Geometry of the GPI diagnostic in NSTX. In (a) is a sketch of the vessel outer wall (as seen from the plasma) showing the reentrant GPI viewport, the manifold from which the gas puff emerges, and the approximate angle of the local magnetic field. The 3-D structure of the turbulence is shown as a “filament,” and the 2-D projection of a filament with the GPI radial vs poloidal plane is shown as a “blob.” In (b) is an equilibrium at the L-H transition time for #135042, along with the GPI area projected into the (R, z) plane, the location of the GPI manifold (the line just outside the GPI area), and the projection of the rf antenna/limiter in this plane (covering far right around the midplane).
Typical GPI images of the light emission in this experiment. At the top is an L-mode image and at the bottom is an H-mode image later in the same shot, both images taken with exposure times and the same (false) color intensity scale. Also shown is the best estimate for the separatrix location (dashed line) and the shadow of the rf antenna/limiter location (dotted line). These images cover a region of in the radial (horizontal) direction and in the poloidal (vertical) direction, and have a pixel size of . The range of GPI turbulence analysis is shown by the rectangle in the middle.
Time dependence of light emission from the GPI diagnostic within a 1.5 cm wide region just outside the separatrix at . This GPI signal drops rapidly at the H-mode transition at , at about the same time as the standard light emission far from the GPI puff (the latter has a slower response time). At the right are Thomson scattering profiles just before and just after the H-mode transition showing the formation of an edge density transport barrier after the transition. The times of these Thomson scattering data are shown at the bottom left. The radial range of the GPI diagnostic with respect to the outer midplane flux surfaces is also shown at the right, along with the separatrix location (labeled sep.).
Time dependence of the GPI signal during an time period around the L-H transition for the same spatial region, i.e., just outside the separatrix for the same spatial region as Fig. 3. Three successive shots are shown which were taken under identical machine conditions. In each of these shots there are many transient quiet periods preceding the transition, which have a GPI signal level similar to the period after the L-H transition.
Sequence of GPI images for an (70 frames) period showing a typical quiet time (#135044 at 0.237 225–0.237 435 s). Each frame has an exposure time of , and the approximate location of the separatrix is shown by the vertical line in each frame. This quiet period (labeled “Q”) lasts for frames, i.e., . During this time the GPI images look like those seen in H-mode rather than those usually seen in L-mode (see Fig. 2). (enhanced online). [URL: http://dx.doi.org/10.1063/1.3476276.1] [URL: http://dx.doi.org/10.1063/1.3476276.2] [URL: http://dx.doi.org/10.1063/1.3476276.3]10.1063/1.3476276.110.1063/1.3476276.210.1063/1.3476276.3
Fraction of the GPI light emission located outside separatrix vs time for the same three shots as for Fig. 4. The lines are this fraction smoothed over 0.7 ms (200 frames), and the dashed line at 0.15 is just shown for reference. In all cases rapidly drops below 0.15 at the L-H transition (vertical line), but occasionally goes below 0.20 before the L-H transition (quiet periods). Well after the main transition, shows intermittent bursts above for the remainder of the H-mode period.
Radial profiles and relative fluctuation levels of the GPI data for #135042 during an period preceding the L-H transition, sorted according to (fraction of GPI light located outside the separatrix). Part (a) shows that the average radial profile in L-mode (“mean L-mode”) is significantly broader than during H-mode (“mean H-mode”), but the quiet periods in L-mode with look similar to the H-mode profiles. Part (b) shows that the relative GPI fluctuation levels for are also similar to H-mode fluctuation levels, i.e., smaller than L-mode in the SOL. The shaded region shows the location of the steep density gradient region in the H-mode phase of the discharge.
