發(fā)布時間：2018-01-11 03:06

  本文關(guān)鍵詞：基于多源遙感數(shù)據(jù)的油菜長勢監(jiān)測　出處：《華中農(nóng)業(yè)大學(xué)》2016年碩士論文　論文類型：學(xué)位論文

  更多相關(guān)文章： 油菜 高光譜遙感 葉面積指數(shù) 逐步回歸 產(chǎn)量估計 長勢監(jiān)測

【摘要】：油菜是我國廣泛種植的油料作物,實(shí)時、精準(zhǔn)、快速估測油菜長勢對于油菜生長診斷與管理以及產(chǎn)量預(yù)測具有重要意義。隨著遙感技術(shù)不斷發(fā)展,遙感數(shù)據(jù)源越來越豐富,使得大范圍實(shí)時監(jiān)測農(nóng)作物長勢成為可能,大量改進(jìn)型光譜特征參數(shù)、植被指數(shù)反演農(nóng)學(xué)參數(shù)為精準(zhǔn)農(nóng)業(yè)提供技術(shù)支持,尤其是高光譜遙感憑借波段連續(xù)性強(qiáng)、光譜信息量大的優(yōu)勢,為實(shí)時快速大面積監(jiān)測作物長勢提供有效信息。充分利用多源遙感數(shù)據(jù)監(jiān)測油菜關(guān)鍵生長期的長勢,建立較為精確的油菜生長遙感監(jiān)測模型并提取油菜種植面積,成為大范圍監(jiān)測油菜長勢監(jiān)測的趨勢。本研究優(yōu)化了不同條件下(栽種方式、施肥水平等)油菜葉面積和葉鮮重的估算模型,采用高光譜技術(shù)反演了試驗(yàn)小區(qū)不同時期葉面積指數(shù),并利用試驗(yàn)小區(qū)不同時期葉面積指數(shù)建立了試驗(yàn)小區(qū)油菜估產(chǎn)模型,探索通過提取遙感參數(shù)反演油菜葉面積指數(shù),然后利用葉面積指數(shù)進(jìn)行大范圍估算油菜單產(chǎn)。上述研究得出結(jié)論如下:1.采用麥夸特法+通用全局優(yōu)化算法對長寬系數(shù)模型的參數(shù)精確計算,建立的受外界干擾較少(栽種方式、施肥水平、生長時期)長寬冪函數(shù)模型,其建模效果和預(yù)測精度比常規(guī)長寬線性模型更高,可準(zhǔn)確估計油菜葉片葉面積和葉鮮重。2.不同時期所建反演LAI模型以二次多項(xiàng)式為主,但最優(yōu)模型差異較大,在苗期時以紅邊參數(shù)為主的建模精度和預(yù)測效果較好,而油菜生長后期(花期、角果期)則是以NLI為代表的非線性指數(shù)建模精度和預(yù)測效果相對較高,全生育所建模型精度較低,難以采用相對固定的光譜參數(shù)和指數(shù)來預(yù)測油菜整個生育期LAI,不同生育期的油菜株型和覆蓋度存在較大的差異,花和角果與葉不同的光譜響應(yīng)特征,采用統(tǒng)一建模預(yù)測LAI不能準(zhǔn)確預(yù)測LAI,不同生長時期需要選擇合適的光譜參數(shù)和植被指數(shù)建立預(yù)測模型。3.將不同時期試驗(yàn)小區(qū)葉面積指數(shù)作為自變量,試驗(yàn)小區(qū)最終產(chǎn)量作為因變量,相關(guān)分析表明,產(chǎn)量與各個時期LAI皆呈顯著的正相關(guān)關(guān)系,與角果期葉面積指數(shù)呈現(xiàn)極顯著正相關(guān),通過逐步回歸分析,利用十葉期、盛花期、角果期的葉面積指數(shù)所建模型精度較高,經(jīng)檢驗(yàn)具有較好地預(yù)測效果。4.利用遙感影像提取的9個常用植被指數(shù)逐步回歸分別反演武穴市十葉期、盛花期、角果期的葉面積指數(shù),利用試驗(yàn)小區(qū)葉面積指數(shù)—產(chǎn)量模型對武穴市油菜單產(chǎn)進(jìn)行估計,結(jié)果表明利用遙感影像提取的參數(shù)可以有效估算三個時期油菜葉面積指數(shù)和油菜單產(chǎn)。
[Abstract]:Rapeseed is a widely grown oil crop in China. The real-time accurate and rapid estimation of rapeseed growth is of great significance for rape growth diagnosis management and yield prediction. With the development of remote sensing technology. Remote sensing data sources are more and more abundant, which makes it possible to monitor crop growth on a large scale in real time. A large number of improved spectral characteristic parameters and agronomic parameters of vegetation index inversion provide technical support for precision agriculture. In particular, hyperspectral remote sensing has the advantages of strong band continuity and large spectral information. It can provide effective information for real-time and rapid monitoring of crop growth and make full use of multi-source remote sensing data to monitor the growth of rapeseed in key growing period. The establishment of a more accurate remote sensing monitoring model of rape growth and the extraction of rape planting area have become the trend of monitoring rapeseed growth on a large scale. The leaf area and fresh weight of rape were estimated by using hyperspectral technique. The yield estimation model of rape was established by using the leaf area index of different periods in the experimental plot, and the retrieval of rape leaf area index by extracting remote sensing parameters was explored. Then using leaf area index to estimate rapeseed