Sea pollution monitoring with relative radiometric calibration and change detection on multispectral remote sensing images

Heavy metals are harmful to our ecosystem and also threaten human's health. In the northern east of Taiwan, the heavy metal pollutants from abandoned copper-gold mines cause the sea color become yellowish. Remote sensing is an effective and efficient tool for this environmental monitoring. Remote sensing images have been widely used for ground material analysis for the last a few decades, and among them, change detection is one of the most important topics. When the object of the earth surface changes, whether by nature or human behavior, the spectrum will also be changed. By comparing the images collected at two different times, we can identify the area of the changes. However, the changes in radiance spectrum are not entirely from the surface changes, part of them came from the atmospheric effects. Therefore, we must calibrate the radiometry before the comparison. There are two ways to calibrate the images: absolute calibration and relative calibration. The former remove all the atmospheric effects. In order to achieve this, we must have the knowledge of the atmosphere condition, which can not be found easily. On the other hand, the relative calibration uses the information of the images from different times and matches the atmospheric condition between them statistically. In this study, we focus on the relative calibration for change detection of the Yin-Yang Sea. Several remote sensing images collected from different times will be used to monitor the changes.