不同物候期的苹果树冠层氮素含量高光谱估测研究

发布时间：2019-04-13 17:15

【摘要】：氮素是植物生长发育所必需的大量元素之一,对于苹果树生长及果实品质的形成起着至关重要的作用。目前,利用高光谱对植被氮素含量的估测研究主要集中在大田作物,而对多年生的苹果树研究较少,且已有研究大都集中在果树的某一个生长时期,而对果树不同物候期的研究较少。因此对苹果树不同物候期的氮素含量进行高光谱估测研究,对快速、无损和准确地估测果树氮素含量,从而确定氮素含量的最佳估测时期,对果树的精准施肥具有重要意义。以山东省栖霞市为研究区,以红富士苹果树为研究对象。在2013年及2014年苹果盛花期(4月下旬)、新梢旺长期(5月中下旬)、春梢停长期(6月中下旬)、秋梢停长期(9月中下旬)四个物候期,利用ASD FieldSpec 4便携式地物波谱仪,对盛果期苹果树冠层光谱数据进行测定,同步采集苹果树冠的叶片,在室内对叶片N素含量进行测定。分析了不同物候期苹果树冠层N素含量变化规律及不同N含量苹果树冠层光谱特征;对原始光谱数据进行了预处理和15种变换,筛选了与氮素含量相关的敏感波长。同时选用了基于位置的和基于面积的高光谱参量以及其它三类植被指数。最后建立了不同物候期苹果树冠层氮素含量最佳估测模型。主要研究结果如下:(1)初步摸清了苹果不同物候期的冠层氮素含量和高光谱特征的变化规律从盛花期到秋梢停长期,苹果冠层氮素平均含量一直处于下降状态,其中从新梢旺长期到春梢停长期氮素含量降低幅度相对较小,相对比较稳定。不同物候期苹果冠层光谱反射率具有相似的变化趋势。在350~780 nm波段范围内,盛花期、新梢旺长期、春梢停长期以及秋梢停长期的反射率相差较小,但盛花期的光谱反射率要高于其他三个时期的光谱反射率,且在680 nm(红谷)附近差值达到最大。在1000~1300 nm形成高的近红外反射平台,盛花期的光谱反射率要明显低于其他三个时期的光谱反射率,而新梢旺长期、春梢停长期和秋梢停长期的反射率相差最大,表现为新梢旺长期光谱反射率春梢停长期光谱反射率秋梢停长期光谱反射率。在1419~2402 nm波段范围内,四个时期的光谱反射率出现交叉。(2)明确了苹果不同物候期冠层氮素含量的敏感波段及敏感高光谱参量绿光(500~560 nm)和近红外区域(780~1100 nm)受冠层氮素含量影响较为显著,是盛花期苹果冠层氮素含量敏感波段,SDg,CIGREEN,GNDVI为盛花期苹果冠层氮素含量敏感高光谱参量。新梢旺长期冠层氮素含量敏感波段主要集中在近红外区域(780~1100 nm),冠层氮素含量敏感高光谱参量为DVI、MSAVI和TVI。春梢停长期苹果冠层氮素含量敏感波长位于红光区域(620~760 nm)和近红外区域(780~1100 nm),λr,MCARI,TCARI为春梢停长期苹果冠层氮素含量估测敏感高光谱参量。近红外区域(780~1100 nm)为秋梢停长期冠层氮素含量敏感区域,DVI、MSAVI、TVI和MCARI这四种植被指数为秋梢停长期苹果冠层氮素含量估测敏感高光谱参量。(3)建立了不同物候期苹果冠层氮素含量的最佳估测模型通过对比不同物候期苹果冠层氮素含量的估测模型的建模和估测精度发现,最佳估测模型都是基于多元逐步回归提取的敏感波长所构建的BP神经网络模型。苹果冠层氮素含量估测精度呈现同样的变化规律,春梢停长期模型精度新梢旺长期模型精度盛花期模型精度。秋梢停长期模型的估测精度较低,模型不稳定。因此春梢停长期可作为苹果冠层氮素含量估测的最佳时期。
[Abstract]:Nitrogen is one of the essential elements of plant growth and development, and plays an important role in the growth of apple trees and the formation of fruit quality. At present, the research on the nitrogen content of the vegetation by using the high-spectrum is mainly concentrated in the field crop, and the research on the perennial apple trees is less, and the research has mostly concentrated on a certain growth period of the fruit tree, and the research on the different phenological period of the fruit tree is less. Therefore, the high-spectral estimation of the nitrogen content in different phenological period of the apple tree is carried out, and the nitrogen content of the fruit tree can be estimated quickly, non-destructively and accurately, so that the optimal estimation period of the nitrogen content is determined, and the precise fertilization of the fruit tree is of great significance. The study area of Qixia City, Shandong Province, is the study object of the Red Fuji apple tree. In 2013 and 2014, the blooming period of the apple (late April), the long-term (late mid-late of May), the long-term spring tip (in late June), and the long-term (late September) period of the spring tip, and the use of the ASD FieldSpec 4 portable ground-feature wave spectrometer, The spectral data of the apple crown layer in the fruit stage was determined, and the leaves of the apple tree crown were collected synchronously, and the content of the leaf N was measured in the room. In this paper, the changes of N content and the spectral characteristics of the crown layer of apple in different phenological period were analyzed. The original spectral data were pre-treated and 15 kinds of transformation were carried out, and the sensitive wavelengths associated with the nitrogen content were selected. At the same time, the location-based and area-based high-spectral parameters and other three types of vegetation indices are selected. In that end, the optimal model of the nitrogen