考虑红边特性的多平台遥感数据叶面积指数反演方法研究

发布时间：2018-04-18 11:21

本文选题：叶面积指数 + 红边　；参考：《中国科学院大学(中国科学院遥感与数字地球研究所)》2017年博士论文

【摘要】：叶面积指数(Leaf Area Index,LAI)是最重要的植被结构参数之一,是作物长势监测、作物估产、肥水管理等精准农业必备的数据源。遥感技术为大面积、及时获取LAI提供了有效手段。红边波段能够用于研究植物养分及健康状态监测、植被识别和生理生化参数等信息,是定量遥感分析的理论基础。利用不同遥感数据估测植被LAI各有其优劣性,叶面积指数反演过程中需要充分挖掘包含红边波段的不同数据源的特点。例如,高光谱数据红边波段数量多、波段窄,但是存在波段间高度相关、数据冗余的问题;包含单个红边波段的多光谱数据,红边波段较宽,比高光谱数据的红边波段缺少了许多细节;包含多个红边波段的多光谱数据,可以反映更多红边区域的光谱细节,并且由于红边区域反射率迅速上升,红边区域内的不同波段之间存在较大差别,在实际反演中需要进行合理选择。本文针对不同遥感数据源的特点,围绕红边波段进行叶面积指数反演研究,主要研究内容及结论如下:(1)基于近地和航空高光谱数据红边波段的叶面积反演方法研究。基于研究区域采集的近地、航空高光谱数据和田间同步试验测量LAI数据,探究航空和地面高光谱数据红边区域对冬小麦LAI的反演能力。首先,建立高光谱植被指数反演模型,进而研究红边波段组合法和传统波段组合、逐波段组合方法对植被指数反演LAI精度的影响,结果显示在红边区域680-750nm波段范围内,波段组合得到的植被指数与LAI的相关性非常高。最后,针对不同肥水条件下叶面积指数的特征光谱及参数随不同试验条件存在差异,本文基于航空和近地高光谱数据,以及田间实测数据,建立了基于高光谱植被指数MSAVI(Modified Soil-Adjusted Vegetation Index),NDVI(Normalized Difference Vegetation Index)和MTVI2(Modified Triangular Vegetation Index 2)的普适性强、精度高的冬小麦叶面积指数估算模型。(2)基于包含单个红边波段的多光谱卫星数据反演作物叶面积指数方法研究。针对一般红边波段代替红波段的改进植被指数多是基于单一时相、单一作物实现LAI估算中存在的对叶绿素含量的干扰因素考虑不足的缺陷,本文提出基于红边波段和红波段进行组合改进的新植被指数ndviredre(red-edgenormalizeddifferencevegetationindex),msrredre(red-edgemodifiedsimpleratioindex)和ciredre(red-edgechlorophyllindex)。依据田间实测的不同生育时期的四种作物(小麦,大麦,苜蓿,玉米)的叶面积指数和与田间试验准同步的rapideye卫星影像,建立基于植被指数的反演模型,结果证明本文提出的植被指数克服了在多时相和多种类型作物的情况下叶绿素含量的变化对lai反演的影响,有效提高了lai的反演精度,比一般红边波段代替红波段的植被指数反演结果的决定系数提高至少10%。(3)基于包含多个红边波段的多光谱卫星数据反演作物叶面积指数方法研究。面对包含多个红边波段的新发射多光谱卫星在作物参数反演中的研究尚未成熟的情况,本文以搭载两个红边波段的sentinel-2卫星为例,针对不同红边波段之间光谱差异、多个红边波段的波段选择等问题,采用三种叶面积指数反演的经典方法:查找表、神经网络和植被指数法,建立冬小麦叶面积指数反演模型。作为对比,同时利用不包含红边波段的landsat8卫星数据反演,由卫星数据、农学信息、地面实测等多元数据,反演了北京顺义区部分样点的冬小麦叶面积指数。结果表明,具有更高的“时-空-谱”分辨率的sentinel-2卫星,比landsat8卫星反演精度更高。sentinel-2卫星搭载的中心波长为705nm和740nm的两个红边波段,比单个红边波段的多光谱数据(如rapideye)提供了更丰富的红边区域波谱信息,以及更多与lai高度相关的基于705nm和750nm的植被指数的选择。本研究可以为搭载多个红边波段的多光谱卫星数据在植被定量遥感中的应用提供理论依据。本文的研究结论表明高光谱数据红边区域680-750nm波段范围内,植被指数与lai的相关性非常高;基于包含单个红边波段的多光谱卫星数据可以通过结合红边波段的改进植被指数,来抑制叶绿素含量等因素的影响,提高lai的反演精度;多个红边波段的多光谱卫星数据的红边波段之间反射率差异显著,提供了比单波段多光谱数据更加丰富的红边波段信息,有利于丰富植被指数类型选择和lai反演模型精度的提升。以上结论可以为高光谱数据、包含一个或多个红边波段的多光谱数据在作物叶面积指数反演中的应用提供理论依据,为作物生长状态监测、农田管理决策提供可靠的参考信息。同时本文证明了sentinel-2卫星搭载的两个中心波长分别为705nm和740nm的红边波段,在叶面积指数反演中具有重要的应用价值,可以为多光谱传感器的波段设计提供参考依据。
[Abstract]:Leaf area index (Leaf Area, Index, LAI) is one of the most important parameters of vegetation, is the crop growth monitoring, crop yield, fertilizer and water management and precision agriculture the necessary data source. Remote sensing technology for large area, timely access to LAI provides effective means. Red edge band can be used to study plant nutrient and health monitoring vegetation, identification and physiological and biochemical parameters and other information, is the theoretical basis for quantitative remote sensing analysis. Using different remote sensing data to estimate vegetation LAI each has its advantages and disadvantages, the need to fully tap the characteristics of different data sources including red edge band inversion of leaf area index in the process. For example, the bands of hyperspectral data of red edge number, narrow band however, there is a high correlation between bands, the data redundancy problem; multi spectral data contains a single red edge band, red edge band is wide, the red edge wavelength of spectral data is missing a lot of details included; Multi spectral data of a plurality of red edge spectral band, can reflect more details of the red edge region, and the reflectance of red edge area increased rapidly, there is a big difference between different bands of red edge region, in the actual retrieval needs reasonable choice. According to the characteristics of different remote sensing data sources, Research on leaf area index back around the red edge band, the main research contents and conclusions are as follows: (1) study on leaf area and ground inversion method of Airborne Hyperspectral Data Based on red edge band. The study area near acquisition based on Airborne Hyperspectral Data and field test data synchronous measurement of LAI, on air and ground hyperspectral data inversion of red edge area winter wheat LAI. First of all, the establishment of Hyperspectral Vegetation Index inversion model, and studies on the red edge band combination method and traditional band combination, by wavelength combination method on vegetation index Effect of LAI inversion accuracy of the results are displayed in red edge region 680-750nm wavelength range, the correlation between vegetation index band combination with LAI obtained is very high. Finally, according to the spectral characteristics and parameters of leaf