Entry | Date | Title | Site | Author | #Graphics |
---|---|---|---|---|---|
192 | Fri 07-Aug-1998 | NUW Sonic Removal | Cathy Jirak | ||
188 | Wed 05-Aug-1998 | UW#2 down | John Militzer | ||
181 | Sun 02-Aug-1998 | Zeroing of NUW1&2 Sonics | Cathy Jirak | ||
173 | Thu 30-Jul-1998 | NUW#1 Zeroed | John Militzer | ||
169 | Wed 29-Jul-1998 | UW1 Rezeroing and problems | Steve Oncley | ||
166 | Tue 28-Jul-1998 | Reset NUW#1 | John Militzer | ||
156 | Fri 24-Jul-1998 | UW's flipped and rotated | Steve Oncley | ||
148 | Wed 22-Jul-1998 | uw sonic geometry | Steve Oncley | ||
138 | Sun 19-Jul-1998 | nuw comparison | Steve Oncley | ||
134 | Sat 18-Jul-1998 | rezeroed uw2 | Steve Oncley | ||
127 | Thu 16-Jul-1998 | numw sonic booms adjusted slightly | Steve Oncley | ||
124 | Wed 15-Jul-1998 | nuw1 fixed | Steve Oncley | ||
122 | Wed 15-Jul-1998 | uw sonics up to 4.5m | All | Steve Oncley | |
120 | Wed 15-Jul-1998 | UW sonics zeroed | Steve Oncley | ||
118 | Tue 14-Jul-1998 | ati.9m boom angle shot | Steve Oncley | ||
116 | Tue 14-Jul-1998 | atm.9m remounted | NCAR | Steve Oncley | |
114 | Mon 13-Jul-1998 | second K-probe back at 1m | NCAR | Steve Oncley | |
105 | Thu 09-Jul-1998 | csat spatial averaging | Steve Oncley | 1 | |
104 | Thu 09-Jul-1998 | 9m ATI taken down | NCAR | Steve Oncley | |
95 | Tue 07-Jul-1998 | wet towel on sonic | NCAR | Steve Oncley | |
89 | Tue 07-Jul-1998 | Sonic tilts calculated | All | Tom Horst | |
80 | Fri 03-Jul-1998 | Deleted OKMN CSAT from covar.config | OKMN | Tom Horst | |
78 | Fri 03-Jul-1998 | Campbell sonic tests | Steve Semmer | ||
77 | Thu 02-Jul-1998 | Campbell sonic going into marigold | Steve Semmer | ||
69 | Wed 01-Jul-1998 | Sonic elevation angles | All | Tom Horst | |
65 | Tue 30-Jun-1998 | Orientation of OKNM CSAT sonic | OKMN | Tom Horst | |
53 | Mon 29-Jun-1998 | Relevel work on both ATI sonics | Steve Semmer | ||
50 | Mon 29-Jun-1998 | Effect of sonic leveling errors on eddy fluxes | Tom Horst | ||
49 | Mon 29-Jun-1998 | Releveled NCAR 4.5m sonic | Tom Horst | ||
21 | Sun 21-Jun-1998 | Sonic boom azimuths | Tom Horst | ||
4 | Mon 08-Jun-1998 | Sonic testing | Gordon Maclean |
Both UW sonics have been removed from the tower and the tower is down. Water was noticed in the space between the boom and the connector end.
Wed. 5 Aug, ~11:00 UW#2 data stream died At ~15:00 the UW2 apparently brought down the nc_server/ingest. After reboot of system and a few power cycles of the UW2, serial communications with the UW2 sonic continued to spit out garbage. Some portions of the message output was OK and the main command menu would print out interspersed with garbage. Removed sensor: water penetrated into the boom around the connector. No obvious water in and around the PC board or for that matter the connectors themselves. Nevertheless, the whopping 1.8mm of rain this morning must've been too much for the UW. NOTE this UW was the one mounted upside down and it may have gotten water into the 2 mounting screws which normally would be on the bottom of the xarm. 6:00 CDT Rain, 1.8mm 11:00 UW sensor data going bad 15:00 UW hung up base ingest.
