Study On Atmospheric Correction Methods For MERSI Image Over Pearl River Delta Region

The FY-3A Medium Resolution Spectral Imager (MERSI) can obtain the changes ofvegetation, ground, snow and thermal environment of urban area, show enormous potentialityin monitoring of land surface character. However, under the influence of absorption andscattering, the satellite measurements of earth’s surface are not real refectance of the object.So we must make atmospheric correction to abtain the real remote sensing information ofearth’s surface. Separating information of earth’s surface from atmospheric signal is the keyfor the application of the remote sensing technique, which influences accuracy of followingapplications such as imager interpretation and classification.Pearl River Delta belongs to industry city in the coastal region, whose optical propertiesare much more complicated than inland region. Standard atmospheric algorithm for remotesensing may cause big errors. It is necessary to develop atmospheric correction models forurban agglomeration in coastal region. This thesis firstly demonstrates the principle andmethods of SMAC, begin with the design characteristics of MERSI channel and thesynchronous meteorological records, combined with the atmospheric characteristics oflocale region, attempts to extract greater factor impact on local environmental from theatmospheric correction model, put forward a kind of simplified atmospheric correctionalgorithm. The algorithm uses ground visibility, vapor pressure and elevation calculate aerosoloptical thickness of0.55, based on the SMAC correction coefficient of the MERSI, removingatmospheric effects from satellite imagery pixel by pixel. Finally, evaluate the credibility ofthe results by normalized difference vegetation index (NDVI). In this thesis, SMAC packagedby Cython language on the Scientific Linux, base on FY-3A data at the Pearl River Delta inclose date, calculate NDVI before and after atmospheric correction, analyses the change ofNDVI sequence, validate the credibility of the atmospheric correction results.The result shows that:1、Dark-object method effective and shortcut，but computational accuracy is low,optimize method is a good idea.2、Importing synchronous meteorologic records into SMAC improves its atmosphericcorrection accuracy. Its semi-empirical coefficient is fit for FY-3A/MERSIatmospheric correction.3、The test method based on NDVI is valid for verify accuracy of the atmosphericcorrection result.