Blog

This will be my template for this round of soil sample processing.

A few notes:

- We forgot to pack the scale, so ours is being sent.  In the meantime, I've borrowed one (generously loaned by John H) with a resolution of 1g.  When our scale arrives, I'll make a few parallel measurements to compare the two scales.

- We couldn't find any good foil cups at the store today to bake the dirt in, so I've made some by bending aluminum foil.  Obviously, they won't tare as consistently as store-bought.

- The coring tool was set up with two 3cm rings on the top (used) followed by two 1cm rings on the bottom (ignored).  Thus, each 3cm sample volume was 3*pi*(5.31/2)^2=66.4cm^3.

# = weighed wet a day later (but kept cold to reduce water loss)

* = weight after a bit of time in the oven :( [forgot to write down first reading], water content will be a bit low, but dry density will be okay

Here is a quick comparison of the weights of various objects with the borrowed scale and ours.  The borrowed scale read within 1g, so I wouldn't change any of the above readings.

ROUND TWO:

Warmer temperatures have both melted snow (and there has been rain) to increase moisture and liquified existing moisture to allow the EC-5s to measure it.

ROUND THREE:

John, Tom, and Steve S

We tared the scale prior to weighing the wet soil (net), weighed the dry soil + pan (total), then subtracted the pan wt to get the dry soil (net).

I mistakenly started weighing the soil in ounces, so we continued this in order to make a consistent table.

Ring diameter = 2 3/32" = 2.094" = 5.319 cm; volume = 66.66 cm^3;  1 oz = 28.35 gm

Qsoil = 1 - dry/wet; h2o volume fraction = (wet-dry)/(1 gm per cm^3)/sample-volume

P.S. (18 Apr 2011): All calculations seemed off in Round 3, reran and edited above entries.

on site: SteveO and Ling Jan 4, 2011

1> The primary purpose of this visit was to fix/replace the Qsoil sensor.  We briefly disconnected the wrong sensor when trying to figure out which cable was Qsoil.  As I started to unbury it from the mote end, the PIC enclosure pulled out of the ground, leaving the sensor and its cable still in place.  I exposed the cable connector, cleaned both ends, and reconnected the PIC.  Readings were then 22.6%, rather than 615% when we first looked at data.  Thus this was a simple connector problem -- the original sensor was never disturbed and the PIC didn't require new calibration coefficients.  I wrapped a bit of tape around the connector before reburying.  I'm really sorry if this was my fault during installation, though I noticed this cable was separate from the others, so perhaps someone else had replaced this sensor earlier?  Also, I think the connector was slightly deformed.  (3:57-4:07pm)

2> clean up the radiometers (4:09-4:15pm) Notice that the sensors are generally dirtier than other sites; recommend to clean up the sensor every visit.

3> take soil samples (4:07-4:15pm), marked as 1:0-3cm and 1:3-6cm (3-6cm sample may not be perfect,since a piece of brick was in the sample tube -- explaining why taking this core took a bit of effort!  I would still use this measurement as typical of the "soil".)  Soil was damp and sticky.

4> shoot the boom , measuring 80 degrees from the back so that it should be 80+180=260.  I simply tried to center mast in the parts of the sonic "claws" that I could see (partially blocked by cables from the sonic electronics box).  (4:25pm)

On-site SteveO and Ling Jan 4, 2011

1> To test one theory of why Vdsm at this site decreases at a faster rate than the other sites, we replaced the charge controller with one of the new ones sent by Chris.  We turned off the dsm during this process (3:01pm to 3:10pm).  Batteries were at 14.6V and solar input was over 20V.

2> slightly stirred rain gauge, small ice chunks left over from yesterday are still floating in the "antifreeze" (3:11pm)

3> re-seated several connections to the router's DC-DC converter (3:05pm)

4> briefly inspected that radiometers were still clean

see subject.

Big flakes coming down at 7:05, stopped by 7:35, restarted 7:45.

Also at base trailer in the morning.

Joe had noted that the trailer computers appeared to be sick during the break, which he attributes to problems related to computer crashes in Boulder.  When we arrived this morning, we rebooted both "flux" (which had a blank screen that wouldn't react to keystrokes) and "pam" (I don't remember why I rebooted it).  "aster" seemed to be fine and once "flux" was up, S+ worked fine.  Ling did restart dsm_server on "flux" and then was able to run cockpit using localhost (I think) as the server.  We found that printing to "canon" on "flux" and "cannon@pam" on "pam" worked.

