中文摘要：

卫星资料辐射定标精度是其定量应用的关键因素。以METOP-A/IASI的高光谱探测资料为传递基准,利用同时星下点观测的交叉定标方法,对FY-3A/MERSI热红外通道的在轨辐射定标精度进行了客观评估,并给出了亮温系统偏差的订正因子。从观测时间差异、卫星观测天顶角和方位角差异、以及目标均匀性四个方面,分析了交叉定标中所用主要匹配近似因子的不确定性。分析结果表明,目标均匀性是匹配误差的主要来源,偏差不确定性小于2%（当亮温偏差约为1K时,不确定性〈0.02K）,其他因素的影响可以忽略。一年多的样本统计及偏差分析结果显示,MERSI的观测亮温明显高于IASI,年平均亮温偏差约（3.18±0.34）K,月平均亮温偏差呈现季节波动特征,波动幅度约0.8K。与相近时期敦煌场和青海湖地同步观测评价结果有非常好的一致性。初步原因分析推断,造成MERSI亮温偏高的原因主要有两个,一是星上黑体发射率被高估,二是光谱响应函数向大气窗区漂移,后者可能为主导因素。

英文摘要：

Accurate satellite radiance measurements are significant for data assimilations and quantitative retrieval applications.In the present paper,radiometric calibration accuracy of FungYun-3A（FY-3A）Medium Resolution Spectral Imager（MERSI）thermal infrared（TIR）channel was evaluated based on simultaneous nadir observation（SNO）intercalibration method.Hyperspectral and high-quality measurements of METOP-A/IASI were used as reference.Assessment uncertainty from intercalibration method was also investigated by examining the relation between BT bias against four main collocation factors,i.e.observation time difference,view geometric difference related to zenith angles and azimuth angles,and scene spatial homogeneity.It was indicated that the BT bias is evenly distributed across the collocation variables with no significant linear relationship in MERSI IR channel.Among the four collocation factors,the scene spatial homogeneity may be the most important factor with the uncertainty less than 2% of BT bias.Statistical analysis of monitoring biases during one and a half years indicates that the brightness temperature measured by MERSI is much warmer than that of IASI.The annual mean bias（MERSI-IASI）in 2012is（3.18±0.34）K.Monthly averaged BT biases show a little seasonal variation character,and fluctuation range is less than 0.8K.To further verify the reliability,our evaluation result was also compared with the synchronous experiment results at Dunhuang and Qinghai Lake sites,which showed excellent agreement.Preliminary analysis indicates that there are two reasons leading to the warm bias.One is the overestimation of blackbody emissivity,and the other is probably the incorrect spectral respond function which has shifted to window spectral.Considering the variation character of BT biases,SRF error seems to be the dominant factor.