中国区域DMSP-OLS与NPP-VIIRS夜间灯光影像校正

本文基于对传统不变目标区域校正法的改进，选择4期辐射定标参考影像对DMSP-OLS影像进行了饱和校正与时间序列连续性校正。通过阈值法去除了NPP-VIIRS影像的背景噪声及异常值，利用对数变换压缩了其灯光辐射值的动态范围，使其亮度分布更接近于DMSP-OLS数据。利用BiDoseResp函数模型对NPP-VIIRS影像进行了一致性校正，获得了1992-2019年长时间序列夜间灯光影像。从定性、定量及校正影像与社会经济参量的相关性方面对校正结果进行了精度验证。结果表明，改进后的校正方法改善了DMSP-OLS影像饱和严重的现象，使其获得了长时间序列纵向可比性。BiDoseResp函数模型可以很好地拟合DMSP-OLS和对数变换后的NPP-VIIRS之间的回归。校正后的长时间序列灯光影像数据与社会经济参量高度契合。校正结果对推广长时间序列夜间灯光影像的广泛应用、深化长时间序列科学问题的研究具有重要价值。

DMSP-OLS and NPP-VIIRS night light image correction in China

Based on the improvement of the traditional invariant target area correction method, this study selects the four-phase radiometric calibration reference image to perform saturation correction and time series continuity correction for DMSP-OLS images.The threshold method was used to remove the background noise and abnormal values of the NPP-VIIRS image, and the dynamic range of the light radiation value is compressed by the logarithmic transformation so that the brightness distribution is closer to the DMSP-OLS data.Then the NPP-VIIRS image was uniformly corrected through the BiDoseResp function model, and the long-time sequence of night light images from 1992 to 2019 was obtained.Finally, the accuracy of the calibration results is verified in terms of qualitative, quantitative, and the correlation between the corrected image and the socio-economic parameters.The results show that the improved correction method improves the severe saturation of DMSP-OLS images, and makes it obtain the longitudinal comparability of long-term series.The BiDoseResp function model can fit the regression between DMSP-OLS and logarithmic transformation NPP-VIIRS well.The corrected long-term sequence of light image data is highly consistent with the socio-economic parameters. The correction results are of great value to promote the wide application of long-time sequence night light images and to deepen the research of long-term sequence scientific problems.