ATN Satellite Telemetry Quality Control

Quality Control Protocols

The Animal Telemetry Network (ATN) DAC quality control (QC) protocols handle animal trajectory and dive profile data and use the ioos_qc Python package (see docs) to implement multiple QARTOD tests and an aggregate rollup flag.

The following ioos_qc tests are applied:

• ioos_qc.argo.speed_test
• ioos_qc.qartod.location_test
• ioos_qc.qartod.gross_range_test

The Location and Speed tests are applied to trajectory data only, while the Gross Range Test is applied to temperature and salinity profile observations. Data are not removed during the QC step, but the QC flag results are included in the processed datasets.

Quality Control Flags and Definitions

Prior to applying the quality control protocols, raw trajectory and profile data for each deployment are first truncated based on deployment start and end dates provided in the metadata. This is a necessary data cleaning step since telemetry tags are often activated and begin pinging their locations before and after they are attached to the animal and officially “deployed”. These out-of-bounds locations may appear either on land or at sea.

Processed, tidy data with QC flags are then submitted to a variety of downstream locations and repositories including the ATN Portal, National Data Buoy Center (NDBC) for integration into the Global Telecommunications System (GTS), DataONE, NCEI, and ERDDAP for increased data accessibility and archival. Please see the ATN DAC Flow Overview for more details. For deployments that meet certain criteria (outlined here), the data are run through the aniMotum model to produce regularized, predicted trajectories at a 2-hour time step that are visualized in addition to the location data on the ATN Portal. The modeled outputs are not currently archived, but the ATN DAC has plans to eventually include these alongside the deployment data.

Raw Data Tidying

The ioos_qc.axds.valid_range_test() is applied to the raw trajectory and profile data to keep only data falling within valid deployment dates. Typically, this data cleaning step involves checking the ingested data against provider-supplied start and end deployment dates, if available. While the raw data will retain the data falling outside the deployment dates, only the clean deployment data will visualized on the ATN Portal and submitted to downstream national repositories such as NCEI and ERDDAP.

Valid start and end deployment dates are either accessed from manually submitted deployment information saved in the ATN Data Registration portal (ADR) or directly from the upstream data source (ATN manufacturer API or server). For the latter, this information is typically stored in the XML metadata file packaged alongside the data files. When both are available, the ADR deployment dates are used as the source of truth, since there is currently no good way of automatically including actual start/end deployment dates in the upstream data source.

The provided deployment date ranges are converted to datetime and used to truncate the data on both ends, to account for time on land before deployment and after retrieval.

Currently, only auto-ingested Wildlife Computer tag data include start and end date metadata (epoch time) in the corresponding XML file, which might look something like this:

<deployment>
    <start>
        <date>1464789420</date>
        <latitude>34.2178</latitude>
        <longitude>-120.5634</longitude>
    </start>
    <end>
        <date>1465776000</date>
    </end>
</deployment>

Trajectory QC Tests

The table below summarizes the tests and thresholds that are set for each ATN trajectory parameter:

Profile QC Tests

The table below summarizes the tests and thresholds that are set for each ATN profile parameter:

The min and max temperature thresholds and salinity upper bound threshold are taken from the Argo Quality Control Manual for CTD and Trajectory Data and are physical sensor bounds.

While the Argo Quality Control Manual suggests a gross range lower bound of 2 psu for salinity, modern conductivity-based salinity sensors can measure near 0 psu conditions with reasonable accuracy and without signal degradation (Menn & Nair 2022). NOAA’s Manual for Real-Time Quality Control defines the Gross Range Test to flag measurements outside of a sensor’s operational limits. Therefore, the lower bound we use here is 0 psu.