Topographic Correction for Landsat 8 OLI Vegetation Reflectances Through Path Length Correction: A Comparison Between Explicit and Implicit Methods
Gaofei Yin,Lei Ma,Wei Zhao,Yelu Zeng,Baodong Xu,Shengbiao Wu
DOI: https://doi.org/10.1109/tgrs.2020.2987985
IF: 8.2
2020-12-01
IEEE Transactions on Geoscience and Remote Sensing
Abstract:Topographic correction is a prerequisite for generating radiometrically consistent Landsat 8 OLI vegetation reflectances in support of temporally continuous and spatially mosaicked applications. Path length correction (PLC) is a physically solid topographic correction method that avoids the involvement of any empirical parameter and is therefore suitable for reproducing the inherent reflectance of vegetation. This article compared two different implementation pathways of PLC, i.e., the explicit method (EM) and the implicit method (IM), which are based on the numerical inverse and analytical approximation of the PLC model, respectively. The results show that both EM and IM can obviously reduce the topographic effects on Landsat 8 OLI vegetation reflectances. EM performed slightly better than IM in eliminating the correlation between the topographic characteristics and the vegetation reflectances: the coefficient of determination between the green/red/near-infrared (Nir) band reflectance and the local illumination was reduced from 0.257/0.148/0.467 for the uncorrected (UNCORR) case to 0.016/0.004/0.012 and 0.027/0.014/0.094 for the EM and IM corrected results, respectively. The coefficient of variation of the three band reflectances across different aspects was reduced from 16.5%/18.5%/18.7% for the UNCORR case to 3.2%/1.8%/0.9% and 5.3%/7.1%/7.3% for the EM and IM corrected results, respectively. In addition, the intraclass reflectance variability was also reduced after both the EM and IM corrections. Nevertheless, due to the ill-posed nature of the numerical inverse process, EM cannot fully reproduce the inherent vegetation reflectances, and the reflectances after topographic correction overestimated the inherent vegetation values. In contrast, the IM can achieve an appropriate tradeoff between topographic effect elimination and vegetation inherent reflectance preservation. In addition, IM is computationally very efficient compared to EM: using an ordinary laptop, IM can finish the topographic correction for a Landsat OLI image within several seconds, while this would take more than 20 h for EM. This article highlights the potential of using IM for generating radiometrically consistent Landsat 8 OLI vegetation reflectances.
imaging science & photographic technology,remote sensing,engineering, electrical & electronic,geochemistry & geophysics