The Research On Green Tide Satellite Remots Sensing Methods Refinement

In the green tide operational remote sensing monitoring,250m resolution MODIS data is the primary source of data. Normalized difference vegetation index is the primary method used for extraction of green tide information. However, studies have found that, due to the MODIS spatial resolution is low, there are a lot of mixed pixels, resulting in the extraction of the green tide significantly larger coverage area, the impact of green tide operational remote sensing monitoring accuracy. To address this issue, the main work of this paper are:1) 4 bands HJ-1 CCD paper using 30m resolution, based on the difference between the band and the specific methods were constructed five new algorithms, and information extraction experiments. The results showed that green and red band structure of the algorithm can not extract green tide, the results of the other five band combination algorithm to extract little difference. Then, further to the green tide information extraction capabilities five algorithms were compared, results showed that:the traditional NDVI algorithm outperforms other algorithms, but the worst red and blue band algorithm built.2) Based unmixing method, using 250m resolution MODIS images containing green tide as linear decomposition of the green tide covering an area of quasi-synchronized 30m resolution HJ-1 CCD image covering an area of green tide build green tide covering an area of extraction of fine model. Pixel Decomposition method compared with the traditional method and mixed NDVI threshold, the method extracted independently verified green tide zone covering an area closer to the true value of prospective, whichever is the true value of an area of about 96%, corresponding NDVI threshold and mixed pixel decomposition method to extract the area were subject to 2.96-fold and 45 percent of true value. In addition, NDVI threshold method and compared this method is not sensitive to changes in NDVI threshold, the extracted value of green tide covering an area relatively stable. In this test the sensitivity of NDVI selected sample areas in order to monitor operational experience, NDVI threshold selected value [-0.20,-0.15] between and found that NDVI threshold method to extract green tide covering an area of increased 1.41 times; unmixing extraction of green tide covering an area of increased 1.11 times; refinement of model extraction of green tide coverage area grew by 1.11 times. Innovation of this paper is mainly reflected in:1) The unmixing method is applied to green tide information extraction, and to good effect.2) Construction of the MODIS fine green tide coverage area extraction model, to a certain extent, improve the use of traditional methods to extract green tide MODIS data coverage area larger problems, and reduce the sensitivity of the NDVI threshold and simple, can be used in the green tide of operational monitoring.