Welcome to EnviroDIY, a community for do-it-yourself environmental science and monitoring. EnviroDIY is part of 
Hi Jake, no “pressure sensor” is going to be able to be linear/accurate if ice is forming and expanding in the protective enclosure round the pressure sensor. My guess is the whole sensor would have to be specially designed to allow for the ice expansion to even survive a freezing event.
My hypothesis (guess) would be that its the nearly freezing is causing an artifact in depth reading. What is intriguing for that depth of water is that there are temperature variations of 4F between 3~5Feb. So unless you could observe the location and validate that there are real depth changes, its the sensor that is generating anomalous readings.
Then from the 10-21feb when the temperature line seems to flatten just at what looks like ~ 32.5?F Either the temperature sensor is failing, or you would be expecting ice to be forming on the surface. As its stream water – and ice formation is complex at round freezing seems
See annotated graph
