AaPd  #HH $ d HHHHff@dFootnote TableFootnote**. . / - :;,.!?  ^; 9^;TOCHeading   EquationVariablesjYyc Y-Z-[.\.]/^/_1`1a0b0e3f3-b-. / 1 0 3!<$lastpagenum><$monthname> <$daynum>, <$year>"<$monthnum>/<$daynum>/<$shortyear>L;<$monthname> <$daynum>, <$year> <$hour>:<$minute00> <$ampm> "<$monthnum>/<$daynum>/<$shortyear>-f<$monthname> <$daynum>, <$year> "<$monthnum>/<$daynum>/<$shortyear>  <$fullfilename>  <$filename>  <$paratext[Title]>  <$paratext[Heading]>  <$curpagenum>@  <$marker1>d <$marker2> (Continued) Figure & Page(Figure<$paranumonly> on page<$pagenum>!Pagepage <$pagenum>9Heading & Page<$paratext> on page <$pagenum>See Heading & Page%See <$paratext> on page<$pagenum>. Table & Page'Table<$paranumonly> on page<$pagenum>i+ (Sheet <$tblsheetnum> of <$tblsheetcount>)-Section & Page%Section<$paranum> on page<$pagenum>/Equation Number`Headinge<$paranumonly>lww 0lxxlyythne>lzz$mohnl{{A;l||r> Aol~"Aol>-fl> l/<$Arle> late[TladiA>l TOCkerl LOFtinlureLOTranlIX1+K9K =&tdedede? @uEbE J$!J >$i?$ @&oA$nB&oC&oD& E$/F&NG&oH&I$aJ&K&L$wM&0N$O&P&Q&R&_{`dad{b;cd|ddoedfd"gGhdfiGjdk&ld$m&nd&od&pd$qd&Tr&s&it&u&vd&wx~&nyd$zd&e{d&|&}d&1~&9d& d$t#}&e~&ed&e& &u& d&$d& d&i$ d$od&nd&o~od$E&F~&Gd$H&Id&J&K&L~&Md&Nd&O&Pd&Qd$R&_d&`&a~&bd&d$d&d&$d&d&fd$d&d&=d>ddCdDdqdrstuvdw xx yy zz E&{ GE|&E&d}&}&d}&d}$$}&~}$d}&%}$&}&'}&d*E&d)E&d+E&,E&d-E&d6E&d1E&~2 E&3!E&5"E&8#E&9$E&=%E&A&{&F'bEL(}&M)&N*E&uAd"udw|dvd$H$ $wxu H$ dUU` Hz %xwyu Hz dUU` H$ &yxzu H$ UU`SOI-MDITitleSU-LPARL Hz 'zy{u Hz UUh$$a3/24/94b ]#^Version HH({z|u HH`, HH)|{u 1HH5` d}~L1cnQ\(~} 1cnQ\(` Hhnyh~} Hhnyh~dH~6 H~6UU`"<$paranum><$paratext><$pagenum> `"<$paranum><$paratext><$pagenum> )US UT`"<$paranum><$paratext><$pagenum> HnO `@ HnO `@HwjHwjTable of Contents SpecificationH- H-UU`"<$paranum><$paratext><$pagenum> H `@ H `@HEHEList of Figures SpecificationH6 H6UU`1"<$paranum><$paratext><$pagenum> HS `@ HS `@HEHEList of Tables Specificationd H~ H~`1, 23 `$<$symbols><$numerics><$alphabetics> $p`> Level3IX &`m> Level2IX >0` UT Level1IX C` atLSymbols[\ ];Numerics[0];A;B;C;D;E;F;G;H;I;J;K;L;M;N;O;P;Q;R;S;T;U;V;W;X;Y;Z L` Ta <$pagenum> ts Hl@mV `@ Hl@mV `@HuHuIndex Specificationdt"d:LiHH; HH" `$SOI-MDI Magnetic Field Observations m>-UT UT`enJ. Todd Hoeksema @LUR UT`'SOI-TN-113, 18 March 1994 g LiaThis technical note describes analysis done on the magnetic data series obtained using the Image u^Stabilization System in October 1993, with the MDI instrument at LPARL. The purposes of this \task were to determine how well the instrument measures the solar magnetic field, determine Wwhether further calibration and characterization measurements were required, and begin ;I;@S;'development of calibration procedures. age V_Described below are the data sets available and the various analyses completed. A fairly large @=number of figures are included to document the conclusions. UP UT`I. The Data Sets  _Magnetic field observations were taken with MDI on October 20 and 29, 1993. These are the only `et]such measurements taken with the Image Stabilization System operating. Possible ISS-on data 3esets include series 502, 503, and 552. The full disk (FD) images in 503 were clipped at the limbs. g+bSeries identified in earlier lists, 561, 562, 522, and 525, were either unsuitable or incorrectly 9@ labeled. aQ ne]The best series, 552, ran 1993:10:29_20:19-20:45 and included a single FD magnetogram at the h_brZbeginning followed by 3 complete MAG sets. Each MAG set contains 4 groups of 10 images: mraXFD, High-resolution (HR), FD Calibration (FDCal), and HR Calibration (HRCal). The 41st th{ cXimage was sometimes a dark frame. The 10 image sequences were: L0 R0 R1 L1 L2 R2 R3 L3 nt cL4 R4 with the letters referring to Right or Left circular polarization and the numerals to tuning wit0 ]positions 0-4. Data were recorded on the VAX with 2x2 summing to achieve a cadence close to ete`4.4 seconds. Thus an entire sequence took about 45 seconds; it took about 36 seconds for the 8 Thma]filtergrams used to calculate the magnetograms. Exposure times were 1000 clicks. Between 40-s@ndWimage groups the ISS had to be relocked because its range was exceeded by image drift.  