摘要
在传统的长宽系数法的基础上,探讨通过叶片长宽和缺刻深度估算叶面积的可行性,以期为生产和科研提供适应性广和准确度较高的厚皮甜瓜叶面积测定方法。以‘长香玉’‘秋月’‘激情’‘红姑’4个品种为材料,测量叶长、叶宽、缺刻深浅等叶片形态特征指标并采用扫描法测定叶面积,共获得1 298份叶片数据。结果表明,叶面积与叶长、叶宽、缺刻深浅之间存在显著的相关性;通过模型拟合和优选,得到以反映叶片大小的变量L·W及反映叶片缺刻的变量N_1·L和N_2·L为自变量构建的厚皮甜瓜叶面积的数学模型S=-13.2897+(L·W)^(0.943)+0.267N_1·L+0.153 N_2·L的预测效果较好,模型拟合结果决定系数R^2和均方根误差RMSE分别为0.988 6和7.686,模型外部验证结果相关系数r和均方根误差RMSE分别为0.995 0和7.201;各个品种外部验证结果,预测值与实测值之间的相关系数r均在0.994 0以上,均方根误差RMSE均小于6.900;在模型中添加缺刻变量后,均方根误差RMSE由9.434降低到7.201。因此,在模型之中导入缺刻变量后可以提高预测叶面积的准确度,增强模型对不同叶形和不同品种叶片的适应性。
In order to determine the leaf area of muskmelon more accuracy and adaptability in non-destructive way during the process of production and scientific research, the feasibility of the method through the leaf size variable leaf length, leaf width and the notch variable to estimate leaf area of muskmelon was explored on the basis of traditional regression coeffi- cient method. Leaf characteristics including leaf length, leaf width, notch depth and leaf area. Leaf length, leaf width and the notch depth of 1 298 muskmelon leaves from 4 cuhivars of muskmelon ＇Changxiangyu＇, ＇Qiuyue＇, ＇Jiqing＇ and ＇Hon- ggu＇ were measured and the leaf area were determined by scanning method. The results indicated that there was a signifi- cant correlation between leaf area and leaf length, leaf width aod notch depth. Through the model fitting and optimization, a optimizing mathematical model S= -13.289 7＋（L. W）0.94＋0.267N1 .L＋0.153 N2 L, which consisted of the size variable L. W and the notch variable N1 L and N2 L to predict leaf area was obtained. Model fitting results showed that the coefficient of determination R2 and the root mean square error RMSE were 0.988 6 and 7.686, respectively. The results of external valida- tion of the model reflected that the correlation coefficient r and the root mean square error RMSE were 0.995 0 and 7.201, respectively. Model validation was carried out on the 4 melon varieties respectively, and the correlation coefficient r which was between predicted values and the measured values of the 4 validation were all higher than 0.9940, meanwhile the root mean square error RMSE were all less than 6.900. When the notch variable added into the model, the root mean square er- ror RMSE reduced from 9.434 to 7.201. The accuracy of predicting the leaf area and adaptability of the model were im- proved by introducing the notch variable.
出处
《中国瓜菜》
CAS
北大核心
2016年第5期25-28,共4页
China Cucurbits And Vegetables
基金
海南耕地改良关键技术研究与示范专项"水稻-瓜菜-绿肥等轮作模式研究与推广"(HNGDg1201504)
