I imagine co2 not being perfectly mixed and maybe wafts of it hitting the sensor? The overal trend is clear though. Try transforming the data with a moving hour

I imagine co2 not being perfectly mixed and maybe wafts of it hitting the sensor? The overal trend is clear though. Try transforming the data with a moving hour average or something.