Leaf color skewed-distribution parameters enhance the stability of phenotype-environment model across different growth cycles of cabbage

Abstract

Greenhouse cultivation enables high yields through multi-cropping under controlled environments. Research on the association between crop growth and environmental meteorological factors is crucial for achieving precise dynamic regulation of environmental factors and efficient crop growth management within greenhouse. To address the stability problems of inversion model of crop phenotype-accumulated temperature during different sowing dates, this study analyzed the relationship between color skewed-distribution parameters of cabbage canopy and environmental accumulated temperature during different sowing dates. Three types of canopy color parameters (depth, distribution, and mixed parameters) were used as independent variables to construct inversion models of canopy color-accumulated temperature, and the model’s stability was tested across various growth cycles. The results showed that the skewed-distribution parameters of canopy images were significantly correlated with the environmental accumulated temperature, and the correlation coefficient was generally above 0.8. Among the models, the one using distribution parameters as the main independent variable demonstrated the highest fitting accuracy. For the same sowing date, the fitting accuracy was 87.11% and 91.16%, while for different sowing dates, it remained approximately 85%. These findings provide a useful theoretical basis and practical reference for stable and accurate inversion of accumulated temperature based on crop canopy color phenotype during different growth cycles, offering new insights for intelligent, high-quality and efficient management of facility agriculture.

Plant phenotyping relevance

キャベツ冠層画像の色パラメータから積算温度を推定するモデルを構築し、異なる作型・生育周期で安定性を検証しており、表現型取得・推定手法が研究の中心です。

Code and data availability

The paper's cabbage canopy image dataset, CGSD parameter tables, meteorological data, and regression models are not publicly deposited; the Data availability statement says they are available only from the corresponding author on reasonable request. No author code or public repository URL is provided.

No evidence-backed public reproduction asset is currently recorded.