search

PNNL-TES / kpsr

Kaizad's Secret River incubation analysis
MIT License
0 stars 0 forks source link

Port assignments #3

Open bpbond opened 5 years ago

bpbond commented 5 years ago

Hi @kaizadp

So the data start with a long (two day) read on port 0, and there's another very long one on port 4 (see graphs below). Is this intended?

k1 k2

kaizadp commented 5 years ago

Hi @bpbond -- no, those were not intended, and I'm not sure what port 0 is. Is that background when no program is running?

I started my experiment on 5/2. Ran just a couple of ports initially, and then began using ports 1-5 and ambient port 16 around 5/2, 7:56 pm, which was 5/2 8:47 pm on the Picarro computer, but 5/3 1:47 am on the Picarro software. I wasn't sure how all these related, and I also wanted to see if/how the R-code would pull the required subset out, so I left it all in. In the valvemap file, I set the start_date_time to 5/2/19, 7:56 pm. Is that what's messing it up? I can go in and remove the long reads on 0 and 4.

bpbond commented 5 years ago

OK, thanks, that helps. I'll come back to this later.

I can go in and remove the long reads on 0 and 4.

No, no! Raw data is sacred. I just want to make sure I understand what's going on, and then we can handle it correctly.

bpbond commented 5 years ago

So there are 6,002 Picarro observations, but after filtering the dataset to drop ambient (valve 16) data, and limiting to first two minutes of data for each port, this drops to 383. Visualized by port and elapsed time:

plot

So we want to focus just on ports 1-5 per above?

kaizadp commented 5 years ago

That sounds about right. Yes, ports 1-5. Thanks!

bpbond commented 5 years ago

Another thing I'm seeing is relatively few measurements per sample. Just out of curiosity, did you change the Picarro settings in this regard? In most cases we only have 3-4 points in the first two minutes (see above), which is pretty much the minimum for computing a flux.

bpbond commented 5 years ago

Changing the max measurement time from 45 to 90 seconds gives us more data to work with, but still not much:

> summarydata
# A tibble: 12 x 10
   samplenum CO2_ppm_s CH4_ppb_s    r2_CO2    r2_CH4 max_co2_time DATETIME                N MPVPosition h2o_reported
       <int>     <dbl>     <dbl>     <dbl>     <dbl>        <dbl> <dttm>              <int>       <dbl>        <dbl>
 1         8    0.0892  -0.00623    0.816     0.816            83 2019-05-02 23:32:10     4           3        2.26 
 2        14   NA       NA          0         0                27 2019-05-03 00:01:57     1           3        0.718
 3        16    2.92   -11.1      NaN       NaN                49 2019-05-03 00:14:36     2           1        2.29 
 4        29    0.265    6.84       0.656     0.656            76 2019-05-03 01:27:46     3           1        0.705
 5        32    1.10   -11.9      NaN       NaN                47 2019-05-03 01:42:32     2           1        0.701
 6        46   NA       NA          0         0                29 2019-05-03 02:42:08     1           1        0.697
 7        60   NA       NA          0         0                 6 2019-05-03 03:06:53     1           3        2.52 
 8        95    0.612   -0.0372    -0.279    -0.279            51 2019-05-03 04:05:25     3           2        1.42 
 9       105   NA       NA          0         0                49 2019-05-03 04:20:46     1           3        2.79 
10       111   NA       NA          0         0                28 2019-05-03 04:29:14     1           1        2.31 
11       117   NA       NA          0         0                28 2019-05-03 04:42:19     1           1        2.31 
12       137    0.0297  -0.0250    -0.0791   -0.0791           54 2019-05-03 20:01:01     3           3        0.833

This was a test run I hope?

kaizadp commented 5 years ago

I set it to measure for 2 minutes, but I didn’t change the sampling interval within that. There is ambient measurement after each core, and I reduced that to 0.5 minute. It looks like the Picarro is measuring every 25-50 seconds, but not consistently. Is that normal? I could adjust those parameters next time.

kaizadp commented 5 years ago

Yes, this was a test run! I wanted to be sure I understood how to use the Picarro and the code before I began an experiment.

So we filter out the measurements with only 2 points. Is that why we have so few values in your summary table?

And I didn’t realize the measurement time was previously set to 45 seconds. That means I’ll have to set the Picarro to measure more frequently next time

bpbond commented 5 years ago

I have opened a "pull request" with script changes--this makes it easy for you to review and see what I changed. If it looks good, click "Merge".

https://github.com/PNNL-TES/kpsr/pull/4