area index of different water and fertilizer conditions with different test conditions are different, the air and hyperspectral data based on field test and data, established Hyperspectral Vegetation Index Based on MSAVI (Modified Soil-Adjusted Vegetation Index), NDVI (Normalized Difference Vegetation Index) and MTVI2 (Modified Triangular Vegetation Index 2) of the universal model for estimating winter wheat leaf area index with high accuracy. (2) study on the index method of multi spectral satellite data containing a single crop the red edge band. Leaf area based on improved vegetation index for red edge band instead of the red band is a single phase based on single crop Considering the deficiencies of interference on the chlorophyll content factors of LAI estimation, this paper proposes a new vegetation index ndviredre modified red edge band and red band (based on red-edgenormalizeddifferencevegetationindex), msrredre (red-edgemodifiedsimpleratioindex) and ciredre (red-edgechlorophyllindex). On the basis of the four kinds of crop field measured in different growth stages (wheat, barley, alfalfa, corn) leaf area index and rapideye satellite image synchronization and field experiment, establish the inversion model of vegetation index based on vegetation index results presented in the paper overcomes the impact of changes in multitemporal and various types of crop chlorophyll content under the inversion of Lai, effectively improve the retrieval precision of Lai. The coefficient of determination of vegetation index inversion results than the general red edge band instead of the red band increased by at least 10%. (3) of the index method of multispectral satellite data includes a plurality of crop leaf area based on red edge band. Facing the new launch of multi spectral satellite contains more than one red edge band on crop in parameter inversion is not yet mature, this is equipped with two red bands of the sentinel-2 satellite as an example, according to the the spectral differences between red edge band, a plurality of red edge band band selection problem, the classical method using three kinds of inversion of leaf area index: look-up table, neural network and vegetation index method, established the winter wheat leaf area index inversion model. In contrast, at the same time using the landsat8 satellite data inversion does not contain red edge band from satellite data, information, agriculture, the measured multivariate data of Winter Wheat in Beijing, Shunyi District and some samples of leaf area index inversion. The results show that the higher the space-time and spectral resolution of sent The inel-2 satellite, the center wavelength accuracy of landsat8 satellite is higher than the.Sentinel-2 satellite for two red bands 705nm and 740nm, compared with multispectral data for a single red edge band (such as rapideye) with red edge area spectrum information more abundant, and more highly correlated with Lai and 705nm based on vegetation index the choice of 750nm. This study can provide a theoretical basis for multi spectral satellite data with a plurality of red edge band in the quantitative application of remote sensing in vegetation. The conclusion of this paper shows that the high spectral data in red edge area 680-750nm wavelength range, the correlation between vegetation index and Lai are very high; multi spectral satellite data contains a single red edge the band through a combination of improved vegetation index based on red edge band, to suppress the influence factors such as chlorophyll content, improve the retrieval precision of Lai; a plurality of red edge band multi spectral satellite number The difference between the reflectance of red edge band according to the significant, provides more information than the red edge band single band multi spectral data, enrich the types of vegetation index and Lai inversion accuracy of the model. The above conclusions can enhance the hyperspectral data, and provide a theoretical basis for the application of multi spectral data contains one or more red edge the band in the crop leaf area index inversion, for crop growth condition monitoring, to provide a reliable reference information for decision making in farmland management. At the same time this paper proves that two wavelength sentinel-2 satellite were red edge band 705nm and 740nm, which has important application value in the inversion of leaf area index, can provide a reference as the band design of multispectral sensors.

【学位授予单位】：中国科学院大学(中国科学院遥感与数字地球研究所)
【学位级别】：博士
【学位授予年份】：2017
【分类号】：S127

【相似文献】