ieVSeries 502 began 1993:10:20_20:14:57 and included 3 MAG series. Image times were not nn`recorded in the 502 images. For that reason and because the weather may not have been as good, @ol,most of the analysis focused on series 552. ib `he%Raw data file names are of the form: $UU`ra2/home0/soi/data/oct552/lev0/vmagmix552.XXXX.fits 4UU` 1/home0/soi/data/oct502/lev0/magneto502.XXXX.fits K`olDReduced data described below currently reside in subdirectories of s bUU`co/home0/soi/todd/MAG x2y e [Sample filtergrams are shown in Figure 1a-j. These show L3 Full Disk, High Resolution, FD sedUhfierHHV EHH%im\Calibration mode, and HR Calibration mode images from series 502 and 552. Extra 502 series es[FD and FDCal images are shown with wrapped scaling for information. There were only a few $@thsunspots at this time. easGUT UT` wII. Data Analysis een_ _RCP and LCP spectral line wavelengths are effectively displaced linearly with magnetic field. amem _Thus the difference between RCP and LCP velocity gives the magnetic field with a scale of 2.84 me0{evcm/s per Gauss for the calculated splitting of the g=1.5 6768Ni I line. Figure 2 shows the FD LCP ectZvelocity computed from the second FD magnetogram sequence on Oct 29 (552B); the image has shdbeen rotated to approximately standard coordinates, with solar north at the top. This, like most V ^and B images with nice backgrounds in this report, has been clipped at a radius of 240 or 241 fpixels to eliminate garbage at and beyond the limbs. The  getlimb program found the radius of arcthe filtergrams to be 243 pixels. The filtergrams for this Dopplergram, L1-L4, were registered as s a@UTdescribed below. I`en A. Centering/Registering  e dThe image rotation due to the heliostat is 900 arcsec per minute, a little more than 2.1 CCD pixels dRC]at the limb. During the 36 seconds needed for one of these magnetograms a limb feature will e"lc^move about half a big pixel. This rotation has not been accounted for in the computations of 0@d ,V_RCP, V_LCP, or individual magnetograms. 2H e cSolar rotation is much slower, moving a disk center pixel 2 km/s or 0.0028"/s. This would only be toVt ^important when combining magnetograms taken several minutes apart. In 30 minutes disk center d@r [pixels will move 5 arc seconds or about one 2x2 VAX pixel. This motion has been neglected. mb|&thfFigure 3 shows the center position and radius determined by the getlimb program for each of isdthe FD filtergrams in series 502 and 552. Because the ISS is relocked to a different home position Thffor each 10-frame set, the mean of each series has been adjusted to fit on a single plot with a range atfof 3 VAX pixels (i.e. 12"; thus 1 tick = 1"). There seem to be no systematic differences in radius, utax0, or y0 for LCP and RCP images. The computed radius typically varies 1"-2" from filtergram to , ifiltergram. Variations in x0 and y0 are slightly larger in 552, typically 4"-5", than in 502, typically i28a1"-3". It is somewhat surprising that the variations are so large. This gives some feeling for aIn\seeing noise in the LPARL laboratory setup. Recall that the expected FD unsummed pixel to VA@on:pixel velocity difference is approximately 100 m/s/pixel.  