【摘要】 针对作物光合速率受到作物自身生理状况和外部环境因素的共同影响,现有模型对不同生理状态叶片光合速率预测中误差较大的问题,将PSⅡ光化学活性这一可表征植物叶片状态的荧光参数引入模型,提出融合PSⅡ光化学活性的茄子光合速率模型建模方法。以不同生长状态的茄子叶片作样本,设计多环境参数耦合条件下的光合速率测试试验,获取1 320组光合速率数据。以上述数据作为样本集,在通过网格搜索完成模型参数优化的基础上,建立基于随机森林的光合速率预测模型。模型均方根误差为1.266 8μmol∕(m~2·s),决定系数为0.950 7。异校验结果表明,该模型预测值与真实值拟合斜率为0.952,截距为0.02,优于不含暗荧光参数的光合速率预测模型,说明引入PSⅡ光化学活性(F_v∕F_o)可有效提高光合速率模型的精度。更多还原

【Abstract】 Photosynthesis process is affected by both external environmental conditions and crop’s physiological status.The existing models appeared to have larger errors when dealing with crops in different growth states.To solve this problem,this study introduced a chlorophyll fluorescence parameter F_v/F_o,which represented the potential activity of PS Ⅱ,as an input,and constructed an eggplant photosynthetic rate prediction model.Taking eggplant plant as the object,we obtained eggplant leaves in different growth states by setting environment gradients.The chlorophyll fluorescence data and photosynthetic rates under different combinations of environmental factors were measured,in total of 1 320 sets of F_v/F_o and the corresponding photosynthetic rates were obtained as the data set.Grid search method was adapted to select the key parameter of the model.Thereby,a photosynthetic rate prediction rate model based on RF was established.The validation result showed that model was of high accuracy with the RMSE of 1.266 8 μmol/(m~2·s) and the R~2 of 0.950 7.The fitting line of the measured and predicted photosynthetic rate had a slope of 0.952 and an intercept of 0.02.Compared with the model only using environmental factors as the inputs,the prediction accuracy of this model was higher,proving that the introduction of F_v/F_o effectively increased the model’s prediction ability for samples of different growth states.更多还原