A New Neighboring Pixels Method for Reducing Aerosol Effects on the NDVI Images

Remote Sensing

Tao Jiang¹  Yunhao Chen¹  Dandan Wang¹  Mengjie Wang¹  Jingling Quan² 

[1] State Key Laboratory of Earth Surface Processes and Resource Ecology, College of Resources Science and Technology, Beijing Normal University, Beijing 100875, China;State Key Laboratory of Resources and Environmental Information System, Institute of Geographical Sciences and Natural Resources Research, CAS, Beijing 100101, China;

关键词: AOD;    aerosol corrected NDVI;    neighboring pixels;    Landsat 8 OLI;   

DOI  :  10.3390/rs8060489

来源: DOAJ

【 摘 要 】

A new algorithm was developed in this research to minimize aerosol effects on the normalized difference vegetation index (NDVI). Simulation results show that in red-NIR reflectance space, variations in red and NIR channels to aerosol optical depth (AOD) follow a specific pattern. Based on this rational, the apparent reflectance in these two bands of neighboring pixels were used to reduce aerosol effects on NDVI values of the central pixel. We call this method the neighboring pixels (NP) algorithm. Validation was performed over vegetated regions in the border area between China and Russia using Landsat 8 Operational Land Imager (OLI) imagery. Results reveal good agreement between the aerosol corrected NDVI using our algorithm and that derived from the Landsat 8 surface reflectance products. The accuracy is related to the gradient of NDVI variation. This algorithm can achieve high accuracy in homogeneous forest or cropland with the root mean square error (RMSE) being equal to 0.046 and 0.049, respectively. This algorithm can also be applied to atmospheric correction and does not require any information about atmospheric conditions. The use of the moving window analysis technique reduces errors caused by the spatial heterogeneity of aerosols. Detections of regions with homogeneous NDVI are the primary sources of biases. This new method is operational and can prove useful at different aerosol concentration levels. In the future, this approach may also be used to examine other indexes composed of bands attenuated by noises in remote sensing.

【 授权许可】

Unknown