on`The motions in the high-res mode are similar. Figure 4a shows a 4x blow-up of a high-res image il 5X(still summed 2x2 in the VAX) near the sunspot in 552B.HR. Figure 4b shows the same CCD Th @e Bpixels for that sequence of 10 HR images arranged in the pattern: 8` R0L0L1R1 tP`(i R1R2L2L3 1h`e L3R3R4L4 y`es L4R0L0L1 dX rausHHY HH'in ghfThe plot scaling is poor, but the motions of the spots can be seen quite easily, particularly between heVR0 & L0 and L1 & R1. The downward motion between R0 and L0 is nearly 5" (about 4 2x2-$@bosummed HR pixels). thC`D #B. Magnetic Fields - Noise Levels on]`if 1. Registration u . XFigures 5A & 5B show Oct. 29 magnetograms computed without and with registration of the bls afiltergrams used to determine the LCP and RCP velocities. The field is proportional to RCP-LCP. sThXThese plots saturate at a moderate level, 200 m/s (i.e. 70 G), and thus emphasize the R[background noise. The magnetograms have been transformed to solar coordinates. There are 0_noticeable large-scale errors close to the limbs. Most of the disk is covered with noise, due Y bmostly to supergranulation, seeing, and the instrument. Areas with strong field stand out fairly  s_well. Cross sections of the registered and unregistered magnetograms are shown in 5C and 5D. he @tw\While there are differences, the reduction in noise is not striking when shown in this way.  eteThe registration of filtergrams reduces the statistical noise in the magnetic field calculation only 5ne\slightly. Histograms of the two magnetograms (shown in Figures 6A & 6B) have been computed t P hout to a radius of 230/243 pixels. The noise level in the two plots is similar, but a little higher in a 00_the unregistered case, particularly near the limbs. Registered magnetic features also show up ram'orbmore clearly, particularly near the limbs. Because no flat-fielding has been done there is a lot 5ofemore instrumental noise in the registered images - e.g. the hair at about 1 o'clock and the vertical hCas_stripes (rotated from their more familiar horizontal orientation). The FWHM of the registered anQne^magnetograms is 110 m/s = 40 G; FWHM for the unregistered filtergrams is 125 m/s = 45 G. The _ot^standard deviation is about 10 m/s smaller (4 G) when registered. Other statistics for these m@nomagnetograms are as follows: l`Unregistered: UU`s 7image_stats in=552B.FD.B.230.fits CLIP=2000 RADIUS<230 eenUU` G values=166208/262144 min=-1784 max=1654 mean=-3.14654 stddev=69.6143 twoUU`( b7image_stats in=552B.FD.B.230.fits CLIP=200 RADIUS<230 terUU`rlG values=164405/262144 min=-200 max=200 mean=-2.56119 stddev=60.6088 `arRegistered: neaUU`us8image_stats in=552B.FDr.B.230.fits CLIP=2000 RADIUS<230 UU`trF values=166209/262144 min=-1712 max=1238 mean=5.71424 stddev=59.6356 UU` h9image_stats in=552B.FDr.B.230.fits CLIP=200 RADIUS<230 lUU`enEvalues=165195/262144 min=-200 max=200 mean=5.73268 stddev=51.5696 a(  GfStatistics were computed with and without the strong field regions (as reflected in the CLIP level). 6bThe "values" entry tells the fraction of valid points in the image, i.e., excluding those outside D@RRADIUS and greater than CLIP. In all cases, statistics are reported in m/s units. \ en]The 'expected' per CCD pixel velocity noise level for an 8-filtergram measurement is 21 m/s, oj b\according to the IPS. Taking the difference of two 4-filtergram velocities with 2x2 summed 40x `pixels, we would estimate roughly 2*sqrt(2)/2 times the noise, i.e., 30 m/s or 10 G. Given the usd[0UUHH\ mHH&UUat]short exposure times, seeing noise, registration induced errors, etc., this compares not too n@n=Aunfavorably with the 50 m/s standard deviation determined above. t0`t .2. Comparison Between RCP and LCP Velocities H ^For comparison we might consider the difference of two RCP or two LCP velocities which should V^have no magnetic signal. Unfortunately, like other pairs of observations, 552A and 552B were d^taken some 13.5 minutes apart, so the difference is dominated by 5-minute power. The FWHM is r@=about 3x greater. More complete statistics are given below: tUU`wi6image_stats in=tf230.Rdiff.fits CLIP=2000 RADIUS=230 UU`sqIvalues=166209/262144 min=-726 max=790 mean=-3.34571 stddev=134.064 