Most instruments and sensors drift away from their calibrated signal-response over time and use. In the lab, especially instruments that can have carry-over contamination from previous samples, it's easy to run a blank sample at adequate intervals and subtract the blank from subsequent samples. It's not really possible to run a sensor blank in the field, so what could be other options? For something like conductivity, would it be feasible to take conductivity calibration standards into the field and test the sensor's response to those standards, recalibrate, and possibly make corrections to previous measurements? How have you addressed signal drift with low-cost sensors?
You might find this thread helpful; it lists open-source packages and code for "cleaning" environmental time series data: https://twitter.com/amalia_handler/status/1151598449063682049 One would have to have some skill with coding to use those, though. I know some of the people in the thread, including the StreamPulse folks at Duke (your profile says you're in Durham, in case you personally want to talk to them directly); I could ask them to weigh in if that would be helpful. I get that ultimately all the info is supposed to go on this site; I just don't know that much about this topic personally, and am new here.
Reply to this comment...
Log in to comment