FLATLAND96: SONIC Messages: 54 Entries..

The u axis on the ATI was bad on the mornings of June 14 and 15.  The
signal appeared to improve significantly around sunrise, suggesting a
temperature or humidity problem.

I visited the site between about 1130 and 1230 CDT.  The electronics
enclosure was dry inside.  Looking at the received signals with the
'scope, it was evident that the u axis was noisier than v and w.  I
tried loosening and retightening the transducers on the u axis, but
this had no effect.  I swapped the u and w receive boards and the noise
followed the board.  I replaced the the u and w boards in their proper
slots and reduced the amplitude on the u receive board about 1 full
turn of the pot.  This made the signal look a lot cleaner.  I also
reduced the amplitude on v a little, so that all 3 boards have the
first negative wave at an amplitude of about 4-5 V.

I could not rezero the u (or v) axis, because I did not have a
field terminal to determine the temperature and humidity.  However,
I suspect that my adjustment made little change in the zero.  We will
rezero on a subsequent visit.

Visited station #3 on June 16, between 1400 and 1500 CDT.

Measured sonic height to be 3.21 m

Looked at signals with scopemeter and saw intermittent noise preceding
received pulse on the u axis.  Tried to adjust the capacitor to reduce 
noise, but this only made it worse.  On my previous visit it was 
apparent that this is a problem with the board, and the daily plots
suggested a temperature-dependent problem.

Sonic at station 3 continues to have spiking at the 1% level on 
the u axis, particularly after sunset.  There is also sporadic
spiking on the v axis and spiking on the w axis.

The sonics at stations 1 and 2 appear to be operating okay.

ATI shipped a new Receive amplifier board for the sonic at site 3.
I visited the station to replace the board around 20:30 - 21:00 gmt.

Before replacing the board, I looked again at the signals with the
scopemeter and confirmed that the noise followed the board when I
interchanged the u and v boards.  Then I swapped the boards back and
replaced the u axis board with the new board from ATI.  The signals
look a lot better with the new board.  After running with the new board
overnight, I will send the old board back to ATI for inspection and
(hopefully) repair.

I zeroed the u axis after confirming with the scopemeter that the
signals had the proper amplitude (first positive wave < 0.5V, next
positive wave ~13V).  The zero is probably only good to no better than
0.5 m/s since the wind was blowing parallel to the u axis and the
zero-wind tube is pretty leaky.  After zeroing, I looked with the
Toshiba at the sonic output and it was on the order of 0.5 m/s and
fluctuating by several tenths of a m/s.  Will need to rezero when the
wind is lighter.

I dried up a little water in the bottom of the electronics box and
resealed the edges of the door with electrical tape.  The hole appears
to be limiting the water depth as intended.

The w axis was quite noisy, generating spike flags for as much as
10% of the data, from 0800 to 1900 CST, which appears to coincide with 
daytime heating (temperatures exceeding 24 degC).

The new board on the u axis continues to perform well.

In case the programmed setting is lost on the Fluke Scopemeter, the
settings required for looking at the sonic received pulses are:

Channel A: 2V DC w/ 10:1 probe
Channel B: 5V DC w/ 10:1 probe
Time sweep: 20 microsec/div
Trigger: B; positive slope; level=3V

The flags plot for site 3 has shown spiking at the 1% level on the w
axis, particularly during the daytime, and the zero level appears to be
wrong on the u axis.  Between about 1200-1300 CST on June 20, I swapped
the u and w receive boards to isolate the cause of the problem.
However, I could not see any problems on the scopemeter with either
signal.  I adjusted the levels of the signals (not much adjustment
required) and then we zeroed these two channels.

Looking at the current plots for June 20-21, it is obvious that this
change eliminated almost all of the flags on the w axis, but now there
are continuous flags (both night and day) on the u axis at about the 
0.3% level.  The rezeroing has fixed the offset problem on the u axis.

The Zimmermans will ship the refurbished Receive board back to us
today, but Gary could not find much wrong with it.  I will try
replacing the bad w board with the refurbished board.  Meanwhile I will
leave the new board in the w channel and the old w board in the u
channel.

I visited site 3 to look at the sonic signals.  They look fine on the
oscilloscope, but I hoped to detect enough errors looking at the flags
in order to perhaps readjust the sonic.  However the number of flags 
was on the order of 0.1%, so no luck.

