Practical approaches for normalizing directional solar-induced fluorescence to a standard viewing geometry

REMOTE SENSING OF ENVIRONMENT

卷:255

Article

Hao, Dalei¹  Zeng, Yelu¹  Qiu, Han¹  Biriukova, Khelvi²  Celesti, Marco²  Migliavacca, Mirco³  Rossini, Micol²  Asrar, Ghassem R.⁴  Chen, Min¹ 

[1] Pacific Northwest Natl Lab, Joint Global Change Res Inst, College Pk, MD 20740 USA

[2] Univ Milano Bicocca, Dept Earth & Environm Sci DISAT, Remote Sensing Environm Dynam Lab, Piazza Sci 1, Milan, Italy

[3] Max Planck Inst Biogeochem, Hans Knoll Str 10, D-07743 Jena, Germany

[4] Univ Space Res Assoc, Columbia, MD USA

关键词: Solar-induced chlorophyll fluorescence;    Angular normalization;    Directional effects;    Near-infrared reflectance of vegetation;    Red reflectance of vegetation;    Kernel-driven model;   

DOI  :  10.1016/j.rse.2020.112171

来源: Elsevier

【 摘 要 】

Recent advances in remote sensing of solar-induced chlorophyll fluorescence (SIF) have improved the capabilities of monitoring large-scale Gross Primary Productivity (GPP). However, SIF observations are subject to directional effects which can lead to considerable uncertainties in various applications. Practical approaches for normalizing directional SIF observations to nadir viewing, to minimize the directional effects, have not been well studied. Here we developed two practical and physically-solid approaches for removing the directional effects of anisotropic SIF observations: one is based on near-infrared or red reflectance of vegetation (NIRv and Red(v)), and the other is based on the kernel-driven model with multi-angular SIF measurements. The first approach uses surface reflectance while the second approach directly leverages multi-angular SIF measurements. The performance of the two approaches was evaluated using a dataset of multi-angular measurements of SIF and reflectance collected with a high-resolution field spectrometer over different plant canopies. Results show that the relative mean absolute errors between the normalized nadir SIF and the observed SIF at nadir decrease by 3-6% (far-red) and 6-8% (red) for the first approach, and by 7-13% and 6-11% for the second approach, compared to the original data, respectively. The effectiveness and simplicity of our proposed approaches provide great potential to generate long-term and consistent SIF data records with minimized directional effects.

【 授权许可】

Free