Measuring solar wind outflow velocities and understanding the structure of the source region of the high speed solar wind is one of the outstanding problems in solar physics. Observations of outflow velocities in coronal holes (regions of open coronal magnetic field) have recently been obtained with the Solar and Heliospheric Observatory (SOHO) spacecraft. Velocity maps of Ne7+ from the bright resonance line Ne VIII 770 Å, formed at the base of the corona, show a relationship between outflow velocity and chromospheric magnetic network structure, suggesting that the solar wind is rooted at its base to this structure, emanating from localized regions along boundaries and boundary intersections of magnetic network cells. This apparent relationship to the chromospheric magnetic network, as well as the relatively large outflow velocity signatures at the intersections of network boundaries at mid-latitudes are a first step in understanding better the complex structure and dynamics at the base of the corona and the source region of the solar wind.

Here we present observations obtained with the Solar Ultraviolet Measurements of Emitted Radiation (SUMER) instrument on SOHO. SUMER is one of twelve instruments on the ESA/NASA SOHO spacecraft, which was launched on 2 December 1995. SOHO is in a halo orbit around the Lagrange point L1, about 1.5 million kilometers sunward of Earth with a constant view of the Sun. The SUMER instrument (7) is a stigmatic ultraviolet (500-1610 Å or 50-161 nm) grating spectrograph with an off-axis silicon carbide (SiC-CVD) telescope mirror. The telescope mirror can be scanned along two axes to provide pointing and raster images using one of several possible spectrometer entrance slits (typical angular size of slit is 1.0 x 300 arc sec2). Overlapping first and second order stigmatic spectra are imaged onto a 1024 x 360 pixel microchannel plate detector (8) with a dispersion of 42-45 mÅ/pixel (4.2-4.5 x 10-3 nm/pixel) in first order and 21-22.5 mÅ/pixel in second order. A grating scan mechanism permits selection of the individual bandpass (40 Å in first order, and 20 Å in second order) which is imaged onto the detector at a given time. A more complete description of the SUMER instrument can be found in (7, 9, 10).

Two sets of observations obtained with the SUMER spectrometer provide high resolution two-dimensional spectroheliograms of a 9 x 5 arc min2 region in a mid-latitude quiet region (22 September 1996 00:40-08:15 UT) and an open magnetic field region of the north polar coronal hole (21 September 1996 00:15-07:30 UT) (Fig. 1). The observational sequences, spectral bandpass, and location of the spectral bandpass on the detector were identical for both sets of observations to minimize any systematic observational or instrumental effects when comparing the two sets of observations. Each spectroheliogram (11) contains full spectral information permitting the construction of line-of-sight velocity maps as well as intensity images in emission lines formed over a wide range of temperatures and heights in the solar atmosphere, from the chromosphere (Si II 1533 Å, formed at 104 K), through the chromosphere-corona transition region (C IV 1548 Å, formed at 105 K) to the base of the corona (Ne VIII 770 Å, formed at 105.8 K).

The spectral bandpass selected for these studies (Fig. 2a) extends from 1530-1550 Å in first order and 765-775 Å in second order. The overlapping spectral orders permits us to observe many more spectral lines in a given spectral window on the detector and these lines can be used as wavelength and dispersion references to determine the precise wavelength scale of the instrument. The disadvantage is the increased likelihood of weak blends in the overlapping spectral orders which may complicate the analysis by requiring multi-peak Gaussian fitting routines. However, the first order SUMER instrument efficiency (near 1540 Å) is much lower than the second order efficiency (at 770 Å), thus minimizing this complication. To assess the effect of the numerous weak first order Si I lines near the second order Ne VIII 770 Å coronal line, we used a first order High Resolution Telescope and Spectrograph (HRTS) spectrum (12) from a chromospheric bright point, which produces strong Si I intensities, and scaled it to the SUMER spectrum using common unblended Si I lines in the spectral bandpass (Fig. 2b). We fit the original SUMER profile to a Gaussian distribution, subtracted the scaled first order HRTS spectrum, and then fit once again the resulting Ne VIII line profile to a Gaussian to measure any possible line shifts due to the Si I blends. The result showed a redshift of less than 1 mÅ or 0.03 pixels in the Gaussian line center of the unblended profile with respect to the original SUMER profile, which is within the uncertainty of the measurement (±0.1-0.2 pixels). So for these observations, the overlapping first order Si I blends do not appear to significantly effect the Gaussian fit line centers or wavelengths.