UU`)6image_stats in=tf230.Ldiff.fits CLIP=2000 RADIUS<230 UU`Evalues=166209/262144 min=-677 max=898 mean=12.3002 stddev=132.081 t im]There is a systematic offset of undetermined origin of about -15 m/s between the RCP and LCP = tavelocities that would correspond to a zero-level field offset of -5.5 Gauss. The WSO mean solar d@,magnetic field on Oct 29, 1993 was +0.22 G. id of[The FD velocities were calculated from registered filtergrams and derotated to account for nfohe^heliostat motion. Only the inner 230/242 of the disk was considered. Histograms comparing B  i\and the differences described above are shown in Figure 7. Figure 8A shows the registered, le!gidrotated LCP velocity difference map; for variety the values are shown for the full array. The gray /09ascale saturates at 400 m/s. The signals fall off toward the limb, confirming that most of the 2==2_noise comes from the oscillations. Figure 8B shows plots of two rows and a column for both LCP v=1K@and RCP differences. sc`unCThe corresponding calibration images had the following statistics: zUU`esMimage_stats in=FDCal230.B.fits CLIP=2000 RADIUS=230(cal-mode magnetic field) nUU`Fvalues=166209/262144 min=-154 max=169 mean=14.8703 stddev=33.5514 UU`itKimage_stats in=FDCal230.Rdiff.fits CLIP=2000 RADIUS=230 (cal-mode RCP-RCP) nfoUU`heGvalues=166209/262144 min=-224 max=111 mean=-59.0101 stddev=34.7467 HisUU`B Kimage_stats in=FDCal230.Ldiff.fits CLIP=2000 RADIUS=230 (cal-mode LCP-LCP) 7.UU`thDvalues=166209/262144 min=-222 max=110 mean=-67.9079 stddev=34.8775 ma  v\The difference in the cal-mode RCP & LCP means is +9 m/s or about +3 G. The mean of the FD malTcal-mode magnetic field is +15 m/s or about +5 G. Cal-mode images were left in CCD frs.bcoordinates. The standard deviations are about the same and about half the standard deviation of Ythe standard MDI magnetogram. A FD cal-mode image is shown in Figure 9A; the gray scale Uatasaturates at 100. The LCP cal-mode velocity shown in Figure 9B ranges from -880 left of center 6@to +300 near the right limb. .+ UUaJust for information, the ephemeris Sun-Earth line-of-sight velocity differs by only 20 m/s over U9@16@this time period, so there should be no non-linearity problems. Q imZSince the noise is dominated by the 5-minute power, it would be worthwhile making two RCP _@09TDopplergrams in the same minute (and likewise two LCP Dopplergrams) for comparison. erdm mn HH mHH*od` i% 3. Single Wavelength "Magnetogram" e  ev_The simplest magnetic proxy is derived from the difference of RCP and LCP filtergrams taken at nda.. Ythe same wavelength divided by the sum. The ratio should be roughly proportional to the <caXmagnetic field strength. Figure 10A shows the (RCP-LCP)/(RCP+LCP) ratio for filtergram toJhtatuning position 1 (nominally -120 m), again for set 552B. The two images have been registered, dXm/cso significant CCD irregularities are present. In each of these figures the mean has been removed fe fand the gray level saturates at 20 intensity counts. Near the center of the disk the same features tth_are present as in the magnetic field maps. The contrast is significantly reduced and only the gstrongest field regions show up. Smaller regions and regions near the limbs are difficult to detect. `Fig. 10B shows the result for filter tuning position 2 (-40m). The sensitivity to B is better ererfbecause of the difference in the slope of the spectral line. The registration required was small, but d \Fig. 10C shows how much poorer the signal becomes when the filtergrams are not registered. wsP+`Figures 10D & 10E show tuning positions 3 and 4; note that the polarity has changed because the sim[instrument is tuned in the other wing of the spectral line. The background noise has more nt. fcstructure in it and varies significantly in location from frame to frame. This may well be due to ity