基于高光譜遙感的水稻生長(zhǎng)監(jiān)測(cè)研究
發(fā)布時(shí)間：2018-08-01 11:31
【摘要】：快速、實(shí)時(shí)、準(zhǔn)確的獲取農(nóng)田生態(tài)環(huán)境和作物長(zhǎng)勢(shì)信息是精準(zhǔn)農(nóng)業(yè)實(shí)施的重要基礎(chǔ)前提,也是現(xiàn)代精準(zhǔn)農(nóng)業(yè)發(fā)展的關(guān)鍵技術(shù)瓶頸之一。本論文把水稻作為研究對(duì)象,以不同施肥(氮肥、生物質(zhì)炭肥)梯度下的小區(qū)試驗(yàn)作為依托,綜合運(yùn)用高光譜遙感、生理生化參數(shù)測(cè)試以及數(shù)理統(tǒng)計(jì)等技術(shù)手段,分析不同施肥(氮肥、生物質(zhì)炭肥)條件下水稻在不同生育期的冠層高光譜特征、葉綠素含量(SPAD)特征和葉面積指數(shù)(LAI)特征,分生育期的形式建立基于原始光譜反射率參數(shù)、“三邊”參數(shù)的水稻葉綠素含量、葉面積指數(shù)的高光譜估算模型,并利用決定系數(shù)(R2)、均方根誤差(RMSE)、相對(duì)誤差等評(píng)價(jià)指標(biāo)驗(yàn)證預(yù)測(cè)模型的精度。本研究主要取得了以下結(jié)果:(1)對(duì)水稻在不同生育期以及不同施肥(氮肥、生物質(zhì)炭肥)條件下的冠層光譜的變化規(guī)律進(jìn)行研究。結(jié)果表明:從拔節(jié)期到乳熟期,水稻冠層光譜反射率在可見(jiàn)光區(qū)域是不斷增大的,而在近紅外區(qū)域是先增大后減小的。隨著施氮水平的提高,水稻冠層光譜反射率在可見(jiàn)光范圍內(nèi)總體是呈現(xiàn)逐漸降低的趨勢(shì),而在近紅外波段是逐漸增高的。不同生物質(zhì)炭水平下,水稻在可見(jiàn)光波段的冠層光譜的反射率差別不明顯,而到達(dá)近紅外波段,冠層光譜反射率的差異較明顯并且施炭處理的水稻冠層光譜反射率的值大于不施炭處理的冠層光譜反射率的值。因此,該研究結(jié)果能夠?yàn)楦顚哟蔚乩盟竟趯庸庾V信息監(jiān)測(cè)水稻生長(zhǎng)狀況提供一定的理論基礎(chǔ)。(2)分析了水稻冠層原始光譜、導(dǎo)數(shù)光譜與葉綠素含量(SPAD)、葉面積指數(shù)(LAI)的相關(guān)關(guān)系。在可見(jiàn)光區(qū)域,水稻原始光譜反射率與葉綠素含量、葉面積指數(shù)在拔節(jié)期、抽穗期與灌漿期呈現(xiàn)出負(fù)相關(guān),在“紅邊”處,由負(fù)相關(guān)變成正相關(guān);水稻的導(dǎo)數(shù)光譜與葉綠素含量、葉面積指數(shù)之間的相關(guān)系數(shù)在拔節(jié)期、抽穗期與灌漿期的一些波段處高于原始光譜反射率與葉綠素含量、葉面積指數(shù)的相關(guān)系數(shù);乳熟期水稻的原始光譜、導(dǎo)數(shù)光譜與葉綠素含量、葉面積指數(shù)的相關(guān)系數(shù)較低,因此,在建立水稻葉綠素含量、葉面積指數(shù)的高光譜估算模型過(guò)程中,不適宜使用乳熟期的光譜數(shù)據(jù)。(3)應(yīng)用2014年的水稻冠層光譜與葉綠素含量(SPAD)、葉面積指數(shù)(LAI)數(shù)據(jù),在水稻拔節(jié)期、抽穗期與灌漿期基于原始光譜反射率參數(shù)(綠峰位置、綠峰反射率、綠峰面積、紅谷位置、紅谷反射率、紅谷面積、綠峰面積與紅谷面積的比值與歸一化值)、“三邊”參數(shù)(藍(lán)邊位置、藍(lán)邊振幅、藍(lán)邊面積、黃邊位置、黃邊振幅、黃邊面積、紅邊位置、紅邊振幅、紅邊面積、紅邊面積與藍(lán)邊面積的比值與歸一化值、紅邊面積與黃邊面積的比值與歸一化值)建立葉綠素含量、葉面積指數(shù)的估算模型,并應(yīng)用2013年水稻冠層光譜與葉綠素含量、葉面積指數(shù)的相關(guān)數(shù)據(jù)對(duì)所建的預(yù)測(cè)模型進(jìn)行精度檢驗(yàn)。結(jié)果表明:運(yùn)用原始光譜反射率參數(shù)反演水稻葉綠素含量時(shí),在拔節(jié)期應(yīng)優(yōu)先考慮紅谷面積,抽穗期、灌漿期的紅谷反射率、紅谷面積反演效果都很好。使用原始光譜反射率參數(shù)反演水稻葉面積指數(shù)時(shí),拔節(jié)期,優(yōu)先考慮綠峰面積與紅谷面積的歸一化值,抽穗期,紅谷反射率、紅谷面積的反演效果相對(duì)較好,灌漿期的綠峰反射率、紅谷面積反演效果相對(duì)較好。應(yīng)用“三邊”參數(shù)反演水稻葉綠素含量時(shí),拔節(jié)期優(yōu)先考慮應(yīng)用紅邊面積與藍(lán)邊面積的比值,抽穗期優(yōu)先考慮使用藍(lán)邊面積,灌漿期紅邊位置、紅邊面積與藍(lán)邊面積的歸一化值反演效果都較好。使用“三邊”參數(shù)反演水稻葉面積指數(shù)時(shí),拔節(jié)期,紅邊面積、紅邊面積與藍(lán)邊面積的歸一化值的反演效果都很好,抽穗期,優(yōu)先考慮紅邊面積,灌漿期,紅邊位置、紅邊面積與藍(lán)邊面積的比值的反演效果相對(duì)較好。
[Abstract]:Fast, real-time, accurate acquisition of farmland ecological environment and crop growth information is an important basic prerequisite for the implementation of precision agriculture. It is also one of the key technology bottlenecks in the development of modern precision agriculture. This paper takes rice as the research object and relies on the plot experiment under the gradient of nitrogen fertilizer (nitrogen fertilizer, biomass charcoal fertilizer). The spectral remote sensing, physiological and biochemical parameters and mathematical statistics were used to analyze the hyper spectral characteristics, chlorophyll content (SPAD) characteristics and leaf area index (LAI) characteristics of rice at different growth stages under different fertilization (nitrogen fertilizer, biomass carbon fertilizer), based on the original spectral reflectance parameters, "three sides". The chlorophyll content of rice, the Hyperspectral Estimation Model of leaf area index, and the evaluation index of R2, RMSE and relative error were used to verify the accuracy of the prediction model. The main results are as follows: (1) under the condition of different growth period and different fertilization (nitrogen fertilizer, biomass carbon fertilizer) The spectral reflectance of the canopy was studied. The results showed that the spectral reflectance of the rice canopy increased from the jointing stage to the milk ripening period, but increased first and then decreased in the near infrared region. With the increase of nitrogen application level, the spectral reflectance of rice canopy was gradually decreasing in the visible light range. In the near infrared band, the reflectance of the canopy spectral reflectance in the visible light band is not obvious, but the reflectance of the canopy spectral reflectance is obvious in the near infrared band, and the value of the canopy spectral reflectance of the rice is greater than that of the canopy spectral reflectance without carbon treatment. Therefore, the results can provide a theoretical basis for monitoring rice growth by using the spectral information of rice canopy more deeply. (2) the correlation between the original spectrum of the rice canopy, the derivative spectrum and the chlorophyll content (SPAD), the leaf area index (LAI), and the original spectral reflectance of rice in the visible region, and the reflectance of the rice were analyzed. The content of chlorophyll, leaf area index at the jointing stage, the heading stage and the filling stage showed a negative correlation, in the "red edge", from negative correlation into positive correlation, the correlation coefficient between the derivative spectrum of the rice and the chlorophyll content, the leaf area index was at the jointing stage, and some bands at the heading and filling stages were higher than those of the original spectral reflectance and chlorophyll. The correlation coefficient of the content, leaf area index, the original spectrum of rice, the derivative spectrum and the chlorophyll content and the leaf area index are lower. Therefore, in the process of establishing the Hyperspectral Estimation Model of the chlorophyll content and leaf area index of rice, the spectral data of the milk ripening period are not suitable. (3) the application of the rice canopy light in 2014. The spectrum and chlorophyll content (SPAD) and leaf area index (LAI) data are based on the original spectral reflectance parameters (green peak position, green peak reflectivity, green peak area, Red Valley location, Red Valley area, Red Valley area, green peak area and Red Valley area ratio and normalization value), and "three edge" parameter (blue edge position). The amplitude of blue edge, blue edge area, yellow edge position, yellow edge amplitude, yellow edge area, red edge position, red edge amplitude, red edge area, red edge area and blue edge area ratio and normalized value, ratio and normalization value of red edge area and yellow edge area and normalization value) set up chlorophyll content, estimation model of leaf area index, and applied the spectrum of rice canopy in 2013. The results showed that the Red Valley area, the heading stage, the Red Valley reflectivity at the grain filling period and the Red Valley area inversion effect were good. When the rice leaf area index was retrieved by the ejection parameter, the jointing period was given priority to the normalized value of green peak area and Red Valley area. The inversion effect of the heading stage, the Red Valley reflectivity and the Red Valley area was relatively better. The green peak reflectivity and the Red Valley area inversion effect was relatively better. When the chlorophyll content was retrieved with