The Detection Algorithm Of Marine Oil Spill Based On Time Series Using MODIS

Currently, the oil spill accident has already become one of the major and serious pollutions in the marine environment, of which the ecotope has been badly devastated by series of pollutions. While the world is running increasingly frequent economic exchanges, the number of ways for ocean transportation booms incredibly fast as well. As for the result, the marine environment has been seriously damaged by various external human activities. Normally, the oil spills in the sea generally caused by situations like:the explosion of offshore oil drilling platform, the oil exploration during the blowout and leakage, or the tanker ran aground during shipping, grounding and collision etc. The oil spill accident always features characteristics of:high frequency, widely distributed, highly influence of hazard. Therefore, the most accurate and timely detection and identification of the oil spill area plays key role in disaster recovering and reduce the extent of the harm in the bud. To find out the best way to detect and identify the oil spill has been on the top of the problems many countries in the world are urgent to solve.With the rapid development of spacial information technology, the various of remote sensors carried on satellites can obtain a mass of satellite images of the sea surface in a short time, with high spatial and temporal resolution, which has greatly met the current demand for oil spill monitoring field. With the remote sensing technology by satellites, we can cope with the oil spill event happens in a large region without caped by the limited detection area for planes, and can be working real-time in any case, which makes up for short of the airborne remote sensing. Therefore, the technology of oil spill monitoring is trending to the way, of which the satellite based remote sensing dominates, supplemented by combining with airborne remote sensing.The state of the art in oil spill detection in marine by using satellite remote sensing are: Identify the oil spill information by interpreting the satellite image according to the characteristics of the solar spectral reflectance in a certain single band or with several bands combination cause by the oil film which covered on the sea surface. The methods for interpreting are:directly interpreting and identifying by experts experience, identifying by comparative analysis, or logical reasoning, etc.. But with all of them in common, it requires rich knowledge of experts'experience. What's more, they are all carried on the base of under conditions of being noticed with the oil spill accident information about when and where. However, given the consideration of frequent oil spill incidents nowadays, the real-time monitoring plane and the monitors can't make sure will capture the information of every single incident, especially when the weather conditions are too bad to aviate the monitoring aircrafts. Therefore, we need find out other methods which can be real-time monitoring and detecting in any case. In this paper, given the situation of the research background in oil spill monitoring, we take the trial area of Bohai Sea, where the oil spill incident happens in the most frequency, the method carries out with a length of time-series data composed with many MODIS LIB remote sensing data sets in different phases. Based on the abnormal solar reflectance display on the satellite image when oil spill incident happens, it is easily to identify some abnormal events by Gaussian distribution statistical theory with the actual optimum threshold value, which is composed by the mean value and standard deviation of the time-series data sets. Thus, we could find out the time of the abnormal events (oil spill) happened, and the spacial distribution of the oil spill area. However, for some area where is covered by the very thin oil film could only display a subtle difference in the solar reflectance cause by the oil film and the clean sea water, while the Gaussian distribution method would not entirely identify the fine difference, here we propose another methods wavelet analysis, which is so called by "mathematical microscope".Wavelet analysis can zoom in the details of the signals, by which we could enlarge the difference of the solar reflectance caused by oil film and clean sea water., which helps to identify the oil spill area more accurately. This paper presents an method in monitoring the space-time dynamic process of a time-series of the satellite images, which is coordinate with the future development of the current geographic information science (Geographical Information Science) with the trend from the traditional static approach completion of the geographic space to the "space state" studying. The future geographical research focuses on geographical process or dynamic phenomena instead of static geographical structure. The methods in the paper improves the accuracy of oil spill monitoring technology and react more timely on the incidents which will play important role for reducing the harmful hazard of the marine environment, and be supportive to the authority to take the countermeasures to solve the problems.