Evaluation of HJ-1A/B CCD Surface Reflectance Products Using the VNIR and MODIS-Based Atmospheric Correction Approaches

The growing development of medium- to high-resolution satellites in China has led to a considerable increase in the quantitative applications using the data; therefore, it is important to produce standard surface reflectance (SR) products operationally from such data. However, there is still a lack of relevant SR systems and SR products. We applied two atmospheric correction algorithms, adaptive to most multispectral satellites with visible and near-infrared (VNIR) bands, to HJ-1A/B charge-coupled device (CCD) instrument data, namely, the VNIR method and the MODIS-based method, with both methods being based on the Second Simulation of the Satellite Signal in the Solar Spectrum, Vector (6SV) code, and the look-up tables. We evaluated the accuracy of the SR by these two approaches for HJ-1A/B CCD images compared with the AErosol RObotic NETwork (AERONET) corrected reflectance for the period July 2011 to June 2012 over China and surrounding regions. We assessed more than 12 million pixels for the 49 spatial circular subsets, with a radius of 5 km, centered at 12 AERONET sites. The evaluation results indicated that both methods were suitable for operational flow, and that the MODIS-based method had better accuracy than the VNIR method, except for the near-infrared band. This conclusion was also validated by comparison with the normalized difference vegetation index products derived from the MODIS-based SR, VNIR SR, and AERONET SR. Additionally, the MODIS-based method showed superior accuracy when the overpass time of HJ-1A/B was more approximate to that of Terra MODIS.