yield on a large scale. The conclusions are as follows: 1. The parameters of the length and width coefficient model are calculated accurately by using the general global optimization algorithm of McQuat method. The power function model with less external interference (planting mode, fertilization level, growth period) was established, and its modeling effect and prediction accuracy were higher than that of the conventional length and width linear model. The leaf area and fresh weight of rape leaves can be estimated accurately. The inverse LAI model built in different periods is mainly quadratic polynomial, but the difference of the optimal model is great. In seedling stage, the modeling accuracy and prediction effect of red edge parameter were better, while the late growth stage (flowering period, pod stage) of rapeseed was the nonlinear exponential modeling accuracy and prediction effect represented by NLI. The precision of the whole growth model is low, it is difficult to use the relative fixed spectral parameters and indices to predict the whole growth period of rape Lai. There are great differences in plant type and coverage of rape at different growth stages. The spectral response characteristics of flower, pod and leaf were different, and LAI could not be predicted accurately by using unified modeling to predict LAI. Different growth stages need to select the appropriate spectral parameters and vegetation index to establish a prediction model .3. take the leaf area index of different periods as the independent variable and the final yield of the experimental plot as the dependent variable. Correlation analysis showed that there was a significant positive correlation between yield and LAI in each stage, and a very significant positive correlation between yield and leaf area index in pod stage. Through stepwise regression analysis, the ten-leaf stage and flowering stage were used. The accuracy of the model established by the leaf area index of the pod period is high, and it has good prediction effect by testing. 4. Nine common vegetation indices extracted from remote sensing image were used to invert the ten leaf period and the full flowering stage of Wu-acupoint City respectively by stepwise regression. The leaf area index of the pod stage was estimated by using the leaf area index-yield model of the experimental plot to estimate the yield per unit of rape in Wucao city. The results showed that the parameters extracted from remote sensing images could effectively estimate rape leaf area index and rape yield in three periods.
【學(xué)位授予單位】：華中農(nóng)業(yè)大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2016
【分類號】：S565.4;S127

【參考文獻(xiàn)】

相關(guān)期刊論文 前5條

1 謝巧云;黃文江;蔡淑紅;梁棟;彭代亮;張清;黃林生;楊貴軍;張東彥;;冬小麥葉面積指數(shù)遙感反演方法比較研究[J];光譜學(xué)與光譜分析;2014年05期

2 王敏;曲存民;劉曉蘭;謝景梅;李加納;;溫度脅迫下甘藍(lán)型油菜苗期生理生化指標(biāo)的研究[J];作物雜志;2013年02期

3 吳炳方,張峰,劉成林,張磊,羅治敏;農(nóng)作物長勢綜合遙感監(jiān)測方法[J];遙感學(xué)報;2004年06期

4 ;Relations Between Red Edge Characteristics and Agronomic Parameters of Crops[J];Pedosphere;2004年04期

5 林忠輝,莫興國,項(xiàng)月琴;作物生長模型研究綜述[J];作物學(xué)報;2003年05期

，

本文編號：1407847