REAP

I've made a sketch of the site based on Eric's excel doc. The graffle is next to this png in svn:
https://code.ecoinformatics.org/code/reap/trunk/usecases/terrestrial/design/mclaughlin_nutnet_reap_layout.png

We're planning to instrument 2 reps of 4 treaments. It's unlikely the 3rd rep will be instrumented (with station positioned as it is now to the North, this means the Southern rep).

We're planning to run cables for light bars and soil moisture (cs616) in pvc down the aisles (instead of diagonals) to minimize disturbance. Cables are expensive, station should be located to minimize cost.

Short Cut does not allow multiple sensor types to go into the AM16/32 multiplexer. We want to use it with 8 cs616 and 8 light bars. I'm continuing to talk with Campbell to verify the configuration is possible before I place the order. I'm also trying to see if we can get differential measurements of the light bars.

Using Short Cut I've discovered the multiplexer can be wired to accept the cs616 control lines, this is a relief. All 4 logger control ports will be used with this configuration.

To power all 8 cs616s we'll need to buy a busbar from radio shack for about $5.

With the multiplexer in our enclosure, we'll be about out of space. We'll probably need a bigger enclosure if we later want to insert a computer.

Campbell contact has mux w/ simulated cs616s and light bars working in CRBasic. He hopes to send me the program by end of day. I'll then add our other sensors to double check we're not short on any ports. After this I can place the orders.

Submitted order in form of 2 MSI preliminary requisition forms to Mark S. and MSI Purchasing.

According to the form, each of these will have to go through campus purchasing (instead of MSI), which may take 1-2 weeks to process instead of 2 days. This and the McLaughlin schedule means we may have to wait until late Jan for deployment.

Received Campbell Sales Order Acknowledgment. Est Ship Date: 01-06-08.

For the Campbell order, MSI sends an update: "estimated time of arrival for your order is 12/29/08".

Apogee sends ETA for their order: 1/22/2009

UPS sends update: Campbell order scheduled for delivery 12 Jan. It's in Salt Lake City today. 43 lbs.

Campbell order has arrived. Apogee order should arrive by Wed or Thurs next week.

Apogee order arrived yesterday, so we now have both orders. Things left to buy include pvc, couplings, and a dc power bar.

I'm now trying to coordinate schedules for the deployment, mainly the two reserves.

A tentative plan is Feb 10 - 13.
A better plan gives more time, e.g. Feb 15 - 20.

Overview (not necessarily in this order, maybe get Seabloom lab to do some things beforehand):

- fly to portland, rent van, drive 1hr to Baskett, run tests to determine problem(s) with PAR, record last minute station metadata, disassemble station, retrieve computer, radio, etc from usfw office, drive 1hr to OSU, retrieve new campbell and apogee gear

- buy supplies (power bar, pvc, food)

- test new setup - wire up new gear, reprogram datalogger, check values

- drive 10hrs to McLaughlin

- deploy, test radio and internet links, check data values

- drive 8hrs home

I'm scheduling for week of Feb 15th.

Apogee order arrived some time ago.

Feb 16 is a holiday, Baskett office is closed. I can still disassemble the station this day, but have asked Eric to retrieve the computer and radio from inside in advance.

- buy fish tape
- determine pvc lengths and layout

map of site:
http://www.google.com/maps/ms?ie=UTF8&t=h&msa=0&msid=112440998775831561410.000459b3fa51ddac74e41&ll=38.869652,-122.419925&spn=0.017976,0.034161&z=15

Flying to PDX morn of Feb 15. Rented a 10' Budget truck to drive south, a bit obnoxious, but the cheapest soln for a one-way, and will offer shelter from rain.

Station has been deployed to McLaughlin.
There are many loose ends and improvements I would like (not listed in this entry) to tie up and make; once I have this list in order, and have discussed it with Paul, Cathy and Eric, I may request a second trip out sometime soon. Paul and Cathy have offered to take care of some of these for us.

Deployment went roughly as follows.

Day1 - Feb 15, Sun
---
- Fly to PDX, cab to Budget, drive truck to Baskett.
- Remove cable ties, pull up staples and cables. Put yellow tape on uphill soil moisture and light bar.
- Downhill light bar cable has been chewed through. Downhill soil moisture has been gnawed into, bare wires show.
- Buy supplies, stay night in Dallas.

Day2 - Feb 16, Mon
---
- Disassemble station and fence, carry down hill, load into truck.
- Drive to OSU, unload in lab w/ Eric.
- Work in lab, staying with Eric and Elizabeth from here on.

