湖北农业科学 ›› 2026, Vol. 65 ›› Issue (9): 194-199.doi: 10.14088/j.cnki.issn0439-8114.2026.09.031

• 信息工程 • 上一篇    下一篇

玉米农田日光诱导叶绿素荧光反演研究——以辽宁省为例

余鹏程1, 屈亚维1, 任慈2, 徐洁1, 骆元家1, 谢光雄1   

  1. 1.柳州工学院土木建筑工程学院,广西 柳州 545616;
    2.广西大学土木建筑工程学院,南宁 530004
  • 收稿日期:2026-06-10 出版日期:2026-09-25 发布日期:2026-09-17
  • 通讯作者: 任慈(1997-),女,辽宁锦州人,硕士,主要从事结构工程方向研究,(电子信箱)1335598767@qq.com。
  • 作者简介:余鹏程(1997-),男,河南信阳人,助教,硕士,主要从事农业生态遥感研究,(电子信箱)924011360@qq.com;共同第一作者,屈亚维(2005-),男,陕西渭南人,在读本科生,专业方向为农业生态遥感,(电子信箱)2036837304@qq.com
  • 基金资助:
    柳州工学院大学生创新创业训练计划项目(S202513639061)

Study on inversion of solar-induced chlorophyll fluorescence in maize farmland: A case study of Liaoning

YU Peng-cheng1, QU Ya-wei1, REN Ci2, XU Jie1, LUO Yuan-jia1, XIE Guang-xiong1   

  1. 1. School of Civil Engineering and Architecture, Liuzhou Institute of Technology, Liuzhou 545616, Guangxi, China;
    2. School of Civil Engineering and Architecture, Guangxi University, Nanning 530004, China
  • Received:2026-06-10 Published:2026-09-25 Online:2026-09-17

摘要: 日光诱导叶绿素荧光(SIF)是表征植物光合作用状态的重要探针,也是估算植被总初级生产力的重要指标。气象条件通过影响植物生理状态和光合作用过程,进而影响SIF变化。为实现不同气象条件下SIF反演,以辽宁省玉米种植区为研究对象,利用2014—2023年辽宁省21个气象站点数据、基于山地小气候模拟模型获得的气象参数、基于OCO-2数据构建的全球SIF产品(GOSIF)以及气象栅格数据,采用偏最小二乘回归(PLSR)、主成分分析(PCA)、全连接神经网络(FCNN)和岭回归(RR)方法对干旱年和降水适宜年5—9月SIF进行反演。结果表明,站点尺度模型对比中PLSR模型的验证集精度最高,分城市区域RR模型在多数地区取得较好精度,且训练集R2多数达到0.65以上(P<0.05)。除本溪市和丹东市外,其余地区归一化均方根误差(NRMSE)均低于20%。反演结果与GOSIF产品的空间相关性分析表明,多数区域的空间相关系数达到0.7以上,二者空间变化趋势总体一致。通过筛选月平均实际蒸散量、月平均气温、月平均水汽压和月降水量作为输入变量,可有效构建区域玉米农田SIF的反演方法,为利用气象数据开展区域尺度SIF估算与智慧农业遥感应用提供技术参考。

关键词: 日光诱导叶绿素荧光, 干旱, 玉米, GOSIF, 辽宁省

Abstract: Solar-induced chlorophyll fluorescence (SIF) is an important indicator of plant photosynthetic status and a key metric for estimating vegetation gross primary productivity (GPP). Meteorological conditions influence SIF variations by affecting plant physiological status and photosynthetic processes. To achieve SIF inversion under different meteorological conditions, the maize planting areas in Liaoning Province were selected as the study area. Using data from 21 meteorological stations in Liaoning Province during 2014—2023, meteorological parameters simulated by the Mountain Microclimate Simulation Model, the Global SIF product (GOSIF) derived from OCO-2 data, and meteorological gridded datasets, SIF values during May-September in drought years and precipitation-suitable years were inverted using partial least squares regression (PLSR), principal component analysis (PCA), fully connected neural network (FCNN), and ridge regression (RR) methods. The results showed that the PLSR model achieved the highest validation set accuracy in site-scale model comparisons, while the city-level regional RR model obtained good accuracy in most regions, with the coefficient of determination (R2) of the training set mostly reaching above 0.65 (P<0.05). Except for Benxi and Dandong cities, the normalized root mean square error (NRMSE) was below 20% in other regions. Spatial correlation analysis between the inverted SIF results and GOSIF products showed that the spatial correlation coefficients in most regions exceeded 0.7, and the spatial variation trends of both were generally consistent. By selecting monthly mean actual evapotranspiration, monthly mean air temperature, monthly mean water vapor pressure, and monthly precipitation as input variables, the SIF inversion method for regional maize farmland could be effectively established. This provided a technical reference for regional-scale SIF estimation using meteorological data and smart agriculture remote sensing applications.

Key words: Solar-induced chlorophyll fluorescence, drought, maize, GOSIF, Liaoning province

中图分类号: