Oftentimes one will see data in their cast that looks erroneous or out of spec, but reviewing the timeline of each cast and the events which transpired can explain these jumps. If you are starting your cast while your CTD is on deck then the time during which the unit is running in air can be spiky or erratic, but this should be solved after the unit has been fully submerged and the pump has activated. The pump on time setting controls how fast the system will turn on the pump after deployment, so filtering out your deckside data can be done by calculating the number of scans to exclude using your pump on time setting, samples to average setting, and native sampling rate of your CTD.
Bubbles in the flow line can also cause spikes in your data towards the start of your deployment if the system isn’t able to normalize at the surface. We typically recommend units stay near the surface for 2-5 minutes in order to allow air bubbles to escape.
Finally, for additional resources in troubleshooting and smoothing data outliers for your CTD data, refer to our documentation on Seabird University.