Uncertainty of cloud parameters on satellite estimation of radiation budget in climate change study

我们的研究成果如下：[高精度估算辐射收支和云辐射强迫的云光学特性的回顾与再检验]至今，我们继续利用GMS-5卫星从1996年开始估算云量、云光学厚度、地表亮度温度和地面向下太阳辐射的逐时数据。这些产品应该用在地面观察到的数据进行验证。我们的团队一直在东亚运作SKYNET观测网，研究气候变化中的云、气溶胶和辐射问题。比较得出了两个明显的趋势；对于晴天，比较显示出相当好的一致性，但对于阴天，在某些情况下差异较大，超过50W/m2。这将对辐射收支的估算产生严重影响。因此，我们明确了原因，并改进了估计算法。给出…最重要的参数是；(1)传感器响应灵敏度变化；(2)数字化噪声；(3)云的非均匀性。关于第(1)项，我们已经分析了GMS-5的这种退化趋势，得到了灵敏度以每年几个百分点的速度下降。在本研究中，我们定量地估计了与MODIS数据的效果。由于GMS可见通道具有低分辨率（6bit， 64步）和对输入亮度的二阶多项式响应，因此它对低反射亮度（即相对较薄的云）高度敏感，而对高反射亮度（如厚云）灵敏度较差。与MODIS数据的比较显示了两者之间的定量差异。此外，MODIS和GMS同时拍摄了相同的云图，但卫星天顶角不同，这意味着从云中接收到的辐射可能会导致云不平行时的误差。对云的三维结构的影响进行了估计。项目中的这些分析结果反映了我们估算云参数和辐射预算的算法。少

We have the research results as follows,[Review and re-examination of optical characteristics of clouds for estimating the radiation budget and cloud radiative forcing with high accuracy]Up to now, we have continued to estimate the hourly data of cloudiness, cloud optical thickness, surface brightness temperature and downward solar radiation at the surface using the GMS-5 from 1996. These products should be validated with the data observed at the surface. Our team has been operating the SKYNET observation network in the East Asia for the research of cloud, aerosol and radiation issues in the climate change. The comparison gave two distinguished trends ; for clear days, the comparison shows the quite good agreement, but for cloudy days it has large discrepancies with more than 50W/m2 in some cases. This would bring serious effects on the estimation of radiation budget. So we have made the reason clear and improved the algorithm for estimation.The most important parameters for giving the … More se errors are ;(1)sensitivity change of sensor response, (2)digitization noise, and (3)inhomogeneity of cloud.Regarding the item (1), we have already analyzed this trend of degradation of GMS-5 and got the decrease of sensitivity with several percents per year. In this research, we have quantitatively estimated the effect compared with MODIS data. Since the GMS visible channel has a low resolution (6bit, 64 steps), and a second-order polynomial response for the input radiance, it is highly sensitive for the lower reflected radiance, in other words, a relatively thin cloud, and the poor sensitive for the high reflected radiance, such as thick clouds. The comparison with MODIS data gave the quantitative difference between them. Also the MODIS and the GMS take the same images of cloud simultaneously, but the satellite zenith angles are different which means the possibility of the radiance received from the cloud can give the error if the cloud is not plane parallel. The effect of 3-dimensional structure of cloud has been estimated. These analytical results in the project reflect our algorithms to estimate the cloud parameters and radiation budget. Less