Fast time dependence of the radial and poloidal profiles of the GPI data between before to after the L-H transition for shot 135044. Part (a) shows the time dependence of the GPI light vs radius across a row of pixels at the vertical center of the images, part (b) shows same time dependence of the GPI light vs poloidal distance down a column of pixels near the separatrix at , and part (c) shows time dependence of the GPI light vs poloidal distance down a column of pixels well inside the separatrix at . The corresponding levels are shown in the bars at the right, where white is and black is . The quiet periods are labeled with a Q, the H-mode period is labeled as H, the L-H transition is the horizontal dashed line, and the separatrix is the vertical dashed line.
In (a) are the autocorrelation functions of vs delay time for the three shots of Fig. 6, averaged over 10 ms preceding the L-H transition, and in (b) are the power vs frequency spectra of for the same data. The autocorrelation functions all have a quasiperiodic structure with a period corresponding to a frequency of . This periodicity of the quiet times in the SOL is also visible in the raw data of Figs. 5 and 8.
Part (a) shows the time dependence during an period across the L-H transition for shot 135042 for (just outside the separatrix), along with several other turbulence properties computed from the same GPI image data. The approximate time of the quiet periods is marked with shaded vertical bars in part (a), and the L-H transition with a thin vertical bar. Part (b) shows these same quantities over a longer 17.5 ms period for the same shot. These turbulence quantities averaged over a time interval of around each time point. Positive corresponds to the electron diamagnetic direction.
Time-delayed cross-correlation functions between and other turbulence quantities during a 3.5 ms period just preceding the L-H transition in #135042 (0.239 505–0.243 005 s): (a) vs , (b) vs , (c) vs , and (d) vs S. For this figure the cross-correlations are shown for five adjacent radial regions within the box in Fig. 2, including the case for used for Fig. 10 (green). The cross-correlations of and all show a peak near zero delay, but the time of this peak correlation changes systematically with radius. Cross-correlations of and S are strongest near the separatrix.
The time evolution of several of the quantities of Fig. 11 for the radius during the preceding the L-H transition for the same three shots as in Fig. 9. In part (a) is the magnitude of the first negative peak of the autocorrelation function of , which is a rough measure of the size of the oscillating feature at , and below that is the corresponding frequency of this feature. In part (b) is the magnitude of the peak of the cross-correlation functions between and near zero delay time, and below that is the delay time to this peak. In part (c) is the magnitude of the peak of the cross-correlation functions between and S (near zero delay), and below that the delay time to this peak.
Typical scatter plots of the correlation between and (top), and S with (middle), and and S with (bottom). All cases are for (as in Fig. 11), and all have 1000 time points within a 3.5 ms period ending before the transition. All points are at the time of the peak of the cross-correlation functions. There appears to be a significant statistical correlation between quiet periods with low values of and a positive , as can also be seen in Fig. 10(a), as shown by the linear fit to these data. There is a rather wide scatter of vs S for both values of the radial averaging width used for (middle and bottom).
Analysis of the size and frequency of the quiet feature vs time for a larger database of shots, including the data previously shown in Fig. 12(a). The L-mode shots are put at −150 ms and the H-mode shot is put at to bring them into the same plot. There is no clear change in the size or frequency of the feature during the preceding the L-H transition, consistent with the trend seen in Figs. 10 and 12. The feature also appears in L-mode and H-mode shots, so it does not appear to be a feature specific to the L-H transition.
Time dependences of the SOL fraction and the normalized shear S for a period of before the L-H transition (vertical lines) for two additional shots besides #135042 [already shown in Fig. 10(b)]. Although there is a strong modulation of S with at for at least 15 ms before the transition, there are no clear variations in the average value of or S on this timescale, consistent with Figs. 10 and 12. A radial averaging of was used for the velocity gradient in this analysis.
The normalized shear S for the time period up to before the transition for three radial locations for #135042 (left) and for the superposition of all nine shots at the highest frame rate at the same radii (right). There is again no clear variation in S before the transition which could be considered as a trigger for the L-H transition. A radial averaging of was used for the velocity gradient in this analysis.
L-H transition database of fast GPI data.
Article metrics loading...
Full text loading...