Gary Zimmerman has suggested that we may have the gain too high on
the Receive boards so that the "true" first wave is at 10V rather 
than 0.5V.  I tried reducing the gain all the way, but this did not
reduce my "second" pulse below 1 V.  I returned the gain to the 
previous setting, but did not rezero the sonic.

I noted in the dailyplots that this sonic was very different from sites
1 and 2 yesterday:  u:u, w:w and u* were all lower at this site, but
sigma_w/u* looks about right.  Also tc:tc and w:tc were higher than at
the other sites and w:rh20 was also lower.  This was the one of the few
times that the wind was blowing from the field to the south with no
growth, so perhaps this was the cause.  Also, this field was recently
planted, so this also may have changed its albedo, etc.  Looking at
the data from today, it is quite obvious that this site is different
when the wind has a southerly component and a fetch with no vegetation, 
but matches the other sites when the wind has a northerly component 
and has a fetch over the growing soybeans.

ATI sent back the "u" board from station 3 that I asked them to
repair.  Gary Zimmerman said that he could not reproduce the problem
that I saw in the field.  However, he did "change a couple of things
that might be a problem."

I replaced the "w" board, which had been operating in the "u" channel,
with the refurbished board.  I could not see any problem in the signal
with the oscilloscope, either before or after the swap.  However, I
did monitor the sonic spikes in the field.  

Before the swap, with a leaky zero-wind tube on the u axis:

sample count = 2900, u spikes = 7, v spikes = 0, w spikes = 18
sample count = 2953, u spikes = 4, v spikes = 0, w spikes = 28

Replaced refurbished u card in u channel: 

sample count = 2997, u spikes = 0, v spikes = 0, w spikes = 6
sample count = 2970, u spikes = 1, v spikes = 0, w spikes = 10

John also changed the charge controller in the battery box:

sample count = 2945, u spikes = 0, v spikes = 0, w spikes = 14
sample count = 2991, u spikes = 1, v spikes = 0, w spikes = 4

This suggest better performance with the refurbished board, but no
dependence on the battery charge controller.  Note that I have left the
new ATI board in the w channel.

I tried to rezero the u axis, but the wind was blowing moderately 
from the west, so the zero is probably not too accurate.

Site 1

6/20/96  Sonic height = 3.20 m
	 Model SAT-211/3CDKLT, S/N 950601, labeled #3 by NCAR


Site 2

6/18/96  Sonic height = 3.13 m 
         Model SWS-211/3Kl, S/N 920201, labeled #1 by NCAR

6/19/96  Raised sonic height from 3.14 m to 4.54 m.


Site 3

6/16/96  Sonic height = 3.21 m
         Model SWS-211/3Kl, S/N 920202, labeled #2 by NCAR

Beginning 96/06/25, 14-local, 19Z (exactly the moment when Tom left for the
airport!) the v-component at station 3 began glitching well above 1%, and
as much as 10%.

I visited the station later in the day, about 21Z, but didn't look at the
signals.   I had tried to establish a good zero for the U component and had 
climbed on the tower, but didn't notice anything unusual at the time.

I'll check the site this morning....

U Component zeroed.

V Component
Everything appeared OK upon arrival, both inside the electronics box and on the
sonic head assy.
EVE at the time was only showing 1-2 spikes on V.
I went ahead and zeroed u, then put the chamber on V.  For the most part the 
spikes were few and far between, but once i saw a burst of 35.

Signal quality on scope meter appeared OK, but i wonder if it's being
overdriven although i'm not sure.  The initial rize not only was less than 1v
but it was probably .1v on both the V and W, perhaps .3-.4 on U.
During this period i saw spike counts on W that actually were greater than on 
the V

Ground: i tried to re-establish the ground for the electronics box and found it was about .4ohms.  I also reseated the U,V,W boards and cycled power on the box.

The glitches for the most part over an extended period appeared to be
distributed among the 3 components, mostly on V/W with W possibly being worse.
So I decided not to swap the V/W boards, however, looking at the plots
this morning (Thurs, 6/27) the V spikes are by far worse, even though there
was some terrible spiking on W from about 17Z-0Z (ie when i was there).
I now wish i'd replaced and re-adjusted the 2 boards and will probably do that 
today.

This occurred at perhaps 23Z, 96/06/26

The previous position was at: 3.21 meters from the centerline of the boom on
the mast to the ground.