cfincreased sensitivity to seeing and image distortion. A histogram of registered tuning position 2 is ifdshown in Figure 11. (Confusion with differences of "missing" values causes the gaps in the line.) s @he/Statistics for the same image are given below: UU`ws6image_stats in=T.rb2.230.fits -u CLIP=100 RADIUS=230 "UU`* bBvalues=166209/262144 min=-97 max=79 mean=0.640693 stddev=6.99297 ;`in 4. MDI Full Disk Magnetograms maS ^Magnetograms are just the difference of RCP and LCP Dopplergrams. The 8 required filtergrams a_are obtained in about 35 seconds to minimize various annoying effects. These effects are more ausodvisible in series 502 than in series 552. Figure 12 shows a set of full disk magnetograms taken in t.} f]series 502 near noon on October 20, 1993. Samples from each of the 3 magnetogram sequences, uZA, B, & C, are shown, as are vertical and horizontal cross sections of the registered and s funregistered maps. The effect of registration is more noticeable than in series 552 (see Figure 5). s ^This raises the suspicion that either the sky was much worse or the ISS was not performing as .2_well. Differences in the cal-mode images discussed below suggest the ISS performance may have .64@7 \been at fault. Or perhaps the ISS was not properly turned off during cal-mode observations. jڪUU` oEimage_stats in=502B.FDr.B.230.fits CLIP=2000 RADIUS=230 (registered) 檧UU` iFvalues=166192/262144 min=-1959 max=1947 mean=-14.8962 stddev=96.0198 UU`Gimage_stats in=502B.FD.B.230.fits CLIP=2000 RADIUS=230 (unregistered) a sUU`gnEvalues=166191/262144 min=-1990 max=1896 mean=-21.041 stddev=108.484 e es`Series 552 is significantly less noisy than series 502. A complete catalog of magnetograms for ti# ccseries 552 is shown in Figure 13. The signal is good closer to the limb. CCD artifacts are still ion1e bpresent, but features appear the same from frame to frame. Figure 5D shows cross sections of 552 ? oaA-C. Averaging the 4 magnetograms significantly reduces the noise level as shown in Figure 14. cMt bThe weak large-scale polar fields are visible in this plot. The histogram comparing the averaged [ri`magnetograms from series 502 and 552 is shown in Figure 14B. The increased noise is due partly IUi ^to the lack of correction for image rotation of the 502 magnetograms and partly to the poorer wim_seeing conditions. Also note that small motion of features due to solar rotation has not been lued.HH eHH'of@ti>considered. The statistics for the average of series 552 are: UU` t4image_stats in=T.552.FDr.Bav.230.fits CLIP=2000 *UU` fHvalues=166209/262144 min=-1342 max=1511 mean=-2.15014 stddev=40.922 ns6UU`8image_stats in=T.552.FDr.Bav.230.fits CLIP=200 rBUU`evDvalues=165413/262144 min=-200 max=200 mean=-1.97993 stddev=30.0951 p[`si< 5. MDI High Resolution and Calibration Mode Magnetograms s etZThe high resolution maps move by several pixels and have not yet been registered for this  \analysis. Still the maps are interesting, as shown in Figure 15. The 502 maps appear much r im^noisier than the 552 maps. Cross sections of the various maps (before conversion to standard acoordinates) are also shown in the figure. Clearly some registration needs to be done to get the e@^most out of these images, as the details seem to change significantly between measurements. `&Statistics for representative images: ڪUU`image_stats in=502B.HR.B.fits Hv檥UU`4 Bvalues=262144 min=-1814 max=2835 mean=-15.9548 stddev=107.419 UU`inimage_stats in=552B.HR.B.fits 200UU`UU?values=262144 min=-1787 max=2014 mean=0.418617 stddev=119.715 dde ]The origin of the spike in the high-res histogram (Figure 15G) at 0 m/s should not come from t#iodmissing data near the edges (as it does in the full disk images). The origin of the spike remains a . 