本文選題：高光譜　切入點(diǎn)：春玉米　出處：《內(nèi)蒙古農(nóng)業(yè)大學(xué)》2016年博士論文　論文類型：學(xué)位論文

【摘要】：高光譜技術(shù)即為精確農(nóng)業(yè)的一種重要技術(shù)手段,因其具有方便、快捷、高效、對(duì)植株無(wú)損害等優(yōu)點(diǎn),己被大量應(yīng)用于作物生長(zhǎng)監(jiān)測(cè)、作物植株水分監(jiān)測(cè)、營(yíng)養(yǎng)狀況監(jiān)測(cè)、作物產(chǎn)量評(píng)估、品質(zhì)監(jiān)測(cè)等多個(gè)方面。本研究以玉米為研究主體,通過(guò)分析不同種植條件下(不同品種,不同密度,不同施氮量,氮、密互作)玉米冠層及葉片高光譜特征與其相對(duì)應(yīng)的LAI、SPAD值、葉片氮含量和產(chǎn)量的響應(yīng)規(guī)律,明確4個(gè)理化指標(biāo)的敏感波段,并利用光譜指數(shù)NDVI、RVI、DVI構(gòu)建了基于高光譜植被指數(shù)的LAI、SPAD值、葉片氮含量和產(chǎn)量高光譜估測(cè)模型。其主要研究結(jié)果如下：(1)栽培環(huán)境的改變,會(huì)直接引起玉米生理生態(tài)參數(shù)變化,而這種變化又會(huì)因其栽培措施的不同而產(chǎn)生差異,例如：生育期(葉片衰老)、種植密度、施氮量等因素均會(huì)導(dǎo)致玉米LAI發(fā)生改變,但種植密度對(duì)LAI的影響最大,葉片衰老變化次之,施氮量最小。同理,不同栽培條件也會(huì)影響冠層和葉片高光譜特征,在可見(jiàn)光350-760nm波段,冠層光譜與葉片光譜反射率隨著生育期進(jìn)程呈增大趨勢(shì)；不同種植密度下,冠層光譜反射率表現(xiàn)為隨密度的增大而減小,而葉片光譜反射率則表現(xiàn)為隨密度增加呈增大趨勢(shì)；不同施氮量下,冠層光譜與葉片光譜反射率隨著施氮量的增加呈下降趨勢(shì)。在780-1300rnm波段,冠層光譜與葉片光譜反射率隨生育進(jìn)程呈逐漸下降趨勢(shì)；不同種植密度下,冠層光譜反射率隨密度的增大呈增大趨勢(shì),而葉片光譜反射率則無(wú)明顯變化規(guī)律；不同施氮量下,冠層光譜反射率隨施氮量的增加呈增大趨勢(shì),而葉片光譜反射率在這一波段則無(wú)明顯變化規(guī)律。種植密度對(duì)冠層光譜反射率的影響大于施氮量,施氮量對(duì)葉片光譜的影響要大于密度。(2)通過(guò)對(duì)不同栽培條件下的玉米冠層、葉片光譜與LAI、葉片SPAD值、LNC和產(chǎn)量的相關(guān)分析,得出4個(gè)指標(biāo)的反射率光譜敏感波長(zhǎng)主要位于550nm、678nm、710nm和1100nm附近,一階導(dǎo)數(shù)光譜敏感波長(zhǎng)位于500nm、550nm、580-680nm之間、700nm和755nm附近。利用NDVI、RVI和DVI三種光譜參數(shù)構(gòu)建了不同栽培條件下的LAI、SPAD值、LNC和產(chǎn)量高光譜估測(cè)模型,并對(duì)模型應(yīng)用精度進(jìn)行了比較,得出：模型在應(yīng)用于其他栽培條件時(shí)均會(huì)出現(xiàn)較大偏差,其中冠層光譜模型對(duì)其他條件下各指標(biāo)的估測(cè)精度都比較差,尤其是對(duì)LAI的估測(cè)偏差最大。而SPAD值和LNC的葉片光譜模型,具有較高的普適性。(3)不同栽培條件下高光譜參數(shù)值與各指標(biāo)數(shù)值之間定量關(guān)系的差異是導(dǎo)致各指標(biāo)高光譜估測(cè)模型普適性差的根本原因,對(duì)不同栽培條件下的光譜與各指標(biāo)數(shù)據(jù)進(jìn)行綜合分析,可以降低兩者之間的不匹配程度,提高模型的普適性。同時(shí),利用土壤線參數(shù)和葉片光譜與冠層光譜反射率差值參數(shù)可以提高光譜模型的穩(wěn)定性。在所建高光譜模型中,通用性較好的各指標(biāo)估測(cè)模型有：LAI為NDVI(729.3,963.6)和MNDVI(729.3,963.3)模型；SPAD值和LNC為NDVI(729.3,963.6)和mRVI(729.3,963.6)模型；產(chǎn)量為NDVI(R'695.7, R'755.5)和MRVI(550.2,963.6)模型。
[Abstract]:Hyperspectral technology is an important means of precision agriculture, because of its convenient, fast, efficient, has the advantages of no damage to the plant, has been widely used in crop growth monitoring, crop water monitoring, status monitoring, crop yield assessment, many aspects of quality monitoring. In this study, corn the research subject, through the analysis of different planting conditions (different varieties, different densities, different nitrogen, nitrogen, close interaction) characteristics of corn canopy and leaf hyperspectral and the corresponding LAI, SPAD value, nitrogen content and leaf yield response rules, clear 4 physicochemical indexes and the sensitive bands. Using the spectral index NDVI, RVI, DVI constructed Hyperspectral Vegetation Index Based on LAI, SPAD value, nitrogen content and yield of leaf Hyperspectral Estimation Models. The main results are as follows: (1) the cultivation of the environment changes, may be a direct cause of maize physiological and ecological parameters But, this change will have the difference, because of the different cultivation measures such as growth period (leaf senescence), planting density, nitrogen and other factors will lead