Station 3 Visit: 17:05-20:55Z, 96/06/27, JD179

This ATI has been having fits since the beginning.  In particular the V
is continuing to spike and the W did yesterday when i was there as well.
Today, I came out to look at the signals and swap the currently worst board (V)
with the currently best board (U).

3 Changes:
	- Replaced the V Transducers
	- Swapped the "V" and "U" boards
	- Adjusted and zeroed all of the signals

I believe the situation is better now, primarily from the adjustments that
were made, however, the "V" board was still showing a few spikes when i left.
Meanwhile, i'm trying to examine the data to determine whether things have
improved according to the data.
By adjusting/damping the signals down in the 3 axis, i hope that the
spiking will clean itself up.  I definitely noticed 'jumping' in the
signals on the 1st peak that supposed to be below .5V, but after i made the
pot adjustments i didn't see very much of that. (except on the "V" board,
but those all seemed to remain below the .5V level at least).

Note: Zero Chamber now in use does let in a bit of wind so some of the
'signal glitches' i observed may have been related to that.  The winds
varied during the visit from about 5-6ms to calm.

I continuously monitored EVE for spike flags and the Fluke Scopemeter.  This
is what i saw/did:


	Sample	     Spike Accum.	
~Time	Cnt	U   V   W	Comments
12:12	1373	    1		Things appear roughly normal even though i know
				lots of spiking is going on.
12:14	2880	1   1		Saw some noise on the v board even with
			 	the zero chamber.  also some waveform that
				could have been some wind.  The 1st peak was
				seeming to jump as high as .8-1V
12:30				Reseated V transducers.
				Still seeing lots of glitching on V
12:35				Replaced Transducers with spare
				Still seeing glitching on V
12:39	730	   131		Yuk.
12:40				Swapped boards "V" board is now on 'U' slot
				and the "U" board is now in the 'V' position.
				Monitoring "V" board:..still seeing front
				end noise.
12:46	492	1   5
12:47	1138	5   5
12:49	2373	13  5		Noise may be better than before but still
				see a few pulses before 1st peak and it's
				still occassionally over 1V...too
				much. zero looks close.
12:52	910	7		Begin adjusting pot counter clockwise to
				damp the signal a bit.
12:58	1853	3
12:59	2783	4		Signal looks better, maybe a few burps to
				.4-.5 V on the 1st peak.  Tweaking pot a
				bit more.
13:01	900	0   0   0
13:03	1933	3   0   0	First peak is barely perceptible and not
				going above .5V.   The 2nd peak is no
				longer saturating, but is hanging around
				11V more like what the manual indicates.
13:05				Zeroing "V" board in 'U' slot: 30-deg 34%RH
13:11	924	0   0   0	Monitoring "U" board in 'V' slot: signal
				looks pretty reasonable.  The 2nd peak
				is just saturating at 12, the 1st peak is
				hanging around .3, but occassionally burps
				to .5
13:13	2151	0   0   0	Adjust pot CCW a bit. 2nd peak now hanging
				in about 10-12 but not saturating.  1st is
				stable at .1-.3
13:19	2362	3   0   0	Zeroing "U" board in 'V' slot:  30-deg 32%RH
				Trouble with getting wind inside the
				tube...
13:27	1135	2   0   0	Rezeroing "U" board: 30-deg, 37%RH..better.

				Monitoring "W" board.
				1st peak is .3-.7, 2nd is saturating.
				Adjusting pot CCW to get 1st peak to
				.1-.35, 2nd peak 10-12V now.  Otherwise
				this board looks pretty clean even though
				we've had spiking the last few days.
13:30				Zeroing "W": 30-deg 34%RH, relatively calm now.
13:38	1530	0   0   0	Need to Rezero "V" board in 'U' slot
				30-deg, 32%RH
13:47	505	0   0   0
13:49	1684	0   0   0
13:50	2130	0   0   0

Judging by the daily plots, the "V" board that i put in the 'U' position
yesterday is still bad.

However, it appears that the adjustment of the U/W seems to have helped the 
spiking there.  I'll keep the fingers crossed.  Other differences in the
sonic data between this site and 1/2 may be attributable to different field
types and conditions.

Station 3 Visit: 20:25-21:45Z, 96/06/28 JD180

Reason:	The "V" board that was put into the 'U' slot yesterday was still
	spiking heavily.  I called ATI about this and... 