Day 3 - Feb 17, Tue
---
- Work in lab
- Discover uphill light bar also bitten into, rough testing reveals it's bad, will not deploy it.
- Discover amplified bleed-over when sampling empty or open circuit light bar channels. I.e. a gnawed through PAR sensor cable may be difficult to see from the signal, when the circuit opens, the signal becomes a slightly amplified reading of the prior channel. This behavior is currently predictable and repeatable for our logger layout.
- Discuss this and other issues at length w/ Campbell, Jed and Mark. Bleed due to channels input resistance. Mark notes I should utilize open circuit detection. Testing. Feature not working. Software and logger-firmware upgrades, and it works. Feature (e.g. argument 'mv2500C') will now report a very high number if open circuit.
- Setting up and testing MUX and program.
- Modifications to program. Can't use C4 for cs215 despite Campbell's recommendation. Find MUX RES works on C4, freeing C1 for cs215.
- Discover problem with program that ran Baskett. QSOSUN was wired to SE3 but never read. QSOSUN was simply a slightly delayed reading of light bar on SE2. Morale falls. The Mystery of Assumed Bird Waste Related Signal Flip Flop solved.
- Finish McLaughlin program. Scanning every 5 min, sampling every 10.

Day 3 - Feb 18, Wed
---
- Load truck
- Buy and load pvc and supplies from Searing Elect/Plumbing and Home Depot
- Drive to McLaughlin. Arrive 10:30pm.

Day 4 - Feb 19, Thu
---
- Meet with Paul and Cathy, check field station roof for omni, drive to nutnet site and unload, begin work
- Meet Quinn at field station. Mount omni on NE roof-corner reusing their old omni mount. Run antenna cable into Paul's office, put box on newly drilled hole, drive ground rod.
- Go to site, prelim station setup, lay out pvc for sampling to take place in NW corner of all plots. Missing T-junctions, cut holes for close plots.
- I go to field station, radio shot test, Quinn aligns, works.
- Fish cables.

Day 5 - Feb 20, Fri
---
- Verify with Eric - "good, just avoid 1m sampling area within plots"
- Determine layout - roughly 60cm * 60cm from NW corners, light bar West to East, soil moisture to the North (for shadow). Put in all non-fence plot sensors.
- Do fenced plots. Cut fence, cut 1'' hardware cloth patches, apply and zip tie patchs, run in sensors, mend patches with zip ties maintaining 1'' integrity. Snip metal fencepost cable ties (not fence!), bend and carefully enter, measure and insert sensors, exit. Mend fence corners with zip ties.
- Pvc putty seal all conduit openings.
- Labeled all 14 cables by plot number and date, and a few of the other sensors (already labeled). Weatherproof Radioshack labels suck, label a second time with label printed number stickers from Paul. West cables are also marked with 1 ring Red tape, furthest west with 2 rings of Red, likewise Blue on East.
- Look at possible junction box from Paul, determine against.
- Move station a bit to the west-side of plot 28, remove 2 conduits from the west-side runs to give needed cable slack. (west half is bigger?). Legs now line up w/ conduit openings for runs up legs.

Day 6 - Feb 21, Sat, racing the rain
---
- wire mux. too messy. mount and rewire
- no room for surge suppressor away from logger or mux, determine let it dangle.
- wire busbar
- dealing with ground cables - drive rod, some wire, go to shop where Rhett crimps and heat-seals 2 ring connectors on my cables.
- tape omni connector on roof
- test radio shot. works.
- drive tripod feet rods, bundle and zip tie excess sensor cable to legs, putty enclosure holes, roughly bubble level qsosun.
- back at field station, further radio shot testing. works. View => Station Status never seems to work, however.
- Discuss computer (dogbert) and router with Paul. Verify their ip is static. Setting up router (d-link dgl-4100) for external access. Auto-detection of mclaughlin static ip does not work. Manually statically set this ip in dyndns client. Setup our computer with reserved dhcp ip address (check with me for internal address). ssh to dogbert from nibbler(nceas) works. rsync works. VNC from remote works.

Day 7 - Feb 22, Sun
---
- Heavier rain begins at ~7am, roughly coinciding with data fetch over radio having trouble.
- Analyze data. Soil sensors seemingly not responding to rain, rain gage signal seems a bit low.
- Go to site. Check all conduit bell ends tight, all sealed, enclosure sealed and fine. Can't line up yagi, rain and mist, visibility is low. Yagi looks like it's a bit too low, but don't try to adjust in rain on ladder.
- Check inside rain gage, it seems fine. Seal bucket on with putty.
- Additional staples on non-fenced plot bars. More zip ties.
- Return to field station and test radio shots. Many fail, many work.
- Sort truck, return borrowed items, leave fencing and extra conduit. Final chat w/ Paul and Cathy
- Talk with Eric, soil is thick and slow signal change is expected.
- Depart. Drop off borrowed sawzall, drill and ladder to Quinn in Davis. Drive to Pismo, stay the night.

