Internal data quality control

When collecting average current data with your Generation 2 Aquadopp instrument, the averaged data for each cell and each configured average interval is saved to the [filename]_avgd.aqd file. The data format used for the average current data is data format 7. Single ping data is not saved.

To avoid bias from poor-quality pings, the instrument applies an internal quality control before averaging. Note that this internal quality control is only relevant for the average current data, as burst types of measurements like wave bursts use data format 3 and save individual pings in the [filename].aqd file. An overview of all the files saved to the instrument recorder can be found in the following article: Data files

The internal quality control process for the DF7 average current data can be outlined as follows:

1. Filtering of individual pings: All pings with correlation lower than 50 % are discarded. For more information about the correlation parameter, please refer to the article Correlation - theoretical background. In addition a fish filter is applied. The fish filter detects spikes in each beam by comparing the signal strength to the mean values of the other beams.
2. Bin mapping: the data is bin mapped on a ping-by-ping basis. For more information about bin mapping, please refer to the article Tilt
3. Averaging: the bin mapped pings that remains after the correlation filtering are averaged together within one average interval and one cell.
4. Quality Code: each averaged cell is given a specific Quality Code. The Quality Code flags potential data issues but does not discard or modify the data. Each bit indicates a specific quality check.

Percent good

In the DF7 average current data there is also a Percentage Good value which indicates the percentage of valid pings retained in each cell during an average interval after the internal quality control. If the Percentage Good value is 0 %, no valid pings contributed to the average, and the resulting velocity is set to 0. The removal of pings with a correlation lower than 50% is irreversible as this is done prior to the averaging to ensure good data quality.

Quality code

The Quality Code is a 16-bit status code. Each bit in the Quality Code indicates a specific quality check, as listed below:

Bit 0 (0000 0000 0001) - Not used

Bit 1 (0000 0000 0010) - Fish filter
Mask based on spike detection in each beam when compared to the mean values of the other beams.

Bit 2 (0000 0000 0100) - Correlation
Pings with correlation less than 50 % are masked.

Bit 3 (0000 0000 1000) - Pressure
Uses the pressure readings to detect if the collected data is out of water.

Bit 4 (0000 0001 0000) - Not used

Bit 5 (0000 0010 0000) - Sidelobe
Masks cells contaminated by surface sidelobes by comparing the minimum absolute pressure during the average interval to a conservative cutoff value of 1050 hPa for the corresponding surface pressure. For more information about sidelobes, refer to the following the article Sidelobe interference

Bit 6 (0000 0100 0000) - Percent good
Masks the cell if the Percent Good value is below 50 %. The Percent Good is calculated by dividing the number of pings that have passed the correlation test by the total number of pings within the cell and average interval.

Bit 7 (0000 1000 0000) - Not used

Bit 8 (0001 0000 0000) - Bin map
Used when no cells could be mapped to this vertical position, for example due to a tilted instrument.

Bit 9 (0010 0000 0000) - Tilt
Masks entire profile where either the average pitch or roll exceeds 40° through the average interval or the pitch or roll standard deviation exceeds 5°.

If the Quality Code is 0, all controls are passed. Otherwise, the code itself indicates why the data is flagged. If a data point is flagged by several control parameters, the code will be a combination of the respective bits. Note that these checks do not discard or alter the data in any way, only flag it. The exception is the Bin map flag which indicates there was not sufficient data coverage to map data to a specific cell position due to a tilted instrument. This will leave the velocity cell empty.

During post-processing, you can either discard data flagged by one or more quality parameters, or retain it and perform manual quality control.