ATI Hints per Gary Zimmerman
	Regarding the problem with the "V" board, Gary 
	is sending another board overnight if he has a chance to finish
	going over it.  It is one of those sent back earlier.

	Internal ATI Slash / Filter Commands, not in the manual (Apx A)
	/SM	Median Filter
	/SA	Averaging Filter
	The median filter may work better to remove spiking and Gary
	recommended going to it instead of the normal 'averaging mode'.

	Signal Settings:
	The actual trigger threshold is at 2.5V
	We should not oversaturate the signal, as evidenced by the '2nd' peak
	getting all the way up to 12.  Gary said that it really is better for
	it to go anywhere from 8-12.  Absolute min. is 6V. 
	The filter cap also can be tweaked a bit to get the signal down.

	'W' component board can be attenuated more than the others, so a 'bad'
	or oversensative board can be put into that axis and tweaked down to
	help minimize spiking.

Major Changes to ATI Include:
	Set Sonic to /SM filter mode
	Swapped "W" board into 'U' slot
	Swapped "V" board from 'U' slot and put it in 'W' slot
	Adjusted signals as best i could.
	Zeroed more or less.  NOTE the chamber is really letting in too much
		wind .

Signal Quality
	I saw some of the 'phantom' noise signals on the "V" board in 'U' slot.
	These were generally less than 1V, but occassionally more.  The period
	was about the same as the trigger pulse.  The overall signal also was
	quite irregular but this may be attributed to wind getting through
	the zero chamber.   This board was pretty easy to adjust when it got
	to the 'W' slot, and the Signal looked a bit jumpy.

	I had a devil of a time adjusting the "W" board once it got into the
	'U' slot and played around quite a bit.   I still saw phantom signals
	on it, but i think i finally adjusted them out.  EVE was still getting
	spike hits though, but never > 1%

Conclusion:
	I believe we'll still see spikes in the 'U' data overnight even
	with the "W" board there.   That board had been spiking 2 days ago
	until i adjusted it in it's original 'W' slot.
	Need to watch both...

	Cnt	U   V   W	   
15:44				Swapped "W" into U.  "V" into 'W' slot
				Seeing lots of junk on W as well.
15:46	948	1   0   0		
15:48	2038	2   0   0	Attenuated "W" pot CCW.  2nd peak is now 8-10V
				Now the ghosts seem to be gone.  But i'm still
				seeing spike hits on EVE and the signal itself
				looks crummy.
15:51				Setting "V" board in 'W' slot.
				2nd peak is 8-10V.
				Zeroing "V" board in W,  34-deg, 27%
				Should be OK
16:20	2718	3   0   0	"W" board in 'U' slot.  Still seeing jumps
				in 1st peak up to 1.5V 
				and a bit of a 'phantom'
16:24	2692	7   0   0	Still adjusting signal.
16:40				Finally throwing up my hands, have done the
				best adjustment on the "W" board in 'U' slot i
				can.  Zeroed, etc.

Unbelievable!
Judging from the overnight plots, the sonic spiking at station 3 has greatly
improved to almost having nothing at all....wow.

So now, was it the /SM median filter mode, or the board swap, or the board
signal adjustment, or the big question: will it last?

  Sonic spikes were down through out the night. Station 1 and
2 look very good with no or little activity. Station 3 has some
spikes on U. We did a new board today from ATI. I want to
monitor station 3 before changing the board(s).

  Sonics at stations 1 and 2 look good. Station 3 "u" is
spiking. It may be time to replace the driver board.

  Station vist around 20:00 - 20:45 GMT

I chnaged the driver board for the "u" axis at station 3.
The board was adjusted, with zero chamber on, such that the peak
of the first wave was below .250 mv and the second wave was around 
9 volts. I did a zero calibration on all 3 vectors.

   visit time:  15:00 - 16:00 GMT

 Spent time working on the "u" vector. After checking the
signals using the scope and the zero chamber I decided to change
the transducers because the trailing end of the received signal
was dropping off quickly. The length of the signal improved some.
The sonic was zeroed. Will monitor.

	visit time:  22:30 - 23:00 GMT

  I decided to try something different at station 3. So
I tuned down the gain so in light winds the peak of the second
pulse was between 5 and 6 volts. I monitored the spike count in
EVE for about ten minutes. The average count was < 3 for vector
U. The others were zero. Maybe this will be a solution.