the "three edge" parameters, the extraction was extracted. In the festival, the ratio of red edge area to blue edge area was first considered, and the area of blue edge was first considered in heading stage, and the inversion effect of red edge area and blue edge area was better. When the rice leaf area index was retrieved by "three edge" parameters, the extraction period, red edge area, red edge area and blue edge area were returned. The inversion effect of one value is very good, and the back effect of the ratio of red edge area, red edge location, red edge area and blue edge area is better than that of red edge area.
【學(xué)位授予單位】：西北農(nóng)林科技大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2015
【分類號(hào)】：S127;S511
本文編號(hào)：2157437
[Abstract]:Fast, real-time, accurate acquisition of farmland ecological environment and crop growth information is an important basic prerequisite for the implementation of precision agriculture. It is also one of the key technology bottlenecks in the development of modern precision agriculture. This paper takes rice as the research object and relies on the plot experiment under the gradient of nitrogen fertilizer (nitrogen fertilizer, biomass charcoal fertilizer). The spectral remote sensing, physiological and biochemical parameters and mathematical statistics were used to analyze the hyper spectral characteristics, chlorophyll content (SPAD) characteristics and leaf area index (LAI) characteristics of rice at different growth stages under different fertilization (nitrogen fertilizer, biomass carbon fertilizer), based on the original spectral reflectance parameters, "three sides". The chlorophyll content of rice, the Hyperspectral Estimation Model of leaf area index, and the evaluation index of R2, RMSE and relative error were used to verify the accuracy of the prediction model. The main results are as follows: (1) under the condition of different growth period and different fertilization (nitrogen fertilizer, biomass carbon fertilizer) The spectral reflectance of the canopy was studied. The results showed that the spectral reflectance of the rice canopy increased from the jointing stage to the milk ripening period, but increased first and then decreased in the near infrared region. With the increase of nitrogen application level, the spectral reflectance of rice canopy was gradually decreasing in the visible light range. In the near infrared band, the reflectance of the canopy spectral reflectance in the visible light band is not obvious, but the reflectance of the canopy spectral reflectance is obvious in the near infrared band, and the value of the canopy spectral reflectance of the rice is greater than that of the canopy spectral reflectance without carbon treatment. Therefore, the results can provide a theoretical basis for monitoring rice growth by using the spectral information of rice canopy more deeply. (2) the correlation between the original spectrum of the rice canopy, the derivative spectrum and the chlorophyll content (SPAD), the leaf area index (LAI), and the original spectral reflectance of rice in the visible region, and the reflectance of the rice were analyzed. The content of chlorophyll, leaf area index at the jointing stage, the heading stage and the filling stage showed a negative correlation, in the "red edge", from negative correlation into positive correlation, the correlation coefficient between the derivative spectrum of the rice and the chlorophyll content, the leaf area index was at the jointing