Estimating soil and grapevine water status using ground based hyperspectral imaging under diffused lighting conditions: Addressing the effect of lighting variability in vineyards

被引:4
|
作者
Kang, Chenchen [1 ,2 ]
Diverres, Geraldine [3 ]
Paudel, Achyut [1 ,2 ]
Karkee, Manoj [1 ,2 ]
Zhang, Qin [1 ,2 ]
Keller, Markus [3 ]
机构
[1] Washington State Univ, Ctr Precis & Automated Agr Syst, Prosser, WA 99350 USA
[2] Washington State Univ, Dept Biol Syst Engn, Prosser, WA USA
[3] Washington State Univ, Dept Viticulture & Enol, Prosser, WA USA
基金
美国食品与农业研究所;
关键词
Hyperspectral imaging; Diffused lighting; Grapevine; Water stress; Deficit irrigation; Partial least square; Vitis; DEFICIT IRRIGATION; VEGETATIVE GROWTH; REFLECTANCE; ALGORITHMS; PHYSIOLOGY; DROUGHT; STRESS;
D O I
10.1016/j.compag.2023.108175
中图分类号
S [农业科学];
学科分类号
09 ;
摘要
A timely and appropriate level of water deficit is desirable in wine grape production to optimize fruit quality for winemaking. Thus, it is crucial to find a robust and rapid method to assess grapevine water stress in real time. Hyperspectral imaging (HSI) has the potential to detect changes in leaf water status, but the robustness and accuracy are limited in field applications. This study focused on developing ground based approaches for detecting soil and grapevine water status using HSI obtained in diffused lighting conditions. During the 2021 growing season, leaf water potential (& psi;L), stomatal conductance (gs) on the selected leaves and volumetric soil moisture (& theta;v) in the root zone were measured as water status indicators. Spectral data from diffused and direct sunlight conditions were obtained to construct models to estimate plant and soil water status indicators. Partial least squares (PLS) regression models were individually developed to estimate & psi;L, gs, and & theta;v using spectra obtained from direct/diffused lighting conditions, respectively. The results indicated that the & psi;L estimation model using spectral data from diffused lighting performed better than that obtained using direct sunlight, indicated by a higher R2 (0.89 vs. 0.82), a lower RMSE (0.12 vs. 0.15 MPa) and a lower MAE (0.10 vs. 0.11 MPa). The model developed for estimating & theta;v using spectral data under diffused lighting achieved superior performance to the one in direct sunlight in terms of R2, RMSE and MAE (0.90 vs. 0.89 and 1.56 vs. 1.59 %, 1.26 vs. 1.29 %). These results demonstrated that spectral data obtained under diffused lighting can improve model performance by providing a more uniform illumination. Ground based HSI was capable of high-resolution sensing of grapevine water status by estimating & psi;L and gs and map variability within canopies.
引用
收藏
页数:13
相关论文
共 1 条
  • [1] Assessment of unified models for estimating potato leaf area index under water stress conditions across ground-based hyperspectral data
    Luo, Shanjun
    He, Yingbin
    Li, Qian
    Jiao, Weihua
    Zhu, Yaqiu
    Yu, Jinkuan
    Zhao, Chang
    Xu, Ruiyang
    Zhang, Shengli
    Xu, Fei
    Sun, Jing
    Han, Zhongcai
    Li, Chuang
    JOURNAL OF APPLIED REMOTE SENSING, 2020, 14 (01):