to the change of maize LAI, but the effects of planting density on LAI maximum, leaf senescence of nitrogen is minimal. Similarly, different cultivation conditions will also affect the spectral characteristics of canopy and leaf, in the visible band 350-760nm, canopy spectral reflectance and leaf growth process with increased; under different planting density, canopy spectral reflectance was observed with the increase of density decreases, while the leaf spectral reflectance is increased with the increase of density; under different nitrogen levels, canopy spectral reflectance and leaf spectral reflectance with the increase of nitrogen decreased. In 780-1300rnm band, canopy spectral reflectance and leaf spectral reflectance decreased gradually with the growth and development trend Potential; under different planting density, canopy spectral reflectance with increasing density increased, while the leaf spectral reflectance had no significant variation; under different nitrogen levels, the canopy spectral reflectance with the increase of nitrogen increased, while the leaf spectral reflectance in the wave period had no significant variation effect of planting density. The canopy reflectance is greater than the amount of nitrogen, the nitrogen effect on leaf spectra than density. (2) through the canopy of Maize under different cultivation conditions, leaf spectra and leaf SPAD values, LAI, LNC correlation analysis and yield, reflectance spectrum sensitive wavelength draw 4 index are mainly located at 550nm. 678nm, near 710nm and 1100nm, between the first derivative spectra are sensitive to wavelengths of 500nm, 550nm, 580-680nm, 700nm and 755nm. Near by NDVI, RVI and DVI three spectral parameters to build different cultivation conditions of LAI, SPAD The value of LNC, and the yield of Hyperspectral Estimation Models, and the application of the precision of the model were compared to that in the model applied to other cultivation conditions will appear larger deviation, the estimation precision of canopy spectral model for each index under other conditions are relatively poor, especially on the estimation of maximum deviation of LAI and SPAD value. LNC and leaf spectral model, general higher. (3) the difference of spectral parameters under different cultivation conditions and the quantitative relationship between the index value of each index is the result of numerical Hyperspectral Estimation Model of root causes of poor universality, data spectrum of different cultivation conditions and various indicators analysis, can reduce the degree of mismatch between the two, to improve the universality of the model. At the same time, the use of soil line parameters and leaf reflectance and canopy reflectance difference parameters can improve the stability of the model. In the spectrum In the hyperspectral model, the index estimation models with good universality are: LAI is NDVI (729.3963.6) and MNDVI (729.3963.3) model; SPAD value and LNC are NDVI (729.3963.6) and mRVI (729.3963.6) model; output is the model of "729.3963.6" and "X".

【學(xué)位授予單位】：內(nèi)蒙古農(nóng)業(yè)大學(xué)
【學(xué)位級(jí)別】：博士
【學(xué)位授予年份】：2016
【分類號(hào)】：S513
，

本文編號(hào)：1578346