根据MODIS图像中8.5 μm、11 μm、12 μm三波段的亮温判断云相态的原理,建立一个三层的BP人工神 经网络算法。利用该算法分别对中国高纬度区域(N30°～N55°)和低纬度区域(N0°～N30°) 的2008年1月15日和2008年7月15日的两景MODIS图像进行了云相态的识别,并把识别结果与NASA中心的MOD06云相态结 果进行了对比。对比结果表明利用该方法反演云相态的正确率在90%以上,且利用该算法反演得出的云相态 结果中,无法确定的云相态范围减少。

A triple-layer BP neural network was put forward based on the principle of the 8.5 μm, 11 μm, and 12 μm brightness temperature to retrieve the cloud phase in MODIS images. The cloud phase was retrieved based on the artificial neural network in high-latitude region(N30°～N55°) and low-latitude region(N0°～N30°) on January 15, 2008 and July 15, 2008 in China, and the result was compared with the cloud phase of MOD06 of NASA center. The result indicates that the inversion precision of the method is higher than 90%, and the uncertain scope of cloud phase is decreased.