After seeing the accuracy of NUW1 go south again. We decided to rezero. It was noticed that there was a diurnal pattern to the sample loss problem. After zeroing the NUW1 the sample rate stayed at 40 even after the signal began to go. It was then decided to rezero NUW2. So far the sample rate is at 40 where it normally would have 36 or so. Will watch to see how long this lasts. NUW2 U T Rh d=.201 37.9 21.87 o= ??? V d=.20 37.89 21.78 o=.028 W d=.199 37.95 22.02 o=??? NUW1 U d=.201 38.1 21.89 o=3.1 V d=.199 38.1 22.2 o= ?? W d=.198 37.9 22.0 o= .01
Jul 30, 9:45 CDT Zeroed UW#1 This was done because this sensor has been reporting erroneous 'U' data for several days beginning a day or two after being repositioned to point SE. Power cycling hasn't helped. Conditions: T~= 31.4degC, RH~=39% Reported by Sonic: Distance Offset U, (a-axis) 0.199176 1.845867 (egads!) V, (b-axis) 0.199507 .001124 W, (c-axis) 0.197715 not reported! We'll keep an eye on it. Suspect it won't make a big enough difference to correct the problems being observed...
ZEROING THE NUW SONICS: (Notes from Steve Oncley) ------------------------------------------------- > > > Did you have a zero chamber or do we need to make one? Procedure > looks easy. > Yes, there is a suitable chamber there. In the wall cabinet above where the Toshiba was sitting there should be at least 2 yellow "squishy" foam and 1 hard blue foam zero chambers on a middle shelf. The long blue foam chamber is what you want. It opens with a magnetic catch and closes fairly well over the path. You will need to know the temperature and RH when you do the zero. I did the zeros with the sensor on the tower (accessed using the step ladder) and rserial'ed to marigold 202(?) to grab current T/RH reading at 4.5m. You should be able to read the Xtemps and XRH displays as well. The biggest problem is, as you may have noticed already, is that rserial to the UWs gives you the prompt after you've entered the data. I just memorized the prompts: C (to enter calibration mode) Y (in response to: Are you sure?) xx.x (for temperature) yy.y (for rh, in percent) U, V, or W (for the path: U=a, V=b, W=c. HAVE THE CHAMBER ON THE APPROPRIATE PATH BEFORE HITTING RETURN.) (wait for about 40 dots, then hit "Enter" a few times to restart) As you've probably seen in the pages I inserted into the manual, path a is the north-south path, path b has the top transducer to the east, and path c you can figure out from there. Tony drew this in pencil on the top of the boom near the array. If you are going to zero one path, I would go ahead and zero all of them (for the one sonic). My guess is that zeroing isn't going to help. Bringing it inside and running it on the Toshiba may be necessary to figure out what is going on.
July 28, 15:35CDT Power Cycle UW Sonic #1 Reason: For the last 1-2 days this sensor has been reporting erroneous speeds/vectors. The UW's were reoriented on Friday July 24, before this one began acting up. That began on ~ July 26, 7-CDT.
Jul 24, Friday, as of10:00 this morning, I flipped and rotated the UWs. nuw1 is upright and nuw2 is flipped. We'll have to shoot boom angles. After shooting the boom angles: uw1 = 158.5547 uw2 = 202.9708
I've measured the geometry for both UW arrays. However, I'm not terribly confident in them since the paper disks moved a bit on the double-sided tape and did not form a good reference surface. Of course, I also was eyeballing the centers. These readings are mostly within 1mm of similar measurements, which would imply an overall accuracy of 0.3 degrees. I'll work up the actual angles if I have time. uw1 uw2 ab/bb 87.00 86.87 bb/cb 86.74 86.82 cb/ab 87.01 86.78 at/bt 87.40 86.46 bt/ct 87.47 87.25 ct/at 87.03 86.41 ab/at 199.43 201.56 ab/bt 180.39 181.33 ab/ct 180.24 182.80 bb/at 180.48 183.13 bb/bt 200.60 201.46 bb/ct 181.31 182.81 cb/at 179.00 181.36 cb/bt 179.99 180.24 cb/ct 199.00 196.26
I've just started to look at the flow distortion from the parallel UW sonic array. The definitive comparison will have to wait until I have real array geometries loaded. (I need to obtain a calipers at least 8" long, which I could either try to buy tomorrow, borrow from OU, or have John bring out - after borrowing from our shop.) Also, I should redo covars with spike replacement. Perhaps I'll even start that now. In general the comparison is excellent (it should be for parallel!), except for wind directions from the NE. I assume that the wake of uw1 contaminates uw2's measurements for this direction and causes uw2 to read somewhat lower magnitudes, with more scatter. This is consistent with the data. I plan to change to a flipped and rotated configuration shortly before I leave next week.
Because I thought I saw that the sonic #2 array was distorted through the theodolite, we just check the pathlengths using the internal calibration. They all worked out within 1mm of each other, so the array obviously is okay. (It doesn't look too bad up close.) I still have to figure out how to measure the geometry.
I moved both UW sonics about 2mm to the west between the brackets on the S crossarm - in other words, less than 1 degree. I also elevated them a bit with foam rings, about 1mm thick when compressed. This was done to prevent the arms from moving much. I'll now shoot boom angles. P.S. As noted in the status, these sonics were changed to report at 5 samp/s to reduce the data load on ragwort. P.P.S. Boom angles are: uw1: 001 deg 8' 00" uw2: 001 deg 43' 18"
Based on several pieces of evidence (pathlengths of 18.7 cm after zeroing, looking crooked, and having to pivot the mount last night), I realized that the sonic array was damaged during shipping. (I am guessing that this happened on the way TO ATI, since their pathlengths also were 18.x.) After taking it down, one, and later two of the epoxy joints separated. I reapplied epoxy about 12:00 today and baked it in the car from 1:00 - 4:30. I then reassembled and mounted it, and have just rezeroed it. The coefficients are now: A B C pathlength 0.198673 0.199728 0.198278 offset -0.047704 -0.028221 0.018013 The offsets are a bit larger than I would have liked, but the pathlengths now look pretty good (though still not as close to 0.200 as nuw2). FINALLY, we are now starting the parallel part of the flipped and rotated tests on the UWs!! Note: We still need to shoot boom angles.
The UW sonics have been moved up to 4.5m in the nominally parallel configuration. (I had wondered why the data didn't agree better between the UW and ATIs and then realized that the UWs had been at 1m.) Unfortunately, it appears from a distance that the UW1 (actually UW3) array is bent! This also would explain why the zero calibrations found that the paths were 18.8, rather than 20 cm. I noticed that the array had been skewed during shipping, but it may have suffered even more. I'm going back now to bring it down and try realigning it in the jig.
Both UW sonics were just calibrated. nuw1 (east): [SUW3] A B C pathlength 0.187446 0.188175 0.189011 offset -0.011071 -0.009161 -0.000487 nuw2 (west): [SUW4] pathlength 0.199335 0.199497 0.199539 offset -0.013521 0.015395 -0.006920 All the offsets look good to me, and the pathlengths for nuw2 are great. I don't know why the pathlengths for nuw1 are short, but I'll ignore it for now. Now I get to figure out what is wrong with the calibration routine.
Tony and I just shot the boom angle for the 9m ATI with the theodolite. His reading was: 3 degrees 16' 40" My reading was: 3 degrees 14' 08" I think both readings were about of equal weight, so I would average these to 3 degrees 15'. This was shooting the back "U" transducer through the arm supporting the front "U" transducer (and probably the top "W"). I noticed that the 4.5m ATI was facing more easterly, i.e. an angle more like 1 degree?
We've RTV'd the ati K-probe back together. I also did a laboratory (trailer) zero calibration on all 3 paths. It found biases of about 1 cm/s and path- lengths of about 14.6 cm for all paths. The boom is drooping a bit, just due to the weight. I've done as much as I can with the yoke - there is a bit of play in the square part of the yoke. We could try shimming. For the moment, I've just left the droop in, assuming that it can be removed later by tilt.sonic. We still need to shoot the new boom angle. I wouldn't expect it to have changed much.