The data were corrected for nonuniformities in the SUMER detector with a "flat field" obtained on 23 September 1996 (to correct for pixel-to-pixel sensitivity variations) and "de-stretched" (to correct for geometric and thermal distortion) by mapping the data to a grid derived from systematic observations of chromospheric lines. The data were then binned along the slit (y-axis) to increase the signal to noise, and provide approximately square spatial resolution pixel elements, with an effective pixel size of 3 x 3 arc seconds. The emission lines from the spectrum of each spatial resolution element were fit to a Gaussian profile and linear background using a maximum likelihood technique (13, 14) which weighs each data point in the profile according to Poisson statistics (effectively giving more weight to the line center and less weight to the wings in the fitting algorithm). The resulting uncertainties in the Ne VIII 770 Å, Si II 1533 Å and C IV 1548 Å line profile parameters are relatively small, with typical uncertainties in the line center position on the order of ±0.1-0.2 pixels (1 s).

The wavelength of the Gaussian center of each Ne VIII 770 Å profile was measured with respect to the Gaussian center position of the overlapping Si II 1533.432 Å profile for that spatial resolution element. Previous rocket observations (15, 16) of the wavelength of the chromospheric Si II 1533 Å line with respect to an absolute in-flight wavelength calibration lamp have shown that the wavelength of the solar line agrees with the laboratory rest wavelength of 1533.432 Å to within ±1-2 km/s, making this solar chromospheric line a reliable wavelength and velocity reference. The spectrometer wavelength dispersion scale relating the observed wavelength of the Ne VIII line to the Si II line was determined by fitting the Gaussian center position of eight weak nearby first order Si I lines to a polynomial fit. Each of the wavelengths of these Si I lines agreed with this dispersion curve to within 1-2 km/s.

The spatially averaged Ne VIII wavelength for the observations discussed here was 770.411±0.005 Å for the polar region, and 770.419±0.005 Å for the mid-latitude region (Fig. 3). Using the laboratory rest wavelengths published by Bockasten et al. (17) of 770.409±0.005 Å, we find that the observations from the mid-latitude region are systematically redshifted (22) by 4-5±2 km/s. The observations from the polar region tend to be less redshifted, approaching zero redshift at the edge of the polar crown, and then slightly blueshifted (on average) inside the polar coronal hole, consistent with earlier observations (23). Using the more recent solar "rest" wavelengths for Ne VIII determined by Dammasch et al. (19) of 770.428±0.003 Å, we find that the Ne VIII Doppler velocities are, on average, at rest or slightly blueshifted throughout the mid-latitude "quiet Sun" region, with the line-of-sight velocities inside the coronal hole approaching 5-6 km/s toward the blue, suggesting significant outflowing material. In both cases, we see that the observations from the polar region are significantly blueshifted with respect to those from the mid-latitude region, indicative of the coronal outflows associated with the fast solar wind.

However, superposing the same chromospheric network boundaries on the coronal Ne VIII velocity images (Fig. 6) reveals a much stronger correlation. In this case, the coronal velocity field is correlated with the chromospheric magnetic network structure below. Blueshifts, or outflows, tend to occur predominantly on network boundaries, with the largest blueshifts (outflows) occuring at the intersection of network boundaries.

In the polar coronal hole observations (Fig. 6b), the blueshifts appear sharp (bright) along the network boundaries and more diffuse outside of the boundaries, which is consistent with a "funnel-like" expansion of the open magnetic field lines originating along the network boundaries (31-34). It can also be seen that blueshifts do still occur in the mid-latitude region (Fig. 6a), at the intersection of network boundaries, but that the majority of the area, particularly in the cell interiors, is either at rest, or redshifted, depending on the Ne VIII rest wavelength reference.

The magnitudes of the line-of-sight Doppler shifts along the network boundaries in the polar coronal hole between 800-875 arc seconds north of the equator (56 deg. to 65 deg. N solar latitude) are 3-6 km/s (blueshift). These line-of-sight Doppler shifts correspond to radial outflow velocities of 5-12 km/s depending on the location of the observed Doppler shift. Velocities at the intersection of the network boundaries can be greater than the velocities along the network boundary itself, with radial velocities as high as 10-20 km/s.

The relationship between outflow velocity and network structure discussed here is in contrast with recent observations of asymmetries in the wings of the He I 1083 nm absorption line in a polar coronal hole (37). Their results suggest outflows of about 8 km/s found predominantly in the center of supergranular cells, and which are not associated with the bright network structures. However, the formation process of the He I 1083 nm line is complicated (38), making it difficult to directly associate wing asymmetries with outflow velocities. Resolution of this discrepancy may lie in a better understanding of the formation of the He I line.

