Phenology-Aware Sampling Schemes

Background: Deep learning models that process time series data often interpolate observations to create equidistant time series, which may smooth over or obscure short-term phenological signals critical for distinguishing management practices.

Question / Future Work: Future methodological work should explore non-equidistant or phenology-aware sampling schemes for time series data acquisition to better capture short-term phenological signals that might aid in the discrimination between farming systems.