基于玉米冠层结构特点的遥感监测模型研究

发布时间：2018-05-13 00:18

本文选题：春玉米 + 密度　；参考：《石河子大学》2017年硕士论文

【摘要】：【目的】通过分析新疆地区不同密度下春玉米冠层结构特征,冠层叶面积指数、产量与冠层高光谱相关关系,建立不同密度下春玉米冠层叶面积指数和产量的高光谱估算模型,为新疆地区春玉米长势和产量估算提供依据。【方法】在新疆不同春玉米品种、不同密度水平条件下开展大田试验,测定了冠层光谱反射率、叶面积指数、分层叶面积、生物产量、株高、穗位、经济产量等数据,分析在不同密度下累积叶面积指数、相对叶面积密度、生物产量、经济产量与高光谱的相关性,建立了叶面积指数、相对叶面积密度、生物产量、经济产量的相关模型并进行模型精度的检验。【结果】通过开展试验研究,得到如下结果:(1)随着密度的增大,玉米最长和最宽叶片叶位保持大致不变,叶宽随着密度的增大而显著减小,叶长随着密度的增大先增大后减少;单株叶面积减小,基部叶片(1-6叶)叶面积变化不大,下部叶片(7-12叶)叶面积先增大后减小,中部叶片(13-16叶)不同年际间表现出不同差异,上部叶片(17叶及以上)叶面积逐渐减小;群体叶面积增加,中部及上部相对叶面积差异较大;茎干重差异较大,叶干重差异较小。在D3密度下,经济产量与穗位高、相对叶面积密度呈正相关,与生物产量呈现负相关,显著性均少于0.05。(2)不同密度下,春玉米叶面积指数和相对叶面积密度的估算模型不同。叶面积指数的高光谱估算模型在D1,D2,D3密度下分别以RVI[497,935],DVI[720,936],DVI[551,724]为参数拟合的估算模型y=-0.0014x2+0.1201x+2.1747(R2=0.65),y=-30.405x2+10.122x+6.2617(R2=0.49),y=965.98x2-285.68x+29.929(R2=0.65)最好,RMSE分别为0.73、0.34、0.10。相对叶面积密度的高光谱估算模型在D1,D2,D3密度下分别以RVI[1143,947],R945,RDVI[712,552]拟合的估算模型y=-19.588x2+31.649(R2=0.61),y=-26.266x2+20.746x+10.726(R2=0.42),y=-2207.436x2+538.426x-17.601(R2=0.96)精度最高,RMSE分别为0.82,2.45,0.41。(3)不同密度下,春玉米产量的估算模型不同。生物产量的高光谱估算模型在D1,D2,D3密度下分别以NDVI[719,1080],R615,DVI[1020,671]拟合的估算模型y=75.205e2.668x(R2=0.62),y=-13894.287x2+1651.835x+110.938(R2=0.22),y=120.438x2-44.535x+95.499(R2=0.34)精度最高,RMSE分别为1.99,3.61,2.37。经济产量的高光谱估算模型在D1,D2,D3密度下分别以DVI[691,401],DVI[1102,533],NDVI[1122,780]拟合的估算模型y=1004.37e-0.716x(R2=0.58),y=21255.197x2-22028.232x+6757.953(R2=0.54),y=1122.356+19.933lnx(R2=0.39)精度最高,RMSE分别为1.24,1.13,2.37。【结论】利用高光谱遥感可以对不同密度下春玉米冠层结构参数及产量估算。
[Abstract]:[objective] to establish a hyperspectral estimation model of canopy leaf area index and yield of spring maize under different densities by analyzing the correlation between canopy structure, canopy leaf area index, yield and canopy hyperspectral spectrum of spring maize under different densities in Xinjiang. [methods] the field experiments were carried out in different spring maize varieties and different density levels in Xinjiang. The spectral reflectance of canopy, leaf area index and stratified leaf area were measured. Based on the data of biological yield, plant height, ear position and economic yield, the correlation of cumulative leaf area index, relative leaf area density, biological yield and economic yield with hyperspectral data was analyzed, and the leaf area index was established. Relative leaf area density, biological yield, economic yield, and model accuracy were tested. [results] through the experimental study, the following results were obtained: 1) with the increase of density, The leaf position of the longest and widest leaves of maize remained approximately unchanged, the leaf width decreased significantly with the increase of density, the leaf length increased first and then decreased with the increase of density, and the leaf area of single plant decreased, but the leaf area of basal leaf increased slightly. The leaf area of the lower leaves increased first and then decreased, the middle leaves showed different differences among different years, the upper leaves of 17 leaves and more) leaf area gradually decreased, and the population leaf area increased. The difference of relative leaf area between middle and upper part was great, the difference of stem dry weight and leaf dry weight was large, and the difference of leaf dry weight was small. Under D _ 3 density, the economic yield was positively correlated with ear height and relative leaf area density, and negatively correlated with biological yield (< 0.05. 2) under different densities, the estimation models of leaf area index and relative leaf area density of spring maize were different. The hyperspectral estimation model of leaf area index was fitted by RVI [497935] DVI [720936] DVI [551724] at D _ (1) C _ (2) D _ (2) D _ (3) densities, respectively. The estimated model y=-0.0014x2 0.1201x 2.1747R _ (2) O _ (2) 0.65 ~ (0.405x) 10.122x 6.26405x2 10.122x 6.2617R20.495.98x2-285.68x 29.929R2O _ (0.65) was the best one. The hyperspectral estimation model of relative leaf area density was fitted with RVI [1143947] R945N RDVI [712552] at D _ (1) C _ (2) D _ (2) density, respectively. The estimation model y=-19.588x2 31.649 ~ (9) ~ (2) ~ (0.61) ~ (1) ~ 0.61 ~ (1) ~ (1) -26.266x ~ (2) 20.746x 10.726x 10.726x ~ (2 +) R20.42y-2207.436x2 538.426x-17.601C ~ (0.96) the estimation models of spring maize yield were different at different densities. The hyperspectral estimation model of biological yield was fitted with NDVI [719C1080] R615DVI [1020671] under the density of D1C D2D3, respectively. The estimation model yt75.205e2.668xR2ON0. 62OUYYU -13894.287x2 165535x 110.93835x 110.938R2n 0.34) had the highest precision of RMSE of 1.99105e2.61kW 2.37. The hyperspectral estimation model of economic yield was fitted with DVI [691401] DVI [1102533] and NDVI [1122780] respectively under the density of D _ (1) O _ (2) D _ (2) C _ (3). [conclusion] the structural parameters and yield of spring maize canopy under different densities can be estimated by using hyperspectral remote sensing, respectively, at 212555.197x2-22028.232x 6757.953 R20.54 (R2122.356 19.933lnx / R20.39). [conclusion] using hyperspectral remote sensing, we can estimate the structure parameters and yield of spring maize canopy in different density.
【学位授予单位】：石河子大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：S513;S127

【参考文献】