1`mystery for now. Since it is about twice as high as neighboring points, it could be a rounding t?s ]problem somewhere. Zero spikes have been suppressed in most other histograms to account for aM@he?off-limb differences that often produce erroneous zero values. e mo\The calibration maps in 552 are very smooth. A full-disk cal-mode magnetogram was shown in sStZFigure 9. Cross sections of the FDCal magnetograms are shown in Figure 16. The 502 maps 4 ehave large-scale gradients. Because of the large gradients in the cal-mode velocity maps, it is not n00bunlikely that a small misregistration might cause this problem. Misregistrations might be due to inathe ISS reaching a limit, being improperly disabled during cal-mode, or because of a realignment  n`during the calibration time when "it doesn't matter." It is hard to see how seeing, clouds, or @foHsimilar phenomena could cause this kind of problem in calibration mode. aЪUU`:image_stats in=552B.FDCal.B.230.fits CLIP=2000 RADIUS=230 ܪUU`erFvalues=166209/262144 min=-154 max=169 mean=14.8703 stddev=33.5514 誠UU`er;image_stats in=502B.FDCal.B.230.fits CLIP=2000 RADIUS=230 maUU`  sGvalues=166209/262144 min=-692 max=1143 mean=-52.913 stddev=78.2107 StUU`! sBimage_stats in=502B.FDCal.B.230.fits CLIP=200 RADIUS=230  UU`"Evalues=162017/262144 min=-200 max=200 mean=-51.7685 stddev=70.8512 e#  m^The WSO mean solar magnetic field on Oct 29, 1993 was +0.22 G (0.6 m/s); on October 20 it was 1tiW-0.11 G or 0.3 m/s. The MDI measurement is certainly dominated by other effects. The ble? o\uniformity of the cal-mode magnetograms is encouraging. Impressions of the high resolution mMd Scalibration mode magnetograms are similar. Figure 17 shows an example of a HR Cal his[@n =magnetogram and the central row and column of each such map. Dd1609HH vHH)inUU`fi!image_stats in=552B.HRCal.B.fits UU`#ue@values=262144 min=-186 max=151 mean=-15.9465 stddev=32.6213 StUU`$ s!image_stats in=502B.HRCal.B.fits i*UU`% ;values=262144 min=-336 max=151 mean=-84.66 stddev=50.5038 -20C`.7/6. Comparison with Ground-based Measurements me[ fiXKitt Peak, Mt. Wilson and WSO all obtained magnetograms at about the same time as MDI. 1 i MVFigure 18 shows WSO magnetograms for October 20 and 29 and MDI maps averaged to lower wl-_resolution. The comparison is fairly good showing that MDI qualitatively reproduces the large-ibrgrgscale weak field polarity. There are some interesting differences in the stronger field regions. The tra@f ]differences in resolution are so great that it is hard to be quantitative in the comparison.  ]Figure 19 shows the comparison with Mt. Wilson observations. The MDI data have been reduced tfi[in resolution to 128x128 and then interpolated back to 512x512 for comparison with the Mt. im]Wilson data. The maps are remarkably alike (compare 19B and 19D), though the noise level is 620X2.5 to 3 times greater in the MDI data. (MWO FWHM = 3.2 G; MDI FHWM=8.4 G). Note that fiilgthe Mt. Wilson magnetograms have not been corrected for the 1.8 saturation factor in 5250 FeI. how fVThis would reduce the noise factor between MDI and MWO by that amount. IDL's bilinear aiainterpolation was used to reinterpolate the MDI field and this may exaggerate the noise. Notice e e ]that these comparisons were made with single MDI magnetograms, not the series averages. The iut\MDI magnetogram distribution changes width between series 502 and 552, while the Mt. Wilson Fi)@core[are shown in 13 A-D. Each image has been rotated to standard coordinates and saturates at as Lea]200m/s (70 G). The inset at the upper right shows the details in the region around pixel Z@ho+170,280, left of center in the full image. togr th]Figure 14: Averaging the series 552 magnetograms reduces the noise level. 14A shows the / teaverage of all 4 full disk magnetograms out to a radius of 243 pixels. The contrast is good and the rm.\noise level fairly small. The gray scale saturates at 200 m/s (70 G). 