首页 > , Vol. , Issue () : -
GF-4是我国首颗高分辨率静止轨道光学遥感卫星，搭载了一台可见/中波红外凝视相机，由于缺少星上定标系统导致无法提供高频次的星上辐射定标系数。为了更好的实现GF-4/PMS传感器辐射性能监测和数据定量化应用，本文选择了巴丹吉林沙漠南端均匀沙地，利用MODIS作为参考传感器对GF-4/PMS可见近红外波段进行交叉定标。为尽量减少地表不均匀性造成的交叉定标误差，使用较高分辨率的影像（Landsat-8/OLI）搜索巴丹吉林沙漠地区的均匀区域，确定为交叉定标试验区。为进一步减少遥感器成像角度和大气变化的影响，进行了MODIS成像角与GF4成像角角度差约束(小于20?)、成像时刻约束（小于2小时）、云和成像质量问题，共选择2016年至2018年之间有效图像对13组，通过利用MODTRAN模拟计算二者光谱匹配因子进而实现交叉定标系数计算。结果表明：1) 本文计算的辐射定标系数在朗伯体假定和Ross-Li模型假定情况下具有较高一致性，忽略BRDF效应情况造成的计算误差在0.8%~4%之间，定标结果的不确定度在7.4%以内；2) 2016年至2018年GF4/PMS遥感器辐射性能呈现缓慢下降特征，年衰减率小于1%。本文的方法可以有效地提高GF-4/PMS的辐射定标频率和定标精度，实现对GF-4/PMS传感器整个生命周期的辐射性能监测。
GF-4 is the first high-resolution geostationary orbit optical remote sensing satellite in China, equipped with a staring camera covering visible to mid-infrared spectrum. It is pretty to provide on-oribit radiometric calibration coefficients frequently due to lacking on-board calibration device. In order to better monitor the radiometric performance of GF4/PMS and provide the solid basis for quantitative applications, the cross-calibration method was applied to retrieve the calibration coefficients over a select calibration site located in the north of Badan Jaran desert. The cross-calibration site was identified by searching the uniform area in the relative high resolution image (i.e., Landsat-8 /OLI) to minimize the uniform impact on the cross calibration results. To reduce the influence of imaging angles of sensor and atmospheric changing, the restrain was set to the imaging angle difference between MODIS and GF4 (less than 20?), the imaging time difference (less than 2h), the cloud, and the image quality. Totally, 13 image pairs were available from 2016 to 2018 and were used to calculate the cross calibration coefficients after compensating the spectral mathing factors, which were simulated by MODTRAN. The result shows that: 1) the cross calibration coefficients calculated under the Lambertian assumption are highly consistent with that under the Ross-Li BRDF assumption, with uncertainty less than 7.4%, and the relative difference ranges between 0.8% and 4% if the BRDF neglected; and 2) the radiometric performance of GF4/PMS is slowly decreasing from 2016 to 2018, 1% per year. Thus, the proposed method can effectively improve the radiometric calibration frequency of GF-4/PMS with an acceptable accuracy, and can serve the radiometric performance monitoring over the whole life cycle of the sensor.