厚皮甜瓜叶面积预测模型研究

Research on the Prediction Model of Leaf Area of Muskmelon

[目的]探讨厚皮甜瓜叶面积快速准确的非破坏性测定方法。[方法]以长香玉、秋月、激情3个品种为材料,测量叶长、叶宽、叶基到叶尖的距离及缺刻指标等叶片形态特征指标和叶片面积。[结果]叶面积与叶长、叶宽、叶长×叶宽及缺刻指标之间存在显著的相关关系;通过模型拟合和优选,得到了以大小变量B·C、缺刻变量A·E(1/2)/F和丰缺度变量D-C预测厚皮甜瓜叶面积的数学模型为S=0.660 5B·C-1.953 A·E(1/2)/F+3.001(D-C)+6.806,模型拟合的决定系数R2为0.974 3、均方根误差RMSE为8.487;模型外部验证结果,相关系数r为0.985 3、均方根误差RMSE为8.789。[结论]在预测模型之中导入缺刻和丰缺度变量,可以提高预测叶面积的准确度、增加模型的适应性,此模型可以应用于厚皮甜瓜叶面积的非破坏性测定。

[Objective]The research aimed to discuss the non- destructive determination method of the leaf area of muskmelon quickly and accurately. [Method]To Changxiangyu,Qiuyu and Jiqing for materials,the leaf characteristics of three varieties of muskmelon,including leaf length,leaf width,leaf base to tip length,nick index and leaf area were measured. [Result]There was a significant correlation between leaf area and leaf length,leaf width,leaf length × leaf width and nick index. Through the model fitting and optimization,we got a mathematical model S =0. 660 5B·C- 1. 953 A· E（1/2） /F ＋ 3. 001（ D- C） ＋ 6. 806,which consists of the size variable B·C,the nick variable A·E（1/2） /F and the abundance and deficiency variable D- C to predict leaf area. Model fitting results showed that the determination coefficient R2 was 0. 974 3,RMSE was 8. 487. Finally,the results of the model external validation that the correlation coefficient was 0. 985 3,RMSE was 8. 789. [Conclusion] Introducing the nick and fullness into the model can improve the accuracy of predicting the leaf area and increase the adaptability of the model,and the model can be used in the non- destructive determination of the leaf area of muskmelon.