14B, 14C, and 14D iepebinclude data out to only 240 pixels and saturate at levels of 500, 100, and 50 m/s (175, 35,  tcand 18 G) respectively. The inset at the upper right again shows the area around pixel 170,280. m/V14E shows various cross sections of 14A. 14F compares the histograms of the averaged _magnetograms for series 502 and 552. Because time information was unavailable, the 502 series atu@Fmagnetograms were NOT corrected for rotation of the image in the lab.  iobFigure 15: Unfortunately, high resolution filtergrams have not been registered before making opplergrams and magnetograms. Still the resolution is improved over the full disk images llUshown above, as seen in 15A. The noise level is much higher. 15B shows another 552 rm.^magnetogram. Unlike 15A and 15B that saturate at 200 m/s (70 G), 15C saturates at 500 m/s $pe`(175 G). 15D is the high resolution field from series 502. The seeing appears to be worse on , 2 tYthat day. 15E and 15F show cross sections from 552 and 502 high resolution magnetograms 1@]before changing to standard coordinates. A histogram of 552B.HR taken on Oct 29 is shown in aN@.15G. Remember these have not been registered. f oneFigure 16: The center row and column cross section of each of the series 552 full-disk cal-mode ot lXmagnetograms are shown in 16A. The images are flat and have a relatively uniform noise nobe^distribution across the field. 16B shows the same thing for series 502. These are quite non-d # aseHH  HH *ma 1`uniform on the large scale and have a higher small-scale noise level as well. The cause of the 5 @ghnonuniformity is not known. 0. ea_Figure 17: A high-res cal-mode magnetogram from series 552 is shown in 17A. There is some 50<maelarge-scale structure, but not a lot. Cross sections of the three high-res cal-mode magnetograms in nJn aseries 552 and 502 are shown in 17B and 17C. As with the full disk data, there is more noise on eX@er+both small and large scales in series 502. serp caZFigure 18: Magnetograms from various sources can be compared with MDI magnetograms. ~m T18A shows the WSO magnetogram for October 20. They show the large-scale weak-field fes[structure of the photospheric field. 18B and 18C show the corresponding MDI magnetograms. H^For better comparison 18B was constructed from the series 502 average field maps, rebinned to nda32x32 resolution, and finally interpolated back to 512x512; the intermediate binning for 18C was m 0Wto 16x16. The scales saturate at 100 m/s or 35 G. 18D, 18E, 18E', and 18F show the shos Xcorresponding maps for series 552 on Oct 29. 18F saturates at 30 m/s. IDL's bilinear t@mo9interpolation may not be the best to use for comparison. 2 an]Figure 19:  Mt. Wilson magnetograms for 20 October are shown in 19A, 19B, & 19C. The UT derctimes are given as part of the file name. 19A saturates at 70 G, 19B at 35 G and 19C at 20 G. ith X19D shows the MDI magnetogram rebinned to 128x128 and interpolated back to 512x512; the ca\gray scale saturates at 200 m/s (70 G). Histograms of Mt. Wilson and MDI observations are m"@Sshown in 19E on a 30 G scale. Similar plots for 29 October are shown in 19F-19I. ge : ed_Figure 20: The full disk magnetogram for Oct 20 from the KPVT telescope is shown in 20A. diaH wcThe gray-scale saturates at 70 G in 20A, 20B, and 20C. 20B is a close-up of the same area as the 8F VXMDI high-res field (see 15D). 20C has been averaged from 1788 pixels to 596 for better 'sdYcomparison with MDI magnetograms. The corner of the Oct 29 magnetogram was missing, 20D, nrMtcand the seeing was (relatively) poor, 20E. These maps saturated at the indicated values. The Oct es of_30 maps are shown with increasing resolution and increasing gray-scale saturation in 20F, 20G, X19gnaand 20H. The full disk KPVT map reduced to 596x596 pixel resolution is shown in 20I. (Cf. Figs e@ m14A and 15A.)  oncFigure 21: Histograms of the low-field data should indicate the noise level. 