  The sonic continues to be a problem. There were
a number of spikes on "u" throughout the night. In addition
"w" went down at 9:00 GMT this morning. I guess I'll just
spend the day at site 3!

		visit time: 17:30 - 18:30 GMT

  The first thing was to look at the "w" vector. I hooked up
the scope and the signal looked fine. The peak level of the
2nd pulse was above 8 volts. Using EVE, it was seeing spikes
at a 10hz rate!  So I reset EVE. The sonic response, thru EVE,
now looked fine. So something happened in the EVE firmware.

  With respect to the "u" vector, I turned the gain back up
so the second pulse was above 8 volts. This was done with winds
around 3 m/s. I then spent the next 45 minutes monitoring the spike
count thru EVE. Over this time period I did not see a 5 minute
count greater than 3. 

  At this point I decided not to play with anything and let the
system run for another 24 hours.

		visit time:  15:45 - 16:45 GMT
 
  The sonic, at station 2, was moved to a height of
5.67 meters. It is now ~4m above crop height.
The sonic was re-zeroed. I will go out later to shoot
the angle since the boom is now mounted on the circular
mast section.

 New angle: 213 degrees 58' 25"

  The u vector is looking much better today. There were
a number of spikes in the very early morning.

		visit time: 19:40 - 20:00 GMT

  Replaced the driver board for the u vector. The
board removed was the new one I got from ATI. Will monitor
performance. The majority of spikes are below 0.01; however,
this is the only vector showing any problems. It would be
nice to find the source of the problem.
NOTE: the u vector was zeroed.

		visit time: 17:45 - 18:15 GMT

  Switched driver board "u" with driver board "v" to see
if problem on "u" vector was board related. Both "u" and
"v" werer zeroed.

		visit time: 16:30 - 17:00 GMT

  I decided to continue searching for the problem on the
"u" component. There was no major improvements after
switching the "u" and "v" driver boards. So I switched the
transducers. All three axis were zeroed.

		visit time: 18:30 - 19:30 GMT

  The sonic at station 1 was moved to 5.7m. Due to
cloudy conditions the angle of the boom will be done
tomorrow. The boom level will also be adjusted tomorrow
if the winds are down.

  The new boom angle for station 1 sonic is:
224 degree 51' 12"

The cahnge took place yesterday at 19:00 GMT.

		visit time: 18:45 - 19:15 GMT

  Leveled sonic after moving it yesterday. It was
too windy yesterday. I was not able to get alevel
within 0.1 because I ran out of turn-buckle.

		visit time: 20:00 - 21:30 GMT

  I switched the electronic boxes for the sonics at sites 2 and 3.
At site 3 the sonic was adjusted and zeroed. At 2 only adjustments
were made since the sonic is at 5.7m. The stations will be monitored.

  Based on the sonic flag plot, the electronic box is where
the problem is. The "u" vector is now spiking at 2 while
3 looks ok. I let it run through the night.

		visit time: 15:40 -15:50 GMT

 The sonic electronic box has been removed from site 2.
It is in for repairs at the base station. The sonic should
be backup later today.

		visit time: 19:30 - 20:30 GMT

 After checking all the signal lines for conductivity I installed
the sonic electronic box back out at site 2. There was nothing
questionable about the connections from the connector to the
backplane. The driver boards were adjusted but not zeroed. Hope
to zero it later today.

  Sonic at station 1 continues to behave. At station 2 we are
getting the usual spikes on the "u" vector. Station 3 now has some
spikes on "w". This is similar to what we were seeing at station 2
before we changed boxes. This problem may exist in both boxes just
on different vectors.

		visit time: 16:30 - 17:30 GMT

  Tower was lowered and sonic zeroed. Steve O. looked at the
signal levels and tweaked them abit. We are still getting
some spikes.

  Since the change in sonic electronic boxes, station 3 "w"
component has had spikes. It appears these spikes are related
to temperature. The spikes occur during the day time when temperatures
start getting over 25C.

		visit time: 15:00 - 16:00 GMT

  Based on the last few days, the sonic at 3 has been spiking
due to changes in temperature (refer to earlier logbook entry).
This may not have been noticed at station 2, when the electronic
box was there, because the box was shaded by the corn. We moved
the sonic such that it is now on the ground under the PAM electronic
box. A 2' x 2' metal plate was put on top of the box for added
shielding. The sonic was zeroed on all axis.

