Lai Time Series Reconstruction from Sentinel-2 Imagery Using Vegetation Growing Phenology Feature

Naijie Peng,Siqi Yang,Yunzhu Tao,Dechao Zhai,Wenjie Fan,Qiang Liu
DOI: https://doi.org/10.1109/igarss52108.2023.10281924
2023-01-01
Abstract:As an essential input parameter, Leaf Area Index (LAI) plays significant value in global climate models.. Acquisition of LAI products with extensive long-term series is crucial for various applications. Presently, numerous remote sensing inversion algorithms have been proposed for the generation of LAI products. However, the low temporal resolution is a huge challenge for producing LAI at a medium to high spatial resolution scale. In addition, the cloud has resulted in a significant reduction of accessible data. In this study, a LAI reconstruction method considering vegetation phenological period was developed. A total of 241 Sentinel-2 images from Saihanba in northern China and cloud probability data provided by Sentinel Hub were collected. The results indicate that the proposed method can effectively reconstruct LAI value in the data-missing period and areas. This study evaluated the performance of LAI reconstruction by implementing the "leave-one-out" method. The results demonstrate that the RMSE of LAI reconstruction is 0.5509 when using July 31st data as the validation dataset, and 0.3933 when using August 7th data as the validation dataset. In summary, this study demonstrates the potential for the LAI time series reconstruction from high-resolution data.
What problem does this paper attempt to address?