"flux" also was just fine when we returned from the field this evening (and I have logged into it from the motel), so we have not yet been able to duplicate the failure.  (If we do, I'll first try ssh from my Mac on the same LAN.)

However, S+ wouldn't work on "pam".  After some sleuthing, I did "exportfs" and then "/etc/init.d/nfs restart" as root on "flux" followed by "mount /net/local" as root on "pam".  (I'm not sure that "exportfs" was necessary.)  I'm closer and "pam" can now find "/usr/local/bin/Splus7" (it couldn't before), but I still get an execution permission error.  Obviously, I can (and did) ssh from pam to flux to run S+...

On-site SteveO and Ling (Jan 4, 1011, 4:25pm)

1> clean all radiometers. same order and procedure as at hiland3.  Again, all domes looked pretty clean before and after, though I could imagine that there was a tiny frost film on the pyrgeometers. [my best guess from the data is that Rsw.in changed by 1 W/m2 out of 100 W/m2 and Rlw.in not at all.] (4:18PM)

2> replace barometer #2 with #10 (4:26pm), readings 869.56 (4:26pm)

3> take soil samples (4:36pm), marked as 2-0-3cm (full all the way to top) and 2-3-6cm on the sample boxes.  Again, were unable to process so far.

4> The precip gauge had about 10cm of power snow on top of a significant ice plug.  The top of the plug may be at the top of the fluid, but the powder definitely was above that.  I did some poking around to obtain the above information, but otherwise did nothing to the gauge, since it was clear at this site that precip had not yet been counted.  I like the idea of returning to this gauge and heating it, but am not yet convinced that the propane torch suggestion is the best (won't it will destroy the paint on the gauge?) and had not brought it along.  However, I don't yet have a better idea.  We do have extension cords that may be long enough to reach from the van to provide AC power, but the small inverter is only 150W and most sources of heat I can think of would be more than that.  SteveS had suggested talking to JohnH about the ETI gauges, which I have not yet done.  (4:53pm)

5> Shot the boom angle (again with the Datascope) and got 80.8 degrees. 

On  site-Highland3, SteveO and Ling  (Jan 3, 1011, 2:20-3:40pm)

We have done the following:

1> cleaned all radiometer domes.  Decided simply to breath on the domes and wipe off the condensation.  Order was Rsw.in Rlw.in Rsw.out Rlw.out.  The sun was shining during this period, so the effect of cleaning should be measurable.  ?[PS: have looked at the data.  Haze made the radiation somewhat variable, so my best estimate is that cleaning changed Rsw.in by 2--10 W/m2 out of 380 W/m2 and Rlw.in not at all.  Not <too> bad.]  All domes looked pretty clean before and after.  (2:20pm)

2> As per the task list, disinterred (at a cemetary!) the single Tsoil to check/correct its orientation.  Found that it was nearly horizontal at about 2cm depth.  Reinstalled it at an angle that I eyeballed to put the sensor from 1--6cm depth.  Before doing this it read 0.3 C, after 0.81 C.  Note that these temperatures are slightly above freezing.  I noted a thin (~1cm) crust of frozen soil, but underneath definitely wasn't, which all is consistent.  I should also note that there was about 10cm of snow covering the whole area that I attempted to replace when I finished.  Finally, I note that we first unplugged the TP01 while determining which wire was which.  Thus, its values reset to 0, but later appeared to be okay.  (2:32PM)

3> take soil samples (3pm), marked the two sample boxes as 1-0-3cm (5mm shame of being full)and 1-3-6cm.  Unfortunately, there are no scales at the base to weigh them with!!  We'll have to get one sent out :(

4> shoot the boom angle from across the street (3:23pm), readings are 254.2 (Steve's) 254.7 (Ling's).  We did not touch settings on the DataScope or recalibrate it.  I assumed this had been done earlier for this project.

5> break the ice in the rain-gauge (3:32PM),  and take the ice out of the rain-gauge.  The gauge was on the entire time, so these data will have to be edited.  I judged that, since the top of the ~10cm-thick ice plug was just at the top of the antifreeze/oil, this mass had been recorded by the gauge and that it was okay to remove it from the gauge.  In the process, some of the antifreeze splashed on the side (inside) of the gauge, so I wouldn't be surprised by a tip or after this operation as it (slowly) slides back to rejoin the rest of the fluid.  I should point out that there was absolutely no precip today in the area we could observe, so zeroing out the day should be fine.

6> take pictures (3:37pm)

Following the wisdom of Boulder, I have just commented out the crontab entries that turn the router power off at night on hiland3.  Enjoy the extra data and enjoy looking at the power consumption tonight.

