| GF1 satellite is the first satellite of China’s high-resolution Earth observation system and plays an important role in many fields.Now,the calibration of GF1 satellite adopts the calibration coefficient published annually on the official website by the China Resources Satellite Application Center,and the limited number of calibrations makes it difficult to monitor the changes in the radiation performance of satellite sensors,and it is difficult to ensure the quality of remote sensing data by using a fixed calibration coefficient throughout the year.Therefore,it is of great significance to carry out time series calibration studies to monitor the orbital operation status of satellite sensors and to improve the accuracy of calibration.At present,the method of obtaining the time series calibration coefficient is based on a single scene image to obtain an independent calibration coefficient,which does not consider the change law of the radiation performance of the sensor with time,and the orbital running state of the sensor is actually a relatively stable,continuous gradient process,and there should be a certain correlation between the calibration of different moments.In order to monitor the changes in the radiation performance of satellite sensors,improve the accuracy of radiation calibration,and meet the application needs of quantitative remote sensing in the new era.In view of the above questions,this paper will combine the correlation between the time series calibration coefficients to carry out the time series calibration research of GF1 satellite.Taking the Dunhuang National Calibration Field as the benchmark calibration field and the Dazaohuo East Correction Field as the verification field,the GF1-WFV sensor images of the site between 2013 and 2021 are obtained,drawing on the idea of Kalman filtering,and the time series calibration coefficient obtained by the time series calibration method based on the stable target will be reprocessed to realize the time series re-calibration,and obtain the final time series calibration coefficients.Using linear fitting to construct the attenuation model of satellite sensor calibration coefficients,and to analyze the radiation performance of satellite sensor.Finally,the calibration results are verified and evaluated.The main findings are as follows:（1）Re-calibrating based on Kalman filtering can obtain more stable and continuous calibration coefficientsthe volatility of the time series calibration coefficient after re-calibrating is significantly reduced than before re-calibrating,and the time series calibration coefficient is more continuous and stable,which is more in line with the orbital operation state of the sensor: it is a stable,continuous,and gradual change with time process.The relative difference in the time series calibration coefficients of the four WFV sensors in different sites is not more than 10.13%,5.21%,4%,and 4%,respectively,indicating that in addition to the WFV1 sensor,the calibration results of the other three sensors in different sites have good consistency,and most of the results show that the time series calibration method in this paper is more reliable.（2）The radiation performance of the WFV sensor does not decay significantly during orbital operationBased on the sensor attenuation model based on the time series calibration results,the annual decay rate of the four WFV sensors does not exceed 0.22%,0.11%,0.27%,and 0.26%,indicating that the radiation performance of the four sensors is relatively stable during orbital operation,and no significant decay occurs.（3）The calibration results in this article improve the accuracy of the calibrationUsing the time series calibration coefficient fitting formula obtained in this paper to calculate the calibration coefficient,and comparing it with the calibration coefficient published by the China Centre for Resources Satellite Data and Application,it is found that most of the calibration results obtained in this paper are more consistent with the calibration results of the China Centre for Resources Satellite Data and Application,and the relative difference is within 5%,but there are also large differences in individual bands in individual years.From the scatter plot of multi-year calibration results,it can be seen that the multi-year calibration results of this paper show stable changes,while the site calibration results of the Resource Satellite Center show obvious fluctuations in individual bands in individual years.And further in view of the large difference between the two,the calibration results of July-August in this paper and the calibration results of the Resource Satellite Center are compared with the calibration results of Landsat8,which can be seen that the calibration error in this paper is smaller than the calibration error of the Resource Satellite Center,which improves the accuracy of the calibration,and explains that the calibration coefficient calculated by the time series calibration coefficient fitting formula reduces the accidental error that may exist in a single calibration.Using the Landsat8 calibration results to verify the calibration coefficients under different months of the year,it can be seen that the accuracy of the calibration results in this paper is higher than the calibration results released by the Resource Satellite Center,which explains the necessity of time series calibration,and the calibration coefficient corresponding to the time is more accurate than the unified calibration coefficient used throughout the year. |
