LE QUYNH MAI
STUDY ON EXTRACTING LARGE PADDY AREAS FOR THE LOW SPATIAL RESOLUTION SATELLITE DATA
Professor RIKIMARU ATSUSHI, Associate professor TAKAHASHI KAZUYOSHI
Rice is a staple food for half of the world's population. Environmentally, knowledge of rice growing areas such as information on the spatial distribution and temporal dynamics of paddy rice fields is important for the study of trace gas emissions, management of water resources and food security. Therefore, monitoring of paddy land area and paddy land information is very valuable in agricultural management, planning and decision making to ensure the long-term sustainability of agriculture. Rice is also a main food in Japan. Since Niigata prefecture is one of the major rice producing prefectures of Japan, for ensuring food security in Japan, it is necessary to investigate land information by time.
In order to monitor and manage paddy land, remote sensing with satellite imagery has been particularly useful. However, monitoring paddy land area and paddy information require using the high temporal resolution satellite data such as MODIS Terra/Aqua Surface Reflectance Daily L2G Global 250m data. Because the average size of a rice paddy field in Japan is smaller than one MODIS pixel, we extracted large paddy areas using Landsat 8 OLI data. According to Landsat paddy areas, we examined spatial purity that is ratio of the number of pure paddy pixels in one MODIS pixels. Base on spatial purity map, we determined the threshold of spatial purity for extracting MODIS large paddy areas. The result presents that this research can extract 65% paddy areas of the entire target region. However, if we choose separate threshold of spatial purity, the accuracy is higher. For example, in case of Niigata city, we can extract 98% paddy areas compared with statistical data with the threshold is 60%. In case of Joetsu city, with threshold 20%, we can extract 89% paddy areas compared with statistical data.
The result also shows that if we use another kind of DEM for extracting less than 5-degree slope region, this research result may be improved. However, for monitoring the information of paddy land, it does not need to exploit the entire paddy areas exactly like statistics. Therefore, in next step, this research result can be use for monitoring paddy land information
Please use "previous botton" to return the previous page