We installed the replacement U2 transducer in the ati K-probe and have had it running at 1m on the UW tower for about an hour. We'll continue to run it for a while to decide if it is working well enough to mount at 9m. When we do, we will have to: 1. zero it 2. feed a bit more cable up the tower 3. shoot the boom angle.
Tom has e-mailed me the response for spatial averaging for one path from Kaimal's 1968 paper, and I have the data for my model of the spatial averaging for the UW. I have added a plot of these to this entry. Generally, the response with 3 paths at 60 degrees from a sonic with a 10 cm path (the CSAT) to W is about the same as a single vertical path with a 15 cm pathlength. Thus, spatial averaging does not explain the differences we are seeing between statistics from the CSAT and ATI.
To allow us to check why the 9m ATI is spiking on U, I've climbed the tower and taken the sonic head down. (The Krypton is still mounted to the boom at 9m). Note that I had to disconnect the yoke in order to give the cables enough play to allow me to pull the head out of the boom, so TILT CORRECTIONS WILL NEED TO BREAK THE FIT AT THIS TIME. The head is now reconnected and running on channel 201, but is physically at 1m on the UW test tower (which has no other sonic running on it yet). Already the U channel appears to be spiking, so this is a good test configuration. We're going to lunch now. Assuming that it is still bad after lunch, I'll try looking at the signals on the scope. ---------------------------------- 5:30 PM It appears that the u signal is weak, I think going from u2 -> u1 (the 4th of 6 transmissions). I'm going to try to get another transducer sent here as a replacement. Note that this sonic has been off and on all day as I fiddled with it - all at 1m on the UW tower. I've taken it off completely for the night, since it is not weather tight. I'll put it back up to 9m tomorrow.
In an attempt to figure out why the 9m ati is spiking, I wrapped the boom with a wet towel about 1400 CDT. (We hoped that the electronics would be cooled.) This did not affect the spiking, but probably disturbed the krypton humidity measurement. I'll remove it in another 15 min. or so. Towel was removed about 15:05. It was still damp. I also used a finger to wipe a bit of dust off of both U transducers. I doubt that any of this made any difference.
Steve calculated the sonic tilt angles yesterday, using fun.sonic.tilt.dat. We used three days of data beginning June 29 16:30 CDT. Sonic offset pitch roll b1 b2 b3 lean phase cm/s deg deg deg deg --------------------------------------------------------------------- ATIK 4.5m -1 0.4 0.1 -0.012 0.006 -0.001 0.36 170 CSAT 4.5m 3 0.4 0.2 0.029 0.008 -0.003 0.49 157 ATIK 9m -1 -0.4 0.6 -0.015 -0.008 -0.010 0.73 54 These fits included two groups of data, one for wind from the east and a second for wind from the south. We applied this to two sets of data and found that the changes in u* andwere on the order of 1-2% and that the differences between the two 4.5m sonics were less than ~1%.
I had been covaring the CSAT data to compare it to the statistics calculated by OKMN. These statistics were identical with the exception of a small offset in tc (tc' statistics were identical), likely caused by some small difference in the computation of tc from the speed of sound. Now I have deleted the csat from the covar file again in order to avoid needlessly duplicating these statistics in the NetCDF file.
From 8:30 to 10:00 (local), the Campbell sonic was running in an unprompted mode with sync characters. This was a test to see if ASTER would handle this mode and to see if it effected the SDM communication to the OK logger. This mode of operation had no effect on SDM. Refer to document in $ASTER/projects/OASIS98/doc for more information on how to add the sync characters to the sonic output message.
The Campbell sonic is being sampled on channel 206 of marigold. Changes were made to prep.config, channel_config, and covar.config. The sonic is operating in a prompted mode; ie, marigold is sending a "W" to the instrument to get data. The data are also being sampled by OU via the SDM communication. At this time we do not know if we are causing any problems.