Due to a typo in sonic.config.S, the boom azimuth change on July 3rd of the
pam 2 sonic was not used in the the daily plots.

The S code looked like: 

   if (t2 > utime("1996 jul 3 16:30"))
        booazm <- c(boomazm[1], 214.0, boomazm[3])

booazm was being set, instead of boomazm. As a result the azimuth change from
215.2 to 214.0 was not corrected for in the dailyplots from July 3rd to 
July 9th.

The problem was noted and corrected on July 10th.

  We are still getting spikes on the "w" component at
station 3. The events are correlated with the temperature
eventhough we moved the box and added a sun shield. We still
have a spare driver card so we might try that tomorrow.

I've plotted tilts over the last 10 days and get:

site	offset	pitch	roll	lean	phase
	cm/s	deg	deg	deg	deg

PAM1	-4	-0.4	1.2	1.3	70.4
PAM2	-3	0.4	0.6	0.7	126.1
PAM3	-5	1.5	-0.3	1.6	-169.0

Note that the PAM2 plot was pretty scattered, probably because we
changed the tilt after zeroing it Saturday.

All of these values are acceptable to me.

		visit time: 17:00 - 17:30 GMT

  Driver boards "u" and "w" were switched. The pots
were adjusted and "u" and "w" were zeroed.

  After the mast was lowered for cleaning of the krypton
and BPH the angel of the sonic boom was checked. Two measurements
were made.

224 degrees  51'  40"
225 degrees  21'  30"

These are very close to the last measurement made when
the sonic was moved up the tower.

  The spikes on tc are continuing eventhough ther are
not any spikes on u, v, or w. It may be time to power down
and re-start EVE. The last time the key was turned to reset not
powered down. Maybe reset is not hooked up.

I just computed tilts for the last week, which look similar to the previous
week's values.  PAM3 is now "leaning" by 1.8 degrees, which is enough
that we will try to level it tomorrow.

Got the sonic at PAM1 going by vigorously shaking the tower (to try to
get the drop on the w bottom transducer off) and then cycling power.

I'll do this at PAM3 next, but will wait on PAM2 since it is so wet and
muddy.

I got the ati at PAM3 running also.  I knocked THE drop of water off 
and cycled power about 2045 Z.

P.S. PAM2 apparently revived itself about 2130.

We finally moved the sonic at PAM3 to 4.00m.  We didn't measure this new
distance, but did move it up by 0.80m from the previous location.  This
is about 3.7m above the canopy to the south and 3.5m above the canopy to
the north, so it should be okay for a while.

I've looked at the tilts for the last week, and they are pretty similar
to the two previous weeks I've looked at (see previous comments).  This
is somewhat surprising since the PAM3 sonic has been moved recently.
The final tilt corrections should take the moves into account.

P.S. note that all the sonics spiked a bit in the rain overnight.

station 3 had the sonic despiker stuck for the period of jd 215 from
10:00Z to jd 216 05:00Z.

The sonic data for the three sites were examined in order to ascertain
the time intervals over which to integrate data to define the pitch.

Three primary sources of information were considered:
    project long plots of fun.var.plot("alev.atik.3m.pam1/2/3")
    project long plots of tilt <- fun.sonic.tilt.summary("atik.3m.pam1/2/3")
       This function plotted w offset, lean, and azimuth as three panels
    logbook. The logbook was  search to account for specific details indicated
       by features exhibited by the plots.
       
 PAM1
      Averaging period   Start               Stop              Interval
               A         00:00, 18 Jun       22:00, 27 Jun     09 day 22 hr
               B         22:00, 27 Jun       18:00, 09 Jul     11 day 20 hr
	       C         18:00, 09 Jul       17:00, 19 Jul     09 day 23 hr
	       D         17:00, 19 Jul       21:00, 02 Aug     14 day 04 hr
	       E         21:00, 2 Aug        14:00, 23 Aug     20 day 17 hr
                                                               66 day 14 hr
 PAM2
      Averaging period   Start               Stop              Interval
               A         00:00, 18 Jun       14:00, 03 Jul     15 day 14 hr
               B         14:00, 03 Jul       21:00, 10 Jul     07 day 07 hr
	       C         21:00, 10 Jul       16:00, 13 Jul     02 day 19 hr
	       D         16:00, 13 Jul       14:00, 23 Aug     40 day 22 hr
							       66 day 14 hr
 PAM3
      Averaging period   Start               Stop              Interval
               A         00:00, 18 Jun       00:00, 20 Jul     32 day 00 hr
               B         00:00, 20 Jul       16:00, 26 Jul     06 day 16 hr
               C         16:00, 26 Jul       15:00, 04 Aug     08 day 23 hr
	       D         15:00, 04 Aug       12:00, 17 Aug     12 day 21 hr
       Bad data period   12:00, 17 Aug       15:00  19 Aug     02 day 03 hr
	       F         15:00, 19 Aug       14:00, 23 Aug     03 day 23 hr
                                                               66 day 14 hr