Although our observations suggest that the outflows occur predominantly on network boundaries and at the intesections of network boundaries, there is no clear correlation between Ne7+ velocity and Ne VIII intensity (Fig. 5-6). The largest Ne7+ velocities do not correspond to the brightest structures. One might speculate that the energy funneled up through the network goes into either heating local closed loop structures, such as Extreme Ultraviolet (EUV) and X-ray bright points, or accelerating material along open magnetic field regions at the boundaries and intesections of the network structure, but not both at the same time. Whether the energy goes into heating the plasma or accelerating it depends critically on the local magnetic field topology.

These strong blueshifts seen along the network boundaries suggest that these regions may be the only place where the magnetic field is open at these heights and temperatures. At mid-latitudes, where the fields are typically closed above by high temperature loops, the only topologically possible solution for open field lines is at the intersections of the network boundaries. In the polar coronal hole, where the large scale field is open above, the blueshifts become visible along the network boundaries as well as at the intersections.

In summary, these Ne VIII observations reveal the first two-dimensional coronal images showing velocity structure in a coronal hole, and provide strong evidence that coronal holes are indeed the source of the fast solar wind. These observations also show that the observed outflow velocities originate predominantly along chromospheric network boundaries, with the strongest outflows occurring at the intersections of network boundaries. The apparent relationship to the chromospheric magnetic network, as well as the relatively large outflow velocity signatures at the intersections of network boundaries at mid-latitudes, is a first step in understanding better the complex structure and dynamics at the base of the corona and the source region of the solar wind.

1. Krieger, A.S., A.F. Timothy, E.C. Roelof, 1973. Solar Phys., 29, 505.

2. Zirker, J.B., ed., 1977. Coronal Holes and High Speed Wind Streams, Colorado Associated University Press, Boulder.

3. Cushman, G.W. and W.A. Rense, 1976. Ap. J., 207, L61.

4. Rottman, G.J., F.Q. Orrall, J.A. Klimchuk, 1982. Ap. J., 260, 326.

5. Orrall, F.Q., G.J. Rottman, J.A. Klimchuk, 1983. Ap. J. Lett., 216, L65.

6. Fleck, B., V. Domingo and A.I. Poland, eds., 1995. "The SOHO Mission", from Solar Phys., 162, nos.1 &2.

7. Wilhelm, K., W. Curdt, E. Marsch, U. Schühle, A. Gabriel, P. Lemaire, J.-C. Vial, M. Grewing, D.M. Hassler, M.C.E. Huber, S.D. Jordan, A.I. Poland, R.J. Thomas, M. Kühne, J.G. Timothy, O.H.W. Siegmund 1995. Solar Phys., 162, 189.

8. Siegmund, O.H.W, M.A. Gummin, T. Sasseen, P. Jelinsky, G.A. Gaines, J. Hull, J.M. Stock, M. Edgar, B. Welsh, S. Jelinsky, J. Vallerga, 1995. Proc. SPIE, 2518, 344.

9. Wilhelm, K., P. Lemaire, W. Curdt, U. Schühle, E. Marsch, A.I. Poland, S.D. Jordan, R.J. Thomas, D.M. Hassler, M.C.E. Huber, J.-C. Vial, M. Kühne, O.H.W. Siegmund, A. Gabriel, J.G. Timothy, M. Grewing, U. Feldman, J. Hollandt P. Brekke 1997. Solar Phys., 170, 75.

10. Lemaire, P., K. Wilhelm, W. Curdt, U. Schühle, E. Marsch, A.I. Poland, S.D. Jordan, R.J. Thomas, D.M. Hassler, M.C.E. Huber, J.-C. Vial, M. Kühne, O.H.W. Siegmund, A. Gabriel, J.G. Timothy, M. Grewing 1997. Solar Phys., 170, 105.

11. The raster sequence used to produce these spectroheliograms consisted of moving the 1 x 300 arc second slit in roughly 3.0 arc second steps across a 9 arc minute region. This step size, along with an integration time of 150 s per step position provided adequate spatial resolution to resolve fine scale structures and sufficient signal-to-noise and counting statistics to perform high resolution spectroscopy.

12. Brekke, P., 1993. Ap. J. Suppl., 87, 443.

13. Awaya, T. (1979). Nucl. Instr. and Meth., 165, 317.

14. Baker, S. and Cousins, R. D. (1984). Nucl. Instr. and Meth., 221, 437.

15. Hassler, D.M., G.J. Rottman, F.Q. Orrall, 1991. Ap. J., 372, 710.

16. Rottman, G.J., D.M. Hassler, M.D. Jones, F.Q. Orrall, 1990. Ap. J., 358, 693.

17. Bockasten, K., R. Hallin, T.P. Hughes, 1963. Proc. Phys. Soc. London, 81, 522.

18. Fawcett, B.C., B.B. Jones, R. Wilson, 1961. Proc. Phys. Soc., 78, 1223.

19. Dammasch, I.E., K. Wilhelm, W. Curdt, D.M. Hassler, 1999. A & A, in press.

20. Peter, H. 1999. Ap. J., 516, in press.

21. Brekke, P., and D.M. Hassler, 1995. Needs for Improved Laboratory Wavelengths in the EUV, Laboratory and Astronomical High Resolution Spectra, ASP Conf. Series, A.J. Sauval, R. Blomme, and N. Grevesse, eds., 81, 589.

