In environmental management programme, accurate information about the source of pollutions plays an important role in establishing appropriate environment management measures and plans. In order to obtain information on a wide range of the pollutants,...
In environmental management programme, accurate information about the source of pollutions plays an important role in establishing appropriate environment management measures and plans. In order to obtain information on a wide range of the pollutants, satellite or aircraft images were actively conducted. However, due to limitations such as the acquisition cycle and the resolution of the image, it is difficult to apply them to the agricultural area where the variation pattern of the pollution source is irregular and the continuous management is required. To overcome these limitations, researches on UAV have been rapidly increasing in recent years as well as image sensor technology has also been developed. In addition, sensors that can acquire various image data such as multi-spectral and thermal infrared rays were commercialized through UAV technology.
The purpose of this research was to make a subdivided land cover map which is easy to manage the pollutants after acquiring image data of agricultural area by using UAV equipped with RGB, multi-spectral and thermal infrared image sensor. In addition, the characteristics of the pollutants were derived base on various image data, and the possibility of extracting information of the pollutants was then confirmed by using image classification technologies.
In this research, the ortho images generated by the UAV were largely used for two purposes. One was the RGB orthoimage which was used as the material of visual interpretation for the subdivision of the research area and the other was the orthoimage generated by the wavelength band to analyze the spectral characteristics of the land cover. In addition to the existing land cover items, the subdivided land cover map based on the RGB orthoimage newly added seven items that were deemed necessary for the management of the pollution source in the agricultural area such as farm waterway, windrow compost, etc. Therefore, the subdivided land cover map of this research site has the attribute information about 24 items in total, and has the current status information which is not contained in the subdivided land cover map provided by the environmental spatial information service.
The mean value of the spectral information by land cover type was calculated by using the subdivided land cover and the orthoimage of wavelength band produced by this research as the input data of the zonal statistics. The spectroscopic characteristics of seven kinds of orthoimages such as Red, Green, Blue, Red Edge, NIR, TIR and NDVI were verified, and then the potential for extracting pollutants was investigated by supervised and non-supervised image classification. However, it was difficult to search for the characteristics of pollutants due to the similar spectral characteristics of the many pollutants' covering materials and it was confirmed that there was a limit to extract the characteristics of the pollution source with only spectral information. In order to overcome these limitations, it is necessary to develop an image classification method combining not only spectral characteristics but also various spatial factors such as height, square area, etc.