Phenology detection of winter wheat in the Yellow River delta using MODIS NDVI time-series data

Phenology detection has a significant impact on monitoring crop growth and crop yield estimation. Due to short time terrestrial formation, shallow buried depth and high salinity groundwater, soil salinization is serious and has negative influence on crop growth stages in the Yellow River delta. Traditional method for phenology detection using situ data is far more costly and time consuming, moreover, hard to available for all the fields. Soil salinization differs from place to place. The phenological stage varies from low salinization area to high salinization area. MODIS data provides the possibility for regional dynamic monitoring in a timely and accurate way due to its repeated acquisition and broad area coverage. Aim of this study is to detect major phenology and analyze the spatial distribution characters of phenology affected by soil salinization for winter wheat in the Yellow River delta region based on the MODIS NDVI time-series data. Savitzky-Golay filter procedure was selected to denoise. Decision tree method was used to classify winter wheat from other crops and natural vegetation before phenology detection. Phenology was specified by using defined dynamic threshold method. The phenology of green-up stage, heading stage, and harvesting stage was detected. This study concludes that green-up date generally came in early March, headed in early May and harvested in early June. The overall average green-up date occurred in March 4, heading date in May 7 and harvesting date in June 2. The detection result was consistent with the ground observations result on the whole. The spatial distribution of phenology showed a gradual postponement from inland to coast. Heading date and harvesting stage inland might be about 3 days in advance than those near the sea. Green-up stage inland might be about 6 to 10 days earlier than near the coast. Green-up stage was significantly influenced by soil salinity comparing to heading date and harvesting date. Method proposed in thi- paper can be used in phenology detection for winter wheat in Yellow River delta region, which has important guiding significance for crop condition evaluation and phenology detection in other coastal salinization area.