content of the apple tree crown layer of different phenological period was established. The main results of the study are as follows: (1) The changes of the canopy nitrogen content and the high spectral characteristics of the apple in different phenological period from the blooming period to the fall of the autumn and the average nitrogen content of the apple canopy have been in a descending state. The reduction of long-term nitrogen content from the new shoot to the spring tip is relatively small and relatively stable. The canopy spectral reflectance of apple in different phenological period has a similar trend. In the range of 350-780 nm, the long-term and long-term and long-term and long-term and long-term reflectance of the spring-tip and the long-term end of the spring-tip are relatively small, but the spectral reflectance of the blooming period is higher than the spectral reflectance of the other three periods, and the difference in the vicinity of 680 nm (red valley) reaches the maximum. a high near-infrared reflection platform is formed at 1000-1300 nm, the spectral reflectivity of the blooming period is obviously lower than that of the other three periods, And the long-term spectral reflectance of the new shoots and the long-term spectral reflectance of the spring tip stop long-term spectral reflectance. In the range of 1419 ~ 2402 nm, the spectral reflectance of the four periods is crossed. (2) It is clear that the sensitive wave band and the sensitive high-spectrum parameter green light (500-560 nm) and the near-infrared region (780-1100 nm) of the canopy nitrogen content in different phenological period of the apple are obviously affected by the nitrogen content of the canopy, which is the sensitive band of the nitrogen content of the apple in the blooming period, SDg, CIGREEN, GNDDVI is a sensitive high-spectral parameter of the nitrogen content of the apple in the blooming period. The sensitive band of nitrogen content in the long-term canopy of the new-tip-wang was mainly concentrated in the near-infrared region (780-1100 nm), and the sensitive high-spectral parameters of the nitrogen content of the canopy were DVI, MSAVI and TVI. The sensitive wavelength of nitrogen content in the long-term apple canopy at the spring tip is located in the red region (620-760 nm) and the near-infrared region (780-1100 nm). In the near-infrared region (780-1100 nm), the long-term canopy nitrogen content sensitive area, DVI, MSAVI, TVI and MCARI were the sensitive and high spectral parameters of the long-term apple canopy nitrogen content in the autumn. (3) The optimal estimation model of the nitrogen content of the apple canopy of different phenological period was established, and the model and the estimation accuracy of the nitrogen content of the apple in different phenological period were compared. The best estimation model is the BP neural network model constructed based on the multiple stepwise regression. The estimation precision of the nitrogen content in the apple canopy shows the same change rule, and the precision of the long-term model with the long-term model precision of the spring tip is higher than that of the long-term model. The estimation accuracy of the long-term model of the autumn-tip stop is lower, and the model is not stable. Therefore, it is the best time to estimate the nitrogen content in the apple canopy for a long time.
【学位授予单位】：山东农业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：S661.1;S127

【参考文献】