21A compares the 9 On ]histograms of the Oct 29 full-resolution 1788x1788 magnetogram with the 596x596 average that sh_is more comparable with the MDI maps. The full-res histogram has been divided by 6 to show on B ie athe same scale. The 1788 noise level (ignoring the bin-size beating problem) is about 1.5 times  f^greater. 21B relates the MDI noise level in a single magnetogram (552B on Oct. 29) with the ngVOct. 30 KPVT 596x596 noise level. The KPVT histogram has been divided by 20 to match ca`scales. The MDI FWHM is about 5 times the KPVT value. The result is quite similar for the full reatYresolution KPVT curve, 21C. The complete histograms from -700 G to +700 G (-2000 m/s to 9$tiX+2000 m/s) for both KPVT-596 and MDI are shown in 21D in a logarithmic format. The MDI Fi2og`curve has been displaced one order for ease of comparison. The slopes in the distant wings are n @e `similar, showing that the scaling is approximately correct. The Oct. 20 comparison can be made isN@it-in 21E. The MDI distribution is much wider. nf hoTFigure 22: Unfortunately no GONG magnetograms were obtained during the October sit) Vmeasurements. A GONG modulation image from 20 UT on Oct 29 gives a small hint of the @m /comparable resolution of the two instruments. V1!KPduLeftdivRight tdv References Td tdesdhedatd1Cd fd0 dtid-5d" D d}ThFirstid e TOCed orIXofd"# esd#" n =ILf= it n T MHeader. h f> i me rFooter. tif? r header first. f@  footer right. fAT  TableTitleT:Table : . fB CellBody. fffffC fff ff 3HeadingTOC. fffD  ff  2HeadingTOC . fE $Body. fF  CellHeading. $$fG H.Z.CStep. fHPHeading1Body. @J   Header. ff@K   Footer. fffffL CBullet. fM  CellHeading. ~~fND  References References. ~~ fOE EquationEquation Number E:(EQ ). HHfP Body4. $fQ $.@Bullet Bullet Symbol\t. ~ fRHQ fu~3HeadingH:\t..\t FirstBody. fS@ Purpose FirstBody. $$fT H.Z.CStep . fffUSE ff 1Step Step Numberq S:.\tStep. fV Extract. ~~fWD  FirstBodyBody. ~;fXQ u~ ReferenceHead 1Heading Rule\t\tReferencesAppendix. fYHQ u~2HeadingH:\t.\t FirstBody. fZFA  eFigure Table Topf Step NumberF:FIGURE .\tBody. fff[S nffep Step Step Number S:.\t. f\  TableFootnote. ~;f]HQ uf~1Heading 1Heading RuleH:\t.\t FirstBody. $f^ .f$.6.H.Body2 . f_@ Abstract1Heading. ~;f`AQ muGU <~ AppendixHead 1Heading RuleA:\tAppendix \tAppendix. fa@ TNTagAbstract. fb@AuthorTNTag. lfc gngulH: odBody3r. fd $dyBody. fePfTitleaBody. ff UFootnote. fgQHeading  Body. fh CellBody. fiT  TableTitleT:Table : . fj $Formula. fkPAppendix Body. flQ f SubHeading . Body. 666 fm  $caption. $fffn 6.Z.CStep1. fo $.Opening. eTfp .$.6.H.Z. Data.notes. cfq~ p 1HeadingTOC . cfr a  . FigureLOFy. cfs  TableTitleLOTt. ft  SeparatorsIX. fu  SortOrderIX. $fv Level3IX.  fw fLevel2IX. fx  Level1IX. fy y GroupTitlesIX. fz aIndexIX. f{PTitleaBody. . $f} 6.OrrIZ.CStep1. ~ f~ $.. 6.H.fZ.~.Body3. fPHeading1Body. f $Body. $f 6.Z.CStep1. f@AuthorTNTag. f@AuthorTNTag. ~ f $.6.H.Z.. ~.Body3H. f $Body. $fff 6.Z.CStep1. fP Heading2Body. fP Heading2Body. f 6.l...Au4tab. . f 6.l...4tab.    Bullet Symbol   Equation Number  Default Boldte fcourier    $ First Letter 3Appendix Callout Run-In Heading  u Step Number  Default Bold  Superscript  Subscript EmphasisEquationVariables   courier courierEq       tAp Emphasis  aN@eeukeThinSufMediumgDoublehThick@i Very ThinijSingle eeeeeeeeegAHFBFHFBFqHFBFHFBFHFBFFormat A efeeeeefAHFBFHFBFHFBFHFBFHFBFFormat B-=. -Comment-=2-YZ.[\/]^0ab1_`Emas3efd BlackT!WhiteddANReddd@Greendd BluedCyandMagentaed Yellow M.Times.P Times-Roman M.Helvetica.BHelvetica-Bold M.Times.Bi Times-Bold M.Courier.PCourier M.Helvetica.P Helvetica M.Times.BITimes-BoldItalic M.Times.I Times-Italic M.Courier.B Courier-BoldCourierF HelveticaBTimesHRegularHRomanMediumRegularABold RegularItalicO{>bxncv5}Q+] mBot87 ,K ֢MR\rWֶՔ˔Gͯ&}JUgl !}B^N:̕Mvƒtl3a4.W'# ؑA,jX)B WliJ6jI6m3*