Nighttime Terrestrial Radiation Fog Detection Using Time Series Remote Sensing Data

In this paper, we propose a nighttime terrestrial radiation fog detection model using time series data of Multifunctional Transport Satellite-2 (MTSAT-2), which addresses the difficulty in nighttime fog detection with mono-temporal remote sensing data. The nighttime fog is firstly extracted by using monotemporal image. To take advantage of high temporal resolution of MTSAT-2, the temporal curves of brightness temperature difference between band 1 and band 4 and the temporal curves of brightness temperature of band 1 are built based on the nighttime fog detection result of mono-temporal image. The land surface were separated from the nighttime fog and the low clouds by the bright temperature difference accumulate characteristic established by the temporal curves of brightness temperature difference. The nighttime fog are detected by combining three temporal characteristics with support vector classification. The three temporal characteristics are the bright temperature change accumulate characteristic, slope match characteristic and frequency domain singularity characteristic, which are established by the temporal curves of brightness temperature. The experiment results of two days show that the nighttime terrestrial radiation fog detection using time series data has a higher accuracy than the mono-temporal method.