自动多角度光谱观测和遥感技术在碳源汇估算研究中的应用
发布时间：2019-06-11 09:30
【摘要】：陆地生态系统碳循环是当今全球变化研究的热点问题之一,遥感技术由于空间覆盖广并能快捷、非破坏性的对地表植被性质进行探测,在人们探索碳循环过程中具有很好的应用。鉴于此,本研究首先通过改进遥感影像融合方法,为基于遥感的碳收支估算模型提供高时空分辨率地表反射率数据集;其次,在站点观测水平上(站点尺度),利用多角度观测数据,计算冠层尺度光化学反射植被指数的算术平均值,评估光化学反射植被指数监测光能利用率变化的能力,并分析影响光化学反射植被指数与光能利用率相关关系的外部(非生理)因素;然后利用通量塔观测数据得到针对研究区不同土地覆被类型的最大光能利用效率,进而采用NCEP(National Centers for Environmental Prediction)气象再分析资料和MODIS(Moderate-Resolution Imaging Spectroradiometer)光合有效辐射分量(fraction of Photosynthetically Active Radiation,fPAR)数据,将站点观测的总初级生产力(Gross Primary Productivity,GPP)扩展到景观尺度;最后,利用机器学习方法(回归树),footprint模型(Simple Analytical Footprint model on Eulerian coordinates,SAFE-f)和影像融合方法,结合通量塔观测数据和遥感数据,建立完全基于遥感数据的高分辨率净生态系统碳交换量(Net Ecosystem Exchange of CO2)估算模型。得到了以下几点主要结论:(1)优化增强型时空自适应反射率融合模型(Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model,ESTARFM),改进后的影像融合算法预测的类Landsat(Landsat-like)地表反射率具有更高的精度,所合成、预测的高时空分辨率的反射率数据可用于基于遥感数据的碳通量估算模型中。(2)利用自动多角度光谱仪观测的高频率多角度反射率数据,通量塔观测数据分析表明:饱和水气压、浅层土壤温度、总初级生产力、光合有效辐射对光化学反射植被指数和光能利用率的变化都有一定的影响。其中光合有效辐射对两者的变化影响最大,分别为64%和22%;光化学反射植被指数和光能利用率之间具有较好的相关性,在不同的时间尺度上,半小时平均值决定系数(Coefficient of Determination,R2)为0.4084,日平均值决定系数R2为0.7349;在特定的环境因子条件下光化学反射植被指数和光能利用率的相关性表现会更好,光化学反射植被指数对探测胁迫状态下的光能利用率的变化具有较好的敏感性,且敏感性随着环境的改善,如饱和水汽压VPD为20-25hPa、浅层土壤温度为20-25℃、光合有效辐射为300-600 umol.m-2.s-1、总初级生产力40 umolCO2.m-2.s-1的条件下,两者敏感性最佳。(3)采用基于通量塔观测数据的光能利用率模型将站点尺度上的GPP上推到景观尺度。其中,通过光能利用率模型反演得到研究区针叶林和阔叶林的最大光能利用率分别为0.8421 gCMJ-1、1.8082 gCMJ-1,决定系数R2分别为0.7000和0.8345。(4)根据涡度相关通量塔观测数据、遥感时空融合得到的高空间分辨率高时间分辨率数据,利用分类回归树模型,采用机器学习技术构建了高时空分辨率净生态系统碳交换量的估算模型,估算的NEE结果较为合理。总体而言,本文利用遥感数据对陆地表层碳收支相关的光能利用率,GPP和NEE进行了研究,通过研究生态系统碳通量和高光谱连续同步观测,评估了光化学反射指数监测光能利用率变化的能力,并分析了影响光化学反射指数与光能利用率相关关系的外部(非生理)因素,利用影像融合方法提供了高时空分辨率的地表反射率数据,为估算碳收支研究提供了输入数据,为建立高分辨率碳收支估算研究提供参考。
[Abstract]:The carbon cycle of terrestrial ecosystem is one of the hot issues in the research of global change, and the remote sensing technology has a good application in the process of exploring the carbon cycle because of the wide space coverage and the rapid and non-destructive detection of the surface vegetation. in view of that, the present study first provides a high time-space resolution surface reflectance data set for a remote sensing-based carbon revenue and expenditure estimation model by improving the remote sensing image fusion method, and secondly, using multi-angle observation data at the site observation level (site scale), calculating the arithmetic mean value of the canopy-scale photochemical reflectance vegetation index, evaluating the capability of the photochemical reflectance vegetation index to monitor the light energy utilization rate change, and analyzing the external (non-physiological) factors affecting the relation between the photochemically reflected vegetation index and the light energy utilization ratio; The maximum light energy utilization efficiency of different land cover types of the study area is obtained by using the flux column observation data, and then the data of the photosynthesis effective radiation component (fPAR) of the NCEP (National Centers for Environmental Prediction) meteorological re-analysis data and the MODIS (Modern-Resolution Imaging Spectroradiometer) are adopted, The total primary productivity (GPP) of the site observation is extended to the landscape scale; and finally, the machine learning method (regression tree), the footprint model (Simple Analytical Footprint model on Eulian coorates, SAFE-f) and the image fusion method are utilized to combine the flux tower observation data and the remote sensing data, A high-resolution net ecosystem exchange of CO2 model based on remote sensing data is established. The following main conclusions are obtained: (1) The enhanced spatial-temporal adaptive reflectance fusion model (ESTARFM) is optimized, and the modified Landsat-like surface reflectance of the modified image fusion algorithm has higher accuracy, The predicted high temporal resolution reflectivity data can be used in a carbon flux estimation model based on remote sensing data. (2) using the high-frequency multi-angle reflectivity data observed by the automatic multi-angle spectrometer, the data analysis of the flux tower shows that the saturated water pressure, the shallow soil temperature, the total primary productivity, Photosynthesis of effective radiation has a certain effect on the change of the photochemically reflected vegetation index and the light energy utilization rate. The effect of photosynthetically active radiation on the change of the two groups was 64% and 22%, respectively. There was a good correlation between the photochemically reflected vegetation index and the light energy utilization rate. On different time scales, the average of the half-hour mean determination coefficient (R2) was 0.4084. the daily average determination coefficient R2 is 0.749; the correlation performance of the photochemically reflected vegetation index and the light energy utilization rate under a specific environmental factor condition can be better, and the photochemical reflectance vegetation index has better sensitivity to the change of the light energy utilization rate in the detection stress state, And the sensitivity is optimal under the condition that the saturated water vapor pressure VPD is 20-25 hPa, the shallow soil temperature is 20-25 DEG C, the photosynthetic effective radiation is 300-600 umol. m-2.s-1 and the total primary productivity is 40 umolCO2. m-2.s-1. (3) using a light energy utilization model based on the flux-tower observation data to push the GPP on the site scale to the landscape scale. The maximum light energy utilization ratio