| 摘要: |
| 【目的】为掌握花生叶面积指数（LAI）及其动态变化，并用于其生长监测和作物估产， 【方法】2019年在郑州市农业气象试验站进行设置3个不同氮素水平处理的花生试验，利用便携式光谱仪ASD HandHeld 2测量花生冠层高光谱数据，使用LAI2200冠层分析系统采集花生叶面积指数。对原始光谱数据进行倒数对数和导数变换，并选取常见高光谱特征参数建立LAI估算模型，通过模型精度比较从中选择出最优估算模型。【结果】结果显示，LAI单变量估算模型中大部分为指数模型的R2较大，其中VI1（VI1=Rg/Rr。Rg绿峰反射率，Rr红谷反射率）、VI2（VI2=(Rg-Rr)/(Rg+Rr)）和Rr的指数模型R2超过0.68，拟合程度最高。同时利用检验样本计算均方根误差（RMSE），VI2的RMSE值最小，其次是VI1。【结论】综上，以VI2为自变量的指数模型拟合与预测精度均为最高，因此认为以VI2为自变量的指数模型最优，其次为VI1。最后通过多元逐步回归，建立了花生LAI估算的多元回归模型，分析表明多元回归模型模拟精度高于单变量估算模型，当多种光谱参数均可获取的情况下，优先选用多元回归模型对花生冠层LAI进行估算。 |
| 关键词: 花生;叶面积指数;高光谱遥感;模型精度 |
| DOI： |
| 分类号:Ｓ５１２．１；Ｓ３１４ |
| 基金项目:河南省自然科学青年基金（编号：202300410531）；中国气象局农业气象保障与应用技术重点开放实验室开放研究基金（编号AMF201901；AMF201801） |
|
| Estimation of Peanut LAI Based on Canopy Hyperspectral Remote Sensing Date |
|
LI Jun-ling,LI Meng-xia
|
|
Henan Key Laboratory of Agrometeorological Ensuring and Applied Technique,CMA
|
| Abstract: |
| 【Objective】To explore the relationship between hyperspectral reflectance and leaf area index (LAI), 【Method】the experiment was conducted from 2019 in Zhengzhou agricultural meteorological experimental station. Peanut canopy hyperspectral data was measured at different growth stages by using the ASD HandHeld 2, Peanut LAI was collected at the same time by using LAI2200 canopy analysis system. By making the correlation analysis between the original, the speccountdown logarithms, the derivative spectral data and LAI, Peanut LAI estimation model was obtained, and the optimal estimation model was selected through the comparison of model accuracy. 【Result】The results showed that the exponential model of the most parameters in single spectral variables estimated LAI model were with larger R2, including VI1, VI2, Rr parabolic model more than 0.6, the fitting degree were the highest, while the RMSE of VI2 were minimal, followed by VI1. 【Conclusion】So that VI2 is the optimal variable parabola model, and the prediction and forecasting precision are preferable. Analysis of comprehensive multiple stepwise regression model, the sensitive spectral indices for estimating leaf area index of peanut were VI2 and VI1. The analysis shows that the accuracy of multiple regression model is higher than that of single variable estimation model. When a variety of spectral parameters can be obtained, the multiple regression model is preferred to estimate the canopy LAI of peanut. |
| Key words: Peanut; leaf area index; hyperspectral remote sensing; model accuracy |