The logbook was searched to find the sonic anemometer angles.
Note that the first set of angle were given for the cross axis of the sonic
anemometer (= 131 23' 00", 125 10' 48", and 140 11' 50"). 90 degrees was added

 PAM1
       Angle             Start               Stop
       221 23' 00"       00:00, 18 Jun       18:00, 09 Jul
       224 51' 12"       18:00, 09 Jul       14:00, 23 Aug  

 PAM2
       Angle             Start               Stop
       215 10' 48"       00:00, 18 Jun       14:00, 03 Jul
       213 58'25"        14:00, 03 Jul       14:00, 23 Aug  

 PAM3
      Angle              Start               Stop
      230 11' 50"        00:00, 18 Jun       14:00, 23 Aug  

The logbook was searched to find the sonic anemometer heights.

 PAM1
       Height            Start               Stop
       3.20 m            00:00, 18 Jun       22:00, 27 Jun
       4.50 m            22:00, 27 Jun       18:00, 09 Jul
       5.70 m            18:00, 09 Jul       14:00, 23 Aug  

 PAM2
       Height            Start               Stop
       4.54 m            00:00, 18 Jun       14:00, 03 Jul
       5.67 m            14:00, 03 Jul       14:00, 23 Aug  

 PAM3
       Height             Start               Stop
       3.21 m            00:00, 18 Jun        16:00, 26 Jul  
       4.00 m            16:00, 26 Jul        14:00, 23 Aug  

The sonic data for each of the three sites were separated into several separate
intervals, following the analysis given in Logbook entry #287.
The tilt program:
              fun.sonic.tilt.plot("atik.3m.pam1/2/3")
was then operated upon each sonic data period and the coefficients calculated. 

       
  PAM1
  Start          Stop           Interval       woffset     lean     leanaz
A 00:00,18 Jun   22:00,27 Jun   09 day 22 hr   -0.038      1.605     57.7
B 22:00,27 Jun   18:00,09 Jul   11 day 20 hr   -0.039      1.484     50.4
C 18:00,09 Jul   17:00,19 Jul   09 day 23 hr   -0.031      1.178     69.3
D 17:00,19 Jul   21:00,02 Aug   14 day 04 hr   -0.030      1.041     67.9
E 21:00,02 Aug   14:00,23 Aug   20 day 17 hr   -0.033      0.515    110.2

  PAM2
  Start          Stop           Interval       woffset     lean     leanaz
A 00:00,18 Jun   14:00,03 Jul   15 day 14 hr   -0.015      0.932     52.1
B 14:00,03 Jul   21:00,10 Jul   07 day 07 hr   -0.067      0.531    149.5
C 21:00,10 Jul   16:00,13 Jul   02 day 19 hr   -0.006      0.688    125.4
D 16:00,13 Jul   14:00,23 Aug   40 day 22 hr   -0.008      0.696    117.4

  PAM3
  Start          Stop           Interval       woffset     lean     leanaz
A 00:00,18 Jun   00:00,20 Jul   32 day 00 hr   -0.038      1.413   -173.0
B 00:00,20 Jul   16:00,26 Jul   06 day 16 hr   -0.069      1.605   -164.0
C 16:00,26 Jul   15:00,04 Aug   08 day 23 hr   -0.057      1.683   -171.7
D 15:00,04 Aug   12:00,17 Aug   12 day 21 hr   -0.042      1.561   -161.0
E 12:00,17 Aug   15:00,19 Aug   02 day 03 hr         bad data period
F 15:00,19 Aug   14:00,23 Aug   03 day 23 hr   -0.051      2.979    154.5