nwfsc-fram / oceants Goto Github PK

failing file:

2016 > MsJulie > Leg 5 > STBD_CTD7537_DO2976CT3447Op1549_Hauls43to47_2Sept2016.csv

caused by issue #35 where an extra, partially populated, row is being added to the last line of the csv output file from the conversion process

bad file:

2016 > Last Straw > PORT_CTD6367_DO1671CT1788Op833_Hauls62to64_6June2016.csv

example bad file:

2016 > MsJulie > Let 5 > STBD_CTD7537_DO2976CT3447Op1549_Hauls43to47_2Sept2016.hex

Currently when I cannot match a cast to a haul from the data warehouse, I skip adding the following values for the given cast

• depth
• latitude
• longitude
• haulID

I could just use the latitude and longitude from the immediately preceding haul instead as an approximation.

In the example *.raw files that I have for the SBE 19plusV2 from the trawl survey, neither of these flags are set and so I have it hardcoded in to just skip them should they appear in a subsequent *.raw file. Instead, I should probably handle when they are missing or when they appear and are either flagged as yes or no. See parseHex.ts: 190-191.

(node:73880) UnhandledPromiseRejectionWarning: Error: Failed to bind Col "Temperature A/D Counts"
at Col.bind (C:\Todd.Hay\Code\OceanTS\node_modules\apache-arrow\compute\compute\predicate.ts:77:42)
at C:\Todd.Hay\Code\OceanTS\src\sbe19plusV2\equations\temperature.ts:23:26

When the source *.raw file does not properly contain the header information and so the conversion fails. Example file:

2016 > LastStraw > Leg 2 > Hauls 62 to 64

References of interest:

https://github.com/vega/vega-loader-arrow

https://observablehq.com/@vega/vega-and-apache-arrow

https://observablehq.com/@vega/apache-arrow-in-vega-lite

https://vega.github.io/vega/

• Reading
• Appending columns to existing Apache Arrow Table
• Writing

Convert the raw, decimal results over to engineering, decimal results.

• Temperature
• Pressure
• Conductivity
• Salinity
• Oxygen, SBE43
• Fluorescence
• Turbidity
• Oxygen, OPTODE Aanderaa
• Depth

columns include:

• latitude
• longitude
• haulID
• dateTime

See here for details:

apache/arrow#1343 (comment)

Review attached error.log file for the casts of interest

Have inquired with Victor and Seabird, awaiting response. However, I would note that for the 2018 7738 xmlcon file, the OPTODE Aanderaa calibration coefficients are all 0's, as follows, so I can probably just proceed without using them:

  <Sensor index="6" SensorID="64" >
    <OptodeOxygenAanderaa SensorID="64" >
      <SerialNumber></SerialNumber>
      <CalibrationDate></CalibrationDate>
      <DoSalinityCorrection>0</DoSalinityCorrection>
      <DoDepthCorrection>0</DoDepthCorrection>
      <InternalSalinity>0.00000e+000</InternalSalinity>
    </OptodeOxygenAanderaa>
  </Sensor>

References:

https://blog.logrocket.com/node-js-multithreading-what-are-worker-threads-and-why-do-they-matter-48ab102f8b10

https://www.npmjs.com/package/worker-farm

csv output from the main hex > csv processing appear to give incorrect haul and/or start/end date/time stamps. For instance, see haul 201603008002

This in particular is happening with the OPTODE Aanderaa Oxygen sensor in 2017, Last Straw with CTD 7738 where we are missing the calibration coefficients in the xmlcon file for this particular sensor.

For example:

https://codeburst.io/multi-core-nodejs-app-is-it-possible-in-a-single-thread-framework-fe4feaf67f90

Cluster module:

https://www.npmjs.com/package/cluster

Sometimes no hex data is actually recorded in the hex file. Stop subsequent processing of the file when this scenario occurs. See the attached hex file as an example.