Identification of Early Heat and Water Stress in Strawberry Plants Using Chlorophyll-Fluorescence Indices Extracted via Hyperspectral Images.

Abstract

Strawberry (Fragaria × ananassa Duch) plants are vulnerable to climatic change. The strawberry plants suffer from heat and water stress eventually, and the effects are reflected in the development and yields. In this investigation, potential chlorophyll-fluorescence-based indices were selected to detect the early heat and water stress in strawberry plants. The hyperspectral images were used to capture the fluorescence reflectance in the range of 500 nm-900 nm. From the hyperspectral cube, the region of interest (leaves) was identified, followed by the extraction of eight chlorophyll-fluorescence indices from the region of interest (leaves). These eight chlorophyll-fluorescence indices were analyzed deeply to identify the best indicators for our objective. The indices were used to develop machine-learning models to assess the performance of the indicators by accuracy assessment. The overall procedure is proposed as a new workflow for determining strawberry plants' early heat and water stress. The proposed workflow suggests that by including all eight indices, the random-forest classifier performs well, with an accuracy of 94%. With this combination of the potential indices, namely the red-edge vegetation stress index (RVSI), chlorophyll B (Chl-b), pigment-specific simple ratio for chlorophyll B (PSSR b ), and the red-edge chlorophyll index (CI REDEDGE ), the gradient-boosting classifier performs well, with an accuracy of 91%. The proposed workflow works well with a limited number of training samples which is an added advantage.

Plant phenotyping relevance

ハイパースペクトル画像から葉の蛍光指標を抽出し、イチゴの熱・水ストレスという植物状態を推定するワークフローを開発・評価しており、表現型取得と解析手法が中心である。

Code and data availability

The article describes hyperspectral imaging of strawberry plants and machine-learning analysis, but the Data Availability Statement explicitly states that no data are shared, and no public dataset, image repository, or author code URL is provided anywhere in the supplied blocks. The only URLs present are ORCID author i

No evidence-backed public reproduction asset is currently recorded.