I have multiple measurements regarding scientific observations. The problem is that there is a subtle noise pattern caused by the instrument - the wiggles. These wiggles are invisible when looking at the scientific data itself (plotting x vs y), but become clear when I divide one measurement by another taken at a later time (when the wiggle pattern between them changed/shifted). My goal is to correct the "wiggles" for each individual measurement. So far I've tried dividing the measurement by a shifted copy (no success, the data has too many features it seems) and to divide by a smooth function of the data (a kind of polynomial normalization, but no wiggles were found after).