Cloud detection method for NOAA AVHRR images by using local area parameters

For researchers who observe the sea or land surface, the elimination of clouds is the primary important step to process the images. Saunders et al. proposed several cloud classification methods that use a visible channel (CH1) and a near infrared channel (,CH4) of the AVHRR, and it is effective for the European area. However, there are several classification problems when we apply these methods to the Far East area. Therefore, we propose a new method that uses some area-specific parameters: the threshold value for CH1, CH2 and CH4. The former value is related to the elevation of the Sun, which strongly depends on the time and the location. The latter is determined by the long-terin image analysis of a certain area. We analyzed images from the past two years of the Far East area. So we dynamically decided the CH1, CH2, and CH4 threshold for each image. It becomes clear that local area parameters are effective for the cloud elimination