Cross-Calibration Of HJ1-CCD

Radiometric calibration is a process that converts recorded sensor voltages or digitized counts to an absolute scale of radiance that is independent of the image forming characteristics of the sensor. The commonly used high pricisely absolute radiometric calibration method is the field calibration, but it's labor-consuming, requires too much field measuring data and cost a lot of human, material and financial resources, so it only can give a few calibration parameters. The other method for radiometric calibration is cross-calibration which used calibrated sensor to calibrate other sensors, so it is a promising method. However, due to the roughness of the actual surfaces, most of the land objects are not Lambet, differect view-solar geometry has different reflectance, the bidirectional reflectance is important for calibration. For most of the high/moderate resolution remote sensing images, which are small width and vertical observation, so it is difficult to get land surface bidirectional reflectance using these data.To solve the above problem, a methodology was proposed to extract the feature of Bi-directional Reflectance Distribution Functions (BRDF) over Badain Jaran Desert using Landsat-TM/ETM+ and ASTER GDEM data, and simulate surface reflectance for 25 scenes of different Landsat-TM and ETM+ images and compare the simulations with the actual Landsat-TM and ETM+ imageries to verify this method, the result shows that this method precision is higher than the radiometric calibration error requirement of 5% to 7%. Therefore, the nadir viewing Landsat-TM/ETM+ imagery with ASTER GDEM data together can be used to cross-calibrate remotely sensed imagery with large viewing observation, such as HJ1-CCD imagery.Firstly, ASTER GDEM productions were used to retrieve the sun-view geometry of every slope element in remotely sensed imagery, and get surface reflectance data from 13 Landsat7-ETM+ images after atmospheric correction, then extract BRDF feature based on the statistic of these data.Secondly, the BRDF feature data was used to simulate HJ1-CCD images after verification. The surface reflectance was simulated according to HJ1-CCD each pixel's local system view-solar geometry based on the BRDF feature data. We can get apparent radiance using atmospheric parameters. Compared the apparent radiance with DN number get from CCD images, and the cross-calibration parameters can be got. The method of BRDF feature extraction this paper proposed mainly solve the problem that using the vertical observation moderate/high resolution remote sensing data to simulate the BRDF in desert area, this method can be used to fit the BRDF feature of land surface accurately, then the remote sensing images in different land surface conditions can be simulated. This can give an available option for cross-calibration between high and moderate sensors.