Calculated elevation angles of the wind for the period June 29, 16:35 CDT to July 1 8:35 CDT. During this period u ranged from -4 to +8 m/s and v from -5 to +6 m/s. sonic atan(w/u) atan(w/v) 4.5m NCAR 0.4 deg -0.2 deg 4.5m OKMN 0.7 deg -0.5 deg 9m NCAR -0.3 deg -0.3 deg Mean offsets in w were on the order of a few cm/s
Steve shot the orientation of the OKMN CSAT sonic: nominal N/S path 357d 07m 50s nominal SW/NE path 057d 45m 55s nominal SE/NW path 296d 43m 55s
From 14:00 to 16:30 (local) I was playing with the levels of the ATI. I started with the 9m. This required adjustmments on the chain links and turnbuckles on the south guywires. Once this was completed, the 4.5m was adjusted using some panduit ties, a radiator clamp, and some choice words! I talked to ATI about the calibration function for the sensors. It is different from the old units. Therefore the calibration routine for the sonics will need to be modified. The new transfer function is: degrees = counts * 1.5/1024 According to ATI the sensor range is +-4 degrees over a range of +-2048. However it is not that linear over the full range. They recommended using the smaller range. At this time, we are not sure their function is correct. Tom is doing some number crunching to try to confirm this.
Let 'a' be the angle of the u axis above horizontal. Then u_r = u*cos(a) + w*sin(a) w_r = -u*sin(a) + w*cos(a) uw_r = (ww-uu)*sin(a)*cos(a) + uw*(cos(a)^2 - sin(a)^2) (u*_r)^2 = (u*)^2*[1 + (uu-ww)/(u*)^2*sin(a)], cos(a) ~ 1, sin(a)^2 ~ 0 From data, (uu-ww)/(u*)^2 has a value around +5, so a = 1 deg gives an error in u* of about 4%. Assuming w = 0, a = atan(-w_r/u_r) give an angle of 1.64 deg for the ATI sonic and an error in u* of about 7%. For the CSAT sonic, a = -0.63 deg, for a u* error of -3% and thus (u*_r).ATI = 1.1*(u*_r).CSAT The data show (u*_r).ATI = 1.14*(u*_r).CSAT wtc_r = -utc*sin(a) + wtc*cos(a) wtc_r = wtc*[1 - utc/wtc*sin(a)] From data, utc/wtc has a value around -2, so a = 1 deg give an error in w'tc' of about 3.5%. Thus the error in w'tc' for the ATI is about 6% and the error for the CSAT is about 2%, so that wtc_r.ATI = 1.08*wtc_r.CSAT the data show wtc_r.ATI = 1.13*wtc_r.CSAT
Plots of w versus u and v give the following for the three sonics: atan(w/u) w (m/s) atan(w/v) w (m/s) ATI 4.5m -1.64 deg 0.03 -0.70 deg -0.09 ATI 9m -1.29 deg 0.02 -0.15 deg 0.08 CSAT 4.5m 0.63 deg 0.04 -0.04 deg -0.08 We releveled the NCAR 4.5m sonic between 9:30 and 11:00 this morning, using a bubble level. The u axis is level, but the v axis is still high to the west. Steve checked the u axis with the theodolite, looking east along the boom, and it looks good. I looked at the bubble on the OKMN CSAT and it is right on center.
Steve shot sonic boom azimuths with theodolite 4.5m sonic 2d 00m 20s 9m sonic 3d 38m 00s
ATI sonic, S/N 980504, as ragwort channel 201 has showed a few spikes (1 or 2 per hour) on V during testing in the staging area. Other components were clean. Serial #980505 as ragwort 200 showed no spikes. They were configured with a 5ms sampling interval, 10 samples/average, giving 20hz output, 9600 baud, with 2 levels and nsamples also output. The sonics were next to each other, extended from a benchtop, with the V arm of #980505 (which didn't spike) about 3 inches from a wall. Binary mode: In an effort to lower the serial overhead I tried to configure the sonics in binary mode. It mostly worked. U,V,W,Tc looked OK, but the levels (which were one byte each) were always 0. There did not seem to be a way to get nsamples output in binary mode.