JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 97, NO. All, PAGES 16,877-16,885, NOVEMBER 1, 1992 Digisonde Measurements of Polar Cap Convection for Northward Interplanetary Magnetic Field PAUL S. CANNON Defence Research Agency, Farnborough, Hampshire, United Kingdom GEOFFREY CROWLEY AND BODe W. REINISCH Center for Atmospheric Research, University of Massachusetts. Lowell, Massachusetts JURGEN BUCHAU Phillips Laboratory, Hanscorn Air ForceBase, Bedford, Massachusetts Controversy still exists regarding even the average convection pattern when the interplanetary magnetic field (IMF) has a northward component. Using two years of convection data from a Digisonde located at Qaanaaq only 3ø from the corrected geomagnetic pole we have examined the diurnalconvection flow direction variation in the central polarcap whenthe IMF is particularly stable. We find that whenBz is positive, and when By positive and By negative data are treated independently, each exhibits a clear diurnal pattern. Thepauerns are most nearly consistent witha multicell convection model, e.g.Petetara et al. (1984); thereare, however, two anomalies. Our synthesized polarcap convection patterns exhibita polar cap cell centered on 10 corrected geomagnetic localtime (CGLT) when B > 1 nT and13 CGLT when B < -1 nT in Y Y contrast to 06 and18CGLT predicted by themulticell models. Furthermore, in contrast to thesimple multicell models the convection flow patterns for opposite B polarities arenot simple mirrorimages of each other. Y When By< -1 nT the convection is directed across the central polar cap toward 02 CGLT formuch of the day but •hen By > 1nT the flow is tangential to the Qaanaaq geomagnetic latitude for much ofthe day. 1. INTRODUCTION Duringperiods whenthe interplanetary magnetic field (IMF) z component is negative (southward)it is widely accepted that the open magnetospheric model of Dungey [1961] is appropriate. This results in a two-cell ionospheric convection model such asthat proposed by Heppner and Maynard [1987]. The IMF y component, and to lesser extent the IMF x component, are also recognized to play important roles in controlling this convection pattern [e.g. Heppner and Maynard, 1987; Cannonet al., 1991; Cowley et al., 1991]. When the IMF z componentis positive (northward) the convection pattern is thought to be more complex thanthat for the negative (southward) configuration. Considerable controversy exists, however, regarding even the averageor statistical pattern. Maezawa [1976], using ground-based magnetometers, recognized that the simple two-cell pattern is not appropriate to IMF northward conditions, and he deduced that sunward flow should occur in the central polar cap. Magnetospheric satellite electricfield measurements usingS3- 2 andMagsat [Burkeet al., 1979; lijima et al., 1984] confirmed the presence of sunwardconvection,and Burke et al. [1979] suggested a four-cell model of convection, although this pattern could only be clearly inferredon the dayside. Support for the multicellmodel was provided by Potemraet al. [1984] who suggested that a three-cell pattern could occur for Copyright 1992by theAmerican Geophysical Union. Papernumber 92JA01190. 0148-0227/92/92JA-01190505.00 northward Bz and furthermore, that the patternwas dependent upon the IMF By component. WhenBy.-0 four cells were proposed, much the same as those in the Burke et al. [1979] model, butthis reduced to three cells when IByl wasgreater than zero. In the northern hemisphere asBybecomes more positive the polar cap dawn cell expandsand the dusk cell shrinks. When By is particularly strong the expanded cell gives the appearance of a single convectioncell in the polar cap region and the whole pattern appearsto have three cells. When By< 0 the opposite occurs and thedusk cellexpands while the dawn cell shrinks. The three- or four-cell modelwaschallenged by Heppnerand Maynard [1987] who interpreted sunward convection, measured by the DE 2 satellite in terms of a distorted two-cell pattern. The distortion proposed was that of a translational stretching of the evening cell focus toward and beyond noon, accompanied by a rotational twist of the stream lines around the foci. Progressively greater distortion was proposed as Bz increased. Recently,Zhu and Kan [1990] have had some theoretical success in rationalizing the distorted two-cell and multicell models. In their theoretical approach they propose that a number of the differences may be accounted for by anisotropic conductance in the ionosphere. Zhu and Kan [1990] startwith a four-cell magnetospheric model and show that an asymptotic ionosphericflow pattern develops (after 25 min) which is similar to the two-cell Heppner and Maynard [1987] model except that there are additional convectioncells in the polar cap. The purpose of this paper is to report on Digisonde measurements of the convection flow directions in the central polar cap whenBz is positiveand to relate these measurements 16,877