Salinity data from all years through the WildEdit program, as suggested by the Seabird Electronics data processing manual, to eliminate erroneous data (data spikes). In Wild Edit standard deviations were set for pass one to 1, the standard deviations for pass two to 5, and set the scans per block to 100. The histograms of the data suggested that this program identified outlier data. The erroneous data were left in the output file, but were labeled with “-9.990e-29” instead of the original values, and were not included in subsequent analyses. In 2014 the temperature data ran through the program using the same parameters as salinity, and eliminated the erroneous temperature data.
The pump for the thermosalinograph had to be turned off occasionally to prevent the motor from malfunctioning, but the exact timing of pump shut down was not recorded. When the pump was shut off there was no new water being taken in and circulated through the system, so the water temperature artificially increased. Any data readings with temperatures that seemed unreasonably high (SST over 17 C) were assumed to have occurred when the pump was turned off, and both temperature and salinity values were removed from analyses. Almost all of these readings occurred when the boat was stationary.
SSS values of under 30 were eliminated because we sampled in the ocean and lower salinities are not found in saline water. We retained the temperature readings associated with these erroneous salinity values.
Technical issues precluded latitude/longitude from being recorded with thermosalinograph data at Chagulak and Round islands, so we estimated location using dLog location data. We used starting time, scan interval (5 seconds), and scan time from the thermosalinograph (TSG) to calculated real time for each reading. We then used the “mround” function in excel to round the TSG and dlog times to the closest 5 seconds, and then made a query in access to match the lat/longs from dLog data to the closest TSG scan.
Only data from spokes around islands- no transit data.
Because the TSG data was collected every 5 seconds regardless of the speed of the ship, the number of data files in any location varied dramatically. In order to get an average temperature and salinity value for every 3km that we traveled in spokes around islands, a raster file was created from the point files (ArcGIS conversion tools- to raster- point to raster; save in geo database, no numbers in the front of raster name). We created raster cells that were 3000m by 3000m in size, and found the average of the temperature and salinity values from within that cell.
In order to get a representative mean SST and SSS values from each 3km bin, we only used mean values from bins that had at least 30 readings of SST and SSS.
Seasoft V2: SBE Data Processing User's Manual. Software Release 7.21k and later, 5/23/2012. Seattle, WA. www.seabird.com.