叶面积指数
索引(排版)
作物
农业工程
计算机科学
农学
生物
工程类
万维网
作者
Yufan Zhang,Koki Homma,Liangsheng Shi,Xingyu Wang,Qiao Han,Yuanyuan Zha
标识
DOI:10.1016/j.agsy.2025.104357
摘要
CONTEXT Specific leaf area (SLA) is an important trait for quantifying crop growth status and leaf physiological structure. It also serves as a vital crop-specific parameter in the derivation of leaf area index (LAI) and aboveground biomass within crop models. OBJECTIVE The simplified and empirical consideration of SLA in current models ignores its dynamics. To accurately characterize the leaf structural traits, the SLA in the model need to be modified as an updatable continuous variable. METHODS This study used a dual-branch neural network for processing two-view digital images to extract leaf-specific traits. A novel approach was then proposed to modify the SLA from a fixed sequence to an updatable variable by introducing the SLA dynamic function (SDF) into WOFOST model. Meanwhile, a data assimilation framework was developed based on the Ensemble Kalman Filter , enabling SLA and LAI observations to be uptaken simultaneously. A two-year drought-controlled experiment was conducted on rice to evaluate the method under varying water stress conditions. RESULTS AND CONCLUSIONS Results demonstrated that the incorporation of SDF and the assimilation of SLA and LAI significantly improved the accuracy of LAI, aboveground biomass , and grain yield estimations compared to the original model (R 2 = 0.85, RMSE = 1310.05 kg⋅ha −1 ). The incorporation of SLAs further improves model performance and highlights the complementary roles of SLA and LAI in data assimilation . SIGNIFICANCE The technical route developed in this study aims to provide a concise and efficient solution for the integration of crop growth simulation and observation while highlighting the importance of considering dynamic leaf traits in crop modeling to better capture the responses to environmental stresses such as drought.
科研通智能强力驱动
Strongly Powered by AbleSci AI