How to assess data quality in the OptoScale web portal?

The OptoScale web portal shows data and measurements based on images of the fish taken by the bioscope in the pen. For the measurements to be as representative as possible, good and stable data capture is essential.

Best practice is to have one person responsible per site who checks the web portal daily. We recommend checking the operational status and monitoring data capture and data quality for each module.

In short

• Ensure good data capture first. More registrations provide a better basis for the measurements.
• Look at the trend over time, not at individual measurements.
• Deviating measurements do not necessarily mean poor data quality. Check whether there is a natural explanation.

1. Ensure good data capture

An important prerequisite for good data quality is that the Bioscope captures images of a sufficient number of fish. The more registrations over time, the better the basis for accurate estimates and precise data.

Position the bioscope so that the fish naturally swim past the camera. See the user manual for more information on installation and deployment.

What counts as good data capture depends on, among other things:

• The number of fish in the pen
• Whether it is a hatchery or a sea site

2. What to do if data capture is low

If the number of measurements is low over time, we recommend the following steps:

1. Check the bioscope status and data capture. Check the operational dashboard, images, live video and image classes to find the cause.
2. Compare with other bioscopes at the site. Do one or more pens have lower data capture? Compare position and depth with pens that have better data capture.
3. Check where the fish are in the pen. Use the feeding camera to see where the fish are, compared with the Bioscope’s position.
4. Adjust the bioscope. If your checks indicate that the position is affecting data capture, the bioscope should be adjusted.

Common causes of low data capture

• Obstructions in front of the camera
• The fish swim too far from or too close to the Bioscope
• Fouling on the lenses
• The bioscope is hanging at an angle
• The bioscope is at a depth or in a position with few fish

3. How to assess data quality

When data capture is good, we have a solid basis for assessing data quality. The following signals in the data may indicate that the measurements should be examined more closely:

4. What to do if data quality is uncertain

If the measurements seem abnormal or do not match expectations, we recommend the following approach:

1. Check data capture. Make sure data capture is good and stable. If it is low, follow the steps in section 2.
2. Consider whether the situation is temporary. The change may be caused by conditions that are expected to pass. For example, recent activity or operations at the site, reduced visibility/algal blooms, temperature changes, changes in feeding regime or appetite can temporarily affect fish behavior. Look at the trend over time. Do not draw conclusions from individual measurements.
3. Compare the measurements with what is actually happening in the pen. Is there a natural explanation? Disease, reduced feeding or other changes in the condition of the fish can lead to changes in the measurements. Check whether the deviation is due to real factors in the fish group or in operations.
4. Adjust the bioscope if deviations are persistent or unexplained. If the measurements do not make sense over time, or deviate significantly without an obvious explanation, the bioscope’s position should be adjusted.