stage, and some bands at the heading and filling stages were higher than those of the original spectral reflectance and chlorophyll. The correlation coefficient of the content, leaf area index, the original spectrum of rice, the derivative spectrum and the chlorophyll content and the leaf area index are lower. Therefore, in the process of establishing the Hyperspectral Estimation Model of the chlorophyll content and leaf area index of rice, the spectral data of the milk ripening period are not suitable. (3) the application of the rice canopy light in 2014. The spectrum and chlorophyll content (SPAD) and leaf area index (LAI) data are based on the original spectral reflectance parameters (green peak position, green peak reflectivity, green peak area, Red Valley location, Red Valley area, Red Valley area, green peak area and Red Valley area ratio and normalization value), and "three edge" parameter (blue edge position). The amplitude of blue edge, blue edge area, yellow edge position, yellow edge amplitude, yellow edge area, red edge position, red edge amplitude, red edge area, red edge area and blue edge area ratio and normalized value, ratio and normalization value of red edge area and yellow edge area and normalization value) set up chlorophyll content, estimation model of leaf area index, and applied the spectrum of rice canopy in 2013. The results showed that the Red Valley area, the heading stage, the Red Valley reflectivity at the grain filling period and the Red Valley area inversion effect were good. When the rice leaf area index was retrieved by the ejection parameter, the jointing period was given priority to the normalized value of green peak area and Red Valley area. The inversion effect of the heading stage, the Red Valley reflectivity and the Red Valley area was relatively better. The green peak reflectivity and the Red Valley area inversion effect was relatively better. When the chlorophyll content was retrieved with the "three edge" parameters, the extraction was extracted. In the festival, the ratio of red edge area to blue edge area was first considered, and the area of blue edge was first considered in heading stage, and the inversion effect of red edge area and blue edge area was better. When the rice leaf area index was retrieved by "three edge" parameters, the extraction period, red edge area, red edge area and blue edge area were returned. The inversion effect of one value is very good, and the back effect of the ratio of red edge area, red edge location, red edge area and blue edge area is better than that of red edge area.
【學(xué)位授予單位】：西北農(nóng)林科技大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2015
【分類號(hào)】：S127;S511
【共引文獻(xiàn)】
相關(guān)期刊論文 前4條
1 李桂娟;朱麗麗;李井會(huì);;作物氮素營(yíng)養(yǎng)診斷的無(wú)損測(cè)試研究與應(yīng)用現(xiàn)狀[J];黑龍江農(nóng)業(yè)科學(xué);2008年04期
2 唐延林,王秀珍,王珂;利用光譜法測(cè)定水稻生物物理參數(shù)及其與光譜變量的相關(guān)性研究[J];貴州大學(xué)學(xué)報(bào)(農(nóng)業(yè)與生物科學(xué)版);2002年05期
3 李曉利;唐延林;樓佳;;雜交水稻葉片葉綠素含量與光譜吸收率的關(guān)系[J];山地農(nóng)業(yè)生物學(xué)報(bào);2011年01期
4 唐延林,王人潮,王秀珍;對(duì)水稻微分光譜和植被指數(shù)的探討[J];上海交通大學(xué)學(xué)報(bào)(農(nóng)業(yè)科學(xué)版);2003年03期
相關(guān)碩士學(xué)位論文 前4條
1 張曉倩;基于全極化SAR數(shù)據(jù)的水稻識(shí)別與作物參數(shù)估算模型[D];中國(guó)地質(zhì)大學(xué)(北京);2012年
2 李燕強(qiáng);不同冬小麥品種冠層光譜特征及其與農(nóng)學(xué)參數(shù)關(guān)系研究[D];河南農(nóng)業(yè)大學(xué);2012年
3 閻國(guó)倩;植被與土壤混合像元的高光譜偏振特性研究[D];東北師范大學(xué);2013年
4 武婕;玉米成熟期地上生物量及其碳氮累積量的遙感估算[D];山東農(nóng)業(yè)大學(xué);2014年
，本文編號(hào)：2157437
本文鏈接：http://sikaile.net/kejilunwen/nykj/2157437.html
最近更新
教材專著
