Extracting multiple cropping index based on NDVI time series: A method integrating temporal and spatial information

Multiple cropping index (MCI) can be extracted from satellite-derived Normalized Difference Vegetation Index (NDVI) time-series data. However, NDVI time-series data are often affected by noise, e.g. cloud, aerosol. The reconstruction of high-quality NDVI time series is a key to get accuracy MCI. A method integrating temporal and spatial information is developed to remove or reduce the effect of noises on time series. The proposed method mainly consists of two steps: (i) processing of cloudy pixel in spatial domain based on land cover information; and (ii) data processing in the temporal domain (HANTS algorithm). The results show this method can effectively reconstruct NDVI time series and MCI is highly consistent with statistical data. Nevertheless, this method needs ancillary information-land cover data and the quality of reconstructed time series depends on the accuracy of landcover to a certain degree, which may limit the application of the method.