A way to extend the SIF extraction bandwidth for application of crop phenotyping: Considering canopy structure and illumination effects

Abstract

Solar-induced chlorophyll fluorescence (SIF) from hyperspectral imaging is strongly associated with agricultural indices, particularly leaf chlorophyll content (LCC), in crop phenotyping. However, confounding factors such as spectral resolution (SR), canopy structure, and illumination reduce SIF sensitivity and complicate its association with these indices. This study focused on examining the effect of SR and mitigating the influence of canopy structure and illumination to enhance SIF accuracy. This study employed theoretical simulations, experimental validation, and two field experiments comparing devices with varying SRs to extract SIF at 687 nm and 761 nm (SIF Red and SIF NIR ). Initially, SCOPE model was used to simulate radiance trends across different SRs. Secondly, narrow-band imager with 7 nm full width at half maximum (FWHM) were then compared with radiance from sub-nanometer and ASD spectrometers (0.3 nm and 3 nm FWHM). Finally, to modify SIF measurements, canopy structure was quantified using fluorescence escape fraction (f esc ), while absorbed photosynthetically active radiation (APAR) and intensity component from hue-saturation-intensity color model (I HSI ) were used to account for illumination. And driver factor for SIF extraction was explored. Results showed a strong relationship (R 2 ≥ 0.95) between original simulated radiance from sub-nanometer spectrometers and radiance resampled to 3, 5, and 7 nm FWHM. SIF NIR from the narrow-band imager correlated with the other spectrometers (R ≥ 0.64). While lower SR reduced SIF sensitivity, the narrow-band imager still showed potential for estimating LCC. Modified SIF NIR by f esc , APAR, and I HSI , improved correlations with LCC (R = 0.77, 0.74, and 0.86) compared to unmodified SIF NIR (R = 0.65). Estimation model based on SIF modified by f esc , APAR and I HSI presented R P 2 of 0.73, 0.75, 0.84 and outperformed SIF (R P 2 = 0.52), and photosynthetically active radiation and APAR showed R of 0.52 and 0.72 with SIF NIR , which indicated that disentangling confounding factors could enhance the sensitivity of SIF and structural effect governed SIF strongly than illumination effect. This study demonstrated the feasibility of narrow-band imagers for capturing spatial–temporal SIF variability by addressing confounding factors and can provide guidance for narrow-band instrument development for SIF extraction.

Plant phenotyping relevance

狭帯域ハイパースペクトル画像からSIFを抽出し、冠層構造・照明の影響を補正して葉緑素含量を推定する方法を、シミュレーションと実験で検証・改良しており、植物表現型取得が中心である。

Code and data availability

公開論文であることは確認できましたが、現在の公式API・許可済み取得経路では本文を自動取得できませんでした。

No evidence-backed public reproduction asset is currently recorded.