Three-dimensional histogram method for Asian dust identification using MODIS images

Asian dust storm is a meteorological phenomenon in East Asia, which is originated in arid and semi-arid region such as Taklamakan and Gobi deserts. The heavy Asian dusts are occasionally carried by strong winds to Korea, Japan, even to North America. These dusts cause respiratory disease, transportation disturbances, and other disruptions of social and economic activities. Monitoring these dust transports is essential and remote sensing technique and fast computer visualization are required. However, it is not easy to identify Asian dust over Japan Sea, which is relatively low dense and coexists with cloud, fog, and other haze. To overcome this issue, we used a three-dimensional histogram technique developed by J. Kudoh. The original was developed for identifying land, cloud, sea, fog, etc., from NOAA AVHRR reflectances. Instead, we used three MODIS indices, which are Brightness Temperature Difference (BTD), Normalized Difference Dust Index (NDDI), and Normalized Difference Vegetation Index (NDVI). The BTD is sensitive to high-density dust but less sensitive to low-dense. NDDI is hard to discriminate Asian dust from bare land. Using the three indices, the proposed method identified Asian dust very well. Compared to meteorological observing stations, it is found that the observation of low-density dust is required for Japan and the proposal method is useful for this purpose. Advantages of the three-dimensional histogram method are fast detection and identification capabilities of Asian dust from various dust behaviors and this is useful for automatic dust detection.