliosphere in Three-

12 9792 GONZ/ LEZ-ESPARZA AND SMITH: SOLAR CYCLE AND INTERACTION REGIONS Dimensions, edited by R. G. Marsden, pp , D. Reidel, Norwell, Mass., Hundhausen, A. J., R. T. Hansen, and S. F. Hansen, Coronal evolution during the sunspot cycle: Coronal holes observed with the Mauna Loa K-coronameters, J. Geophys. Res., 86, , Mihalov, J. D., A. Barnes, A. J. Hundhausen, and E. J. Smith, Solar wind and coronal structure near sunspot minimum: Pioneer and smm observations from , J. Geophys. Res., 95, , Pizzo, V. J., A three-dimensional model of corotating streams in the solar wind, J. Geophys. Res., 87, , Pizzo, V. J., The evolution of corotating stream fronts near the ecliptic plane in the inner solar system, 2, Threedimensional tilted-dipole fronts, J. Geophys. Res., 96, , Pizzo, V J., Global, quasi-steady dynamics of the distant solar wind, 1, Origin of north-south flows in the outer heliosphere, J. Geophys. Res., 99, , Pizzo, V. J., and J. T. Gosling, 3-D simulations of high latitude interaction regions: Comparison with Ulysses observations, Geophys. Res. Left., 21, 2063, Riley, P., J. T. Gosling, L. A. Weiss, and V. J. Pizzo, The tilts of corotating interaction regions at midheliographic latitudes, J. Geophys. Res., 101, in press, Siscoe, G. L., Structure and orientation of solar-wind interaction fronts: Pioneer 6, J. Geophys. Res., 77, 27-34, Siscoe, G. L., Three-dimensional aspects of interplanetary shock waves, J. Geophys. Res., 81, , Siscoe, G. L., B. Goldstein, and A. J. Lazarus, An east-west asymmetry in the solar wind velocity, J. Geophys. Res., 74(, , Zhao, X., and A. J. Hundhausen, OrganiZation of solar wind plasma properties in a tilted, heliomagnetic system, J. Geophys. Res., 86, , J. A. Gon z lez-esparza, Instituto de Geoffsica, UNAM, Ciudad Universitaria, Coyoacdn, M xico D.F , MEX- ICO ( americo@tonatiuh.igeofcu.unam.mx). E.J. Smith, Jet Propulsion Laboratory, MS , 4800 Oak Grove Drive, Pasadena, CA ( esmith@jplsp.nasa.gov) (Received November 13, 1996; revised January 22, 1997; accepted February 17, 1997.)