Remote Sensing and Detection of Dust Storm in China Using the Thermal Bands of Geostationary Meteorological Satellite

The earth observation system from space includes two kinds of platform:Polar orbit and geostationary satellite. Optical sensors onboard polar satellite have the advantage of high spatial resolution and more spectral bands in visible to infrared regions such as AVHRR/NOAA, MODIS/EOS, MVIS/FY-1C/1D etc. But it can conduct only twice a day. It is not enough for hazardous dust weather whose time scale is very short and moves quickly. It is difficult to understand dust moving and evolution as a whole using polar sensors' observation. Geostationary Meteorological Satellite such as GMS-5, FY-2, GOES and Meteosat can observe the earth continuously all daytime and night at high temporal resolution. Developing an algorithm for remote sensing and monitoring dust event will be very useful for forecast model, environment and climate monitoring and scientific research. Dust cloud is not easy to be discriminated like other strong weather phenomena such as typhoon. It is to be understood of the optical and radiative mechanism of airborne dust. The base theories of remote sensing of airborne dust will be introduced using the thermal and other bands of geostationary sensors. Observation signal of thermal infrared window (8—12 μm) bands have almost no sensitivity to general aerosols with small and thin particle sizes. There are some sensitivity to large and strong dust particles, especially in dust storm or heavy dust storm. The airborne dust can exert two kinds of features on thermal infrared observation signals. Firstly the infrared radiance of ground target into space will be reduced by dust layer and the brightness temperature of the observed underlying target be decreased. This kind of temperature reduction is called infrared difference dust index (IDDI). Secondly the emissivities of airborne dust are different in these two split window bands and produce the negative brightness temperature difference for dust targets. Based on these above theories and the traditional sophisticated multispectral classification technique, a set of algorithms for automatically detecting dust storm is developed using observation data of geostationary meteorological satellite. The first step of this algorithm is to extract the data of all bands from normalized disk image of observation and to conduct the interested region projection and calibration processing. The data reading is not only from present time but also from previous ten days for integrating background brightness temperature image. It is ready for IDDI image integration. The second step is cloud mask processing which is very important for dust discrimination. And then dust determination is conducted using above mentioned theories and methods. It is the key part of this algorithm. The last step is the production output including several types of dust remote sensing. This algorithm can obtain ideal result of dust storm detection and the product of IDDI. This algorithm has already been experimentally run in National Satellite Meteorological Center since 2001. It is not only used for dust detection from data of Japanese GMS-5, but also becomes a useful operational production of the new geostationary meteorological satellite FY-2C in orbit in 2005. In addition, it provides extended potential of quantitative or semi-quantitative remote sensing of dust storm.