Research About Algae Remote Sensing Monitoring In Taihu Lake Based On MODIS Data

With the application of remote sensing technology, cyanobacteria bloom could beobtained effectively, comprehensively and real-timely. Because of it’s multiband,higher time resolution and obtaining for free, the datas obtained from MODIS whichis carried by serial of satellites of USA new generation Earth Observing System(EOS)have been used very widely in the field of remote sensing. Based on the MODIS dataand spectral theory, this paper researches the extraction models for cyanobacteriabloom and establishs a set of systematic processes, which provides scientificprocessing methods and tools for algae in Taihu Lake water monitoring. The focus ofthis paper is mainly reflected in the following three aspects：(1)In order to obtain the remote sensing reflectivity image from MODIS data, thispaper uses the HDF library functions to read the information of bands, latitude andlongitude data, and file attributes, then completes the process which contains cutting,radiometric calibration, geometric correction.(2)Based on the spectral characteristics of water, water body is extracted from thereflectivity image by using the near-infrared band method and water indices method.In order to produce the water mask map of Taihu Lake, it is necessary to remove theinterference around the water body of Taihu Lake. Based on the defects in traditionalconnected component labeling algorithm, this paper propose a new fast labelingalgorithm, which efficiently segment the water binary image, and successfully extractthe mask image of the Taihu Lake.(3) Through studying the spectral characteristics of typical objects of MODISdata, the distribution of cyanobacteria can be obtained with the use of visualinterpretation of single-band grayscale and multi-band color image. Several typicalmethods for cyanobacteria extraction are discussed systematically, including theexisted problems. Then a new model is put forward based on the NDVI algorithm.Combined with the decision tree and Taihu water mask, the classification ofcyanobacteria bloom could be obtained. Experimental result shows that the newmethod which can well interpret the distribution of cyanobacteria is available andmore precise than NDVI algorithm.Through the integrated development of software platforms, the operationalprocesses for cyanobacteria extraction confirmed its efficiency. Compared with the traditional vegetation extraction model, the improved model for cyanobacteria bloomextraction has more advantages, especially in the haze and muddy water interferencecase.