Land cover classification in Western Loess Plateau of china with MODIS imagery

Land cover classification represents one of the most fundamental applications of remote sensing, and is widely used to estimate carbon stocks, hydrological parameters and biogeochemical models. Land cover classification using multitemporal and multi-spectral remote sensing observation data has been an effective technique. This paper adopts a multi-dimension classification method to classify the. land cover in Western Loess Plateau of China using EOS/MODIS remote sensing image and some ancillary data. The imagery used in this study were 30-day composite Normalized Difference Vegetation Index (NDVI) over the year 2003 obtained by the Maximum Value Composite (MVC) technique, and the ancillary data included phonological calendar, agricultural knowledge and existing land cover dataset. Discrete Fourier transform (DFT) as a signal decomposition method can extract periodic responses through expressing a NDVI time series curve with the sum of an additive term and a series of cosine waves and it has been used in land cover classification. Therefore the discrete Fourier transform was applied to the NDVI data set on a per-pixel basis for the whole study areas and images of the additive, amplitude and phase angle for the first two harmonics were produced on a per-pixel basis for each pixel in the NDVI dataset. After integrating the multi-dimension information, the land cover in Western Loess Plateau of China is finally classified using ISODATA unsupervised classification algorithm. At last, the accuracy of the result was evaluated with existing land cover dataset. The tentative result shows good correlations with existing digital land cover map, though the small overestimation or underestimation are recognized in several categories.