Object-Based Analysis and Change Detection of Paddy Field at Hokkaido, Japan

Abstract:

Remote sensing technology has been used in land use and land cover classification. Especially paddy fields is an important cultivated area in Asia. To accurately extract the area is the important indicator to estimate the food production. In this research paddy fields classification in Hokkaido was performed using Topographical features (DEM), Climatic features (accumulated temperature), Spectrometer features (MODIS).

Fig. 1. shows the overview of the analysis methods of this research. The process of this research is carried out in 3 steps.1. Determine the accumulated temperature by retrieving the temperature data from the AMeDAS data. 2. Extract the river from the DEM. Set the elevation of the river to 0 to seek the land elevation around it relatively. 3. Calculate the WI (Water Index) using MODIS band 4(Green)and Band 2(NIR).

Time series NDVI has been corrected by the FFT method (use a low-pass filter). Phenology information was extracted such as vegetation Onset time, Max value and Duration. Result of the classification was compared with the current vegetation map of the Ministry of the Environment. As a result we can confirm that existence of vegetation map around Sapporo and Asahikawa is almost same. but around Obihiro it was clear that overestimate by water index.