Day 8 - Feb 23, Mon
---
- Drive to uscb, unload truck.
- Return truck, bus to airport, fetch car.
- UCSB - begin testing radio shot, can connect with Connect button, but cannot get data.

It should be noted that plot 28, a fenced plot, has not been instrumented. As both light bars were munched on, we were short a light bar (open circuit is being sampled and giving a visible high 2000 reading). One soil moisture was munched on, but the other was ok. This one is in air zip-tied to the tripod (giving a NAN), ready to go when we get another light bar.

Correction to last comment: the soil reflectometer that's not inserted, the one to be put into plot 28, is giving a low number (currently 0.005), not a NAN.

Paul was kind enough to put up our station fence today, and try to better align
the yagi. He said it only seemed slightly off.

Unfortunately I'm seeing the same behavior as earlier this afternoon - I can
only seem to occasionally grab <= 10 records at a time. The rest of the time,
or if I try to grab more, I typically see "transaction failure" "timed out or
resource error" "logger query serial framing error" messages. I'm looking into
this further...

Eric would like to see webplots of the new data. I've renamed the reap01
(Baskett) datasource (which was still named after my office) and inserted
reap02 (McL). The dataturbine workflows need to be updated according, and
workflows that create the data for web plots need to be modified/created and
run.

Have started a new collection schedule this morning (every 10m, with 3 retries 2m apart, no secondary retries) that begins now and only tries to get that last 1 new record, small enough to consistently work (so far at least) with current flaky link. If comms go down for a period of time, on resume this schedule will not fill in any gaps, but it will also not become hopelessly behind and unable to collect any records. Gaps can be filled in manually as needed.
I'll manually catch up the older data, and then splice the two files together into one before svn insertion. Cursory looks at battery voltage yesterday and today show no big drain despite much radio play (so I think this new more aggressive collection interval will be fine). Will re-check voltage in a day or so.

Battery voltage looks good with this collection interval, and I haven't noticed any data gaps.

- emailed Apogee to see about recabling fee for light bars. We need a bar so we can instrument plot 28.

- Shawn Gerrity, Seabloom's new lab manager, will visit in coming weeks. I've asked him if he can rodent proof remaining exposed cables (in-non-fenced-plots and tripod-feet), take photos, try vertically aligning our yagi.

Shawn Gerrity is at the site today for fertilization. We've reoriented the yagi vertically, but so far the radio link is of the same quality as before the change -- still problematic with large transfers. I'll look into this in detail and fix it on my next visit. Shawn will also put some protective tubing on our cable ends, take some photos, get gps coords for the tripod, and make sure the sensors are not in the destructive sampling areas.

In doing some of the tests today, I've introduced at lease two duplicate records in the data-file. I plan to leave these in, and ensure that the DataTurbine source program properly deals with duplicates.

Yesterday I finished catching up the data-gap from before the new schedule was implemented. I also found a large number (though small relative to the amount of records) of missing records, mostly just sets of 1 here and there. I collected all these, so that we can know when in analysis that our first 15 days or so are clean. I'm not planning to do this from here on unless requested, dealing with McLaughlin gaps can be left to the DT actor or other sink (and this will likely become moot if I visit and improve the radio link, making gaps infrequent/non-existent).

Code was added to DatToRBNB (our DT Source) to check for and skip duplicate records. Checking for and warning about gaps would be nice to add in the future.
Code to deal with records that contain NANs needs to be added in the future.

reap02 dataturbine channelmap was refreshed with the new data set-- 15 days of clean McL data available in Kepler for scrutiny.

Eric is ok with instrumenting plot 28 in mid-April.
Apogee recalibration fee is $50 a unit (plus cable lengths I imagine).
I'll try to get our two light bars shipped pronto.
Once Eric is free we'll analyze the data to see if things look good before I go back.

Shipped light bars. Kepler took over, and "pronto" became not very.

They're getting repaired now, and I've put 30m on each, for a total of $150.

Received the 2 fixed light bars (here at UCSB).

Changing this bug to an enhancement request, the station was deployed, work has begun on the engineering view.

Site continues to collect data and power looks good. I/we may make another trip to the site -- in which case the deployment todos implied or listed in this bug should be gone over -- but only after much more software development and hardware testing in the lab for the engineering view.

Original Bugzilla ID was 3650