Due to its spatial resolution, the MODIS instrument offers much potential to monitor specific crops from space. However, only some time series fall adequately in the target crop specific fields while others straddle across different land uses, which consequently dilutes the signal. According to the daily change in orbit, the MODIS observation footprint changes considerably from one day to the next, sampling the vicinity of the grid cell. This study proposes a method to identify which time series are suitable based on the temporal signal-to-noise ratio (SNR) of such daily observations, which are acquired with different observation geometries. The approach is demonstrated over a 30 by 30 km study site in South Dakota (USA) where the time series with high SNR are classified in an unsupervised way into clusters almost exclusively composed of crop specific time series.