22. A "redshift" or Doppler shift towards longer wavelengths indicates object moving away from observer. A "blueshift" or Doppler shift towards shorter wavelengths indicates object moving toward observer.

23. Warren, H.P., J.T. Mariska, K. Wilhelm, 1997. Ap. J. Lett., 490, L187.

24. DeForest, C.E., J.T. Hoeksema, J.B. Gurman, B. Thompson, R. Howard, S.P. Plunkett, R. Harrison, D. Hassler, 1997. A Comprehensive Observation of Polar Plumes, Solar Phys., 175, 393.

25. Frank, Z., T. Tarbell, J. Covington, 1998. EOS, 79, S254.

26. Covington, J., T. Tarbell, P. Brekke, A. Fludra, M. Ryutova, 1998. EOS, 79, S265.

27. Dammasch, I.E., D.M. Hassler, W. Curdt, K. Wilhelm, 1999. Proc. SOHO 7 Workshop, in press.

28. Dere, K., 1986. Ap. J., 305, 947.

29. Brekke, P., D.M. Hassler, and K. Wilhelm, 1997. Solar Phys., 175, 349.

30. Chae, J., H.S. Yun, and A.I. Poland, 1997. Ap. J., 480, 817.

31. Gabriel, A.H., 1976. Phil. Trans. Roy. Soc. London, A281, 339.

30. Gabriel, A.H., 1977. "The Energy Balance and Hydrodynamics of the Solar Chromosphere and Corona", Proc. IAU Colloq., 36, 375.

32. Dowdy, J.F., D. Rabin, R.L. Moore, 1986. Solar Phys., 105, 33.

34. Marsch, E. and C.-Y. Tu, 1997. Solar Phys., 176, 87.

35. Hassler, D.M., K. Wilhelm, P. Lemaire, U. Schühle, 1997. Solar Phys., 375.

36. Wilhelm, K., E. Marsch, B. N. Dwivedi, D.M. Hassler, P. Lemaire, A. Gabriel, M.C.E. Huber, 1998. The Corona Above Polar Coronal Holes as seen by SUMER on SOHO, Ap. J., 500, 1023.

37. Dupree, A.K., M.J. Penn, H.P. Jones, 1996. Ap. J. Lett., 467, L121.

38. Andretta, V., and H.P. Jones, 1997. Ap. J., 489, 375.

39. The authors would like to thank and acknowledge the essential and untiring help of Nicolas Morisset of the SUMER team for solving all of the operations and software problems and system management bugs with enthusiasm and good humor. SOHO is a project of international cooperation between ESA and NASA. EIT images are courtesy of the SOHO/EIT consortium. The SUMER project is financially supported by DLR, CNES, NASA and the ESA PRODEX program (Swiss Contribution). A portion of the work of D.M.H. has been supported by NASA under grant NAG 5-6027 to Southwest Research Institute and by CNRS through the "poste rouge" visiting scientist program. HEM acknowledges the financial support of PPARC.

Fig. 1. Outline of the SUMER observations a) in a mid-latitude, predominantly closed magnetic field region on 22 September 1996, and b) in the open magnetic field region of the north polar coronal hole on 21 September 1996, superposed on Fe XII 195 Å images (formed at roughly 1.5 million degrees) taken at the same time with the Extreme Ultraviolet Imaging Telescope (EIT) instrument on SOHO. Bright regions on the disk indicate hot, dense plasma loops. (Images courtesy of the EIT Consortium.)
 

Fig. 2. a) Sample SUMER spectrum showing the 20 Å bandpass used for these observations. The strong second order line of Ne VIII 770 Å, as well as the first order lines of C IV 1548 Å and Si II 1533 Å span a wide range of temperatures and heights in the solar atmosphere. b) SUMER spectrum of the Ne VIII 770 Å line (upper Gaussian curve) superposed with a HRTS first order spectrum (dot-dashed line) showing the contribution of the small Si I blends. The lower Gaussian curve (dashed line) shows the residual second order Ne VIII 770 Å profile after the scaled, first order HRTS spectrum has been subtracted. The dotted lines superposed on the Gaussian curves show the fits to both Ne VIII profiles (before and after subtracting the contribution of the Si I blends). The two Gaussian line centers differed by 0.03 pixels, illustrating the small effect of the Si I blends on the Gaussian line center.