An object-oriented based daytime over land fog detection approach using EOS/MODIS data
Xiongfei Wen, Liangming Liu, Wei Li, Pei Dong · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2009
A new algorithm is presented for land fog detection from daytime image of Earth Observation System Moderate Resolution Imaging Spectroradiometer (EOS/MODIS) data. Due to its outstanding spatial and spectral resolutions, this image is an ideal data source for fog detection. The algorithm utilizes an object-oriented technique to separate fog from other cloud types. In this paper, MOD35 product is first introduced to exclude cloud-free areas, and high clouds are removed with MODIS 26 band, and then a parameter named Normalized Difference Fog Index (NDFI) is proposed based on Streamer radiative model and MODIS data for fog detection. Through segmenting NDFI image into regions of pixels, and computing attributes (e.g. mean value of brightness temperature) for each region to create objects, each object could be identified based on the attributes selected to determine whether belongs to fog or cloud. Algorithm's performance is evaluated against ground-based measurements over China in winter. The algorithm is proved to be effective in detecting fog accurately based on two different test cases.