The Research Of Dynamical Detection Method For Radiation Fog Over Land Based On Remote Sensing Data

Fog is a cloud in contact with the ground that could be dangerous for human beings. Fog can reduce regional visibility, deteriorate air quality, and it has serious impact on sailing, aviation, highway transportation, military activities, etc. It also produces a great hazard to power equipment, crop growth and human health. It is very important to monitor fog in quasi real-time. Fog detection methods based on remote sensing technique use many kinds of satellite data, from goestationary to polar-orbiting satellites, through observing Earth’s surface repeatedly in large-scale. With the increasing development of satellite remote sensing technology,data source will be more numerous, reliable and stable,which makes fast and dynamic fog detection possible. Fog detection using remote sensing imagery becomes a new focus point.This dissertation covers:(1) The concept, characteristic, and the physical process from the formation to dissipation of radiation fog are summarized, and radiation fog over land distribution, affection in China and the influence caused by the climate change are introduced; (2) The existing theories and methods of fog detection, based on in-situ observation and remote sensing imagery at home and abroad, are summarized; (3) A parameter named Normalized Difference Fog Index (NDFI) is proposed based on analyzing the spectral character of fog and cloud by utilizing the the Streamer radiative transfer model and EOS/MODIS data; (4) According to the limitations of the existing fog detection method using remote sensing data and NDFI feature parameter, an object-oriented fog detection method is proposed based on the algorithm "Combined Mean Shift algorithm and Full Lambda-Schedule algorithm (CMSFLS)" which based on the existing two algorithms:"Mean Shift" and "Full Lambda-Schedule" algorithm and the fog detection method’s performance is evaluated against ground-based measurements over China in winter and the method is proved to be effective in detecting fog accurately based on 3 cases; (5) According to the sequential MTSAT-1R satellite images and the characteristic of fog including the procedure of its generation to dissipation, we can making use of the advantage of high temporal resolution of MTSAT-1R data and two types of fog is proposed in this paper. With the support of non-orthogonal Haar wavelet transform, characteristics of fog in frequency domain are selected, and a method based on sequential images and these characteristics is proposed, the method is proved to be effective in detecting fog accurately based on two cases.According to the researches in this dissertation, some conclusions could be derived as follows:1. Fog is very harmful to transportation, crop growth, power equipment, and human health, etc. With the increasing development of satellite remote sensing technology, data source will be more numerous, reliable and stable, which makes fast and dynamic fog dection based on remote sensing data pobsible.2. The difficulty in remote sensing fog detection lies in distinguishing cloud and fog. In this paper, EOS/MODIS data are used as the important data source for fog detection, NDFI is defined to discriminate fog from clouds,and the NDFI feature parameter is segmented into regions with the CMSFLS algorithm which is also proposed in this paper. By combining the "spectral information", "geometry informtion" and "texture information", the object-oriented method proposed is proved to be effective in detecting fog accurately over China in winter.3. According to the unique feature of fog from generation to dissipation, in this paper, the MTSAT-1R satellite data with high temporal resolution are used as the important data source for fog detection. Fog are defined as two types in this part and the differences of each fog and cloud in the frequency domain is described, experiments show that the method proposed in this paper is proved to be effective in detecting fog dynamically with two cases while experiments also show the method using geostationary satellite data proposed in this paper need to be improved in the future.