Comparison of Winter Wheat Spring Phenology Extraction by Various Remote Sensing Vegetation Indices and Methods

Crop phenology extraction using remote sensing is influenced by satellite data and retrieval method. We conducted a comprehensive evaluation of the accuracy for winter wheat green-up date (GUD) extraction with four VIs (NDVI, EVI, EVI2 and NDPI) and six methods (CCR
_max
, β
_max
, RC
_max
, RT10, RT20 and RT50), comparing with ground-observed GUD in China. We evaluated the performance with correlation coefficient (R), root mean square error (RMSE), and observations (BIAS). The results showed that GUDs extracted from NDPI time series presented the overall highest accuracy. Based on NDPI time series, we also found that GUDs extracted with CCR
_max
method had the highest consistency with ground-observed GUDs. Our finding suggested caution on the uncertainties in phenology extraction with remote sensing methods.