Cross-environment stable leaf canopy skewness-kurtosis indices: Developing transferable biometric correlates for agroclimatic phenotyping models

Abstract

This study proposes a novel, robust and scalable approach for color phenotype modeling, providing a new research perspective for crop phenotype analysis. The color gradation skewness-distribution (CGSD) parameters, as a color characterization indicator that can be widely applied to different crops and ecological environments, possess significant theoretical value and practical application potential. Specifically, we explore the relationship between accumulated temperature—the primary heat factor driving crop development—and crop canopy leaf color, aiming to identify canopy color parameters that can consistently describe crop responses to environmental changes. In this study, we developed and tested inversion models for predicting accumulated temperature in wheat and tobacco crops grown in both laboratory and natural environments across various ecological regions. These models utilized color gradation skewness-distribution parameters derived from digital canopy images. Our results show that some inversion models can predict accumulated temperature responses with high accuracy, achieving 88.95% accuracy for wheat and 77.38% for tobacco. Statistical analysis revealed that, compared to models using parameters related to color depth, those incorporating parameters related to leaf color distribution as independent variables provided more consistent predictions across crops from different ecological regions. This can be explained from the fact that the leaf color distribution parameters are relative values, which are less affected by regional ecological variations than parameters associated with the image color depth, which depend on the absolute value of the color level of leaf image pixels. Our study confirms that the CGSD parameters extracted from the color information contained in digital canopy images can provide a novel approach for accurate monitoring and evaluation of crop growth in diverse ecological environments.

Plant phenotyping relevance

デジタル群落画像から葉色分布指標を抽出し、環境応答の推定モデルを開発・検証しており、植物表現型の取得・計算手法が中心である。

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.