TWH (1/4/11):

Highland Vdsm 1/2/11 at 19:52:30 = 12.80 V (Router turned off overnight)

     lowest Vdsm 1/3/11 at 07:52:30 = 12.11 V

                                             Vdsm loss = 0.69 V

                                 1/3/11 19:52:30 = 12.82 V (Router left on overnight)

     lowest Vdsm 1/4/11 at 07:52:30 = 12.10 V

                                             Vdsm loss = 0.72 V

On Thursday afternoon, JohnM added crontab entries to power this station's router down between 02Z and 14Z to conserve power.  Our intent is to do the same to the other stations if this works.

On Thursday night, the crontab executed, but the router stayed up.  Later that night, I logged into isfs3 and executed the command manually, but the router still stayed up.  We thus concluded that there was a faulty connection or jumper controlling the power.

However, on Friday night the crontab executed and the router went down.  Unfortunately, playa1's rsync took much longer than expected, so the rsync from hiland3 didn't start until after 02Z and thus failed.  Having not seen data coming in, JohnH (and later DaveW) were tasked with rebooting this station to revive it on Saturday morning.

On Saturday afternoon, I changed the crontab to turn off the station at 03Z to give it more time.

On Saturday night, the crontab again executed and rsync worked.

Ling and I have discussed some options.  We have several ideas:

1. rsync should run in parallel on all stations, rather than serially to better define the time window and also allow us to turn off all stations by 01Z.

2. even better would be to have the rsync script (started on porter?) execute the command to turn off the router once the rsync has finished.  The only reason no to do this would be if we wanted staff to be able to log into stations after ~01Z.

3. to fill in the night data, another the rsync script should run in the morning as well.  The evening rsync script would run faster since only half of the data would have to be transferred.

It would be best if Gordon would make these changes, since we think that the rsync script runs under crontab on his account.

From Dec 15 through Dec 19, Vdsm at Highland fell at a rate comparable to other sites.

On the night of Dec 20, Highland fell at a rate slightly faster than other sites.

On the night of Dec 21, Highland fell at a rate comparable to other sites, but cut out at 11.5 V

in the middle of the night.

On the nights of Dec 22 and 23, Highland cut out soon after sunset.

John replaced the Highland batteries on Dec 24 at 7:30 am. 

On the nights of Dec 26-27, Vdsm fell faster than other sites.

On the nights of Dec 28-30, Vdsm fell at a rate comparable to other sites.

On the nights of Jan 1-2 (with the router off), Vdsm fell much faster than at other sites.

Jan 2, ~14:45:  Joe tried cleaning the existing kh2o, SN 1395, with methanol but the voltage never exceeded 2 mv. 

Prior to cleaning, Joe noticed a smudge on the top window and the appearance of rust around the edges.

He replaced SN 1395 with SN 1390, which has an output of 2.7 V.   Unfortunately 1390 did not have the mounting

plate for the electronics in the case, so Joe mounted the electronics temporarily with cable ties.

9:30AM MST 1 Jan 2011

After cleaning in lake effect snow, solar panels were covered in snow, and there was no charging, so the station died. 

2 charged batteries were brought out from the base, and after clearing solar panels, batteries were replaced.

Station was running on departure. Horst was able to ssh into playa1 to confirm it was working.

Solar panels and radiometers cleared of up to 4 inches of accumulated snow. Downward looking SW radiometer had frost on it, the rest were clean after being brushed.

Approximately 9 total inches of snow were now at the site. Subtracting yesterday's 2.5, leaves an additional 6.5 inches of snow had fallen overnight.

1:20pm MST 31 Dec 2010

After battery replacement, hiland3 was not communicating, though it was recording data locally. 

(TWH: I don't think Highland's batteries were replaced.  John Horel recycled power this morning around 7 am,

which perversely stopped the station.)

Horst requested a power cycle of the DSM to restart the cellular modem. 

This was performed and at about 1:25 Horst was able to ssh into the station. 

12:18pm MST 31 Dec 2010

Snow cleared from solar panels at river7. Solar panles had a lot of ice from melting snow falling down. 

Radiometers were clear on arrival. 

Conditions were sunny (scattered clouds) while maintenance was performed. 

12:15pm MST 31 Dec 2010

Spare rad mote 8 has been replaced with new rad mote 17 sent from Militzer.

the new mote 17 was able to communicate with radio, and radiation data is now transmitting. Horst confirmed ability to check data via radio.

Mote 8 contains original data on its microSD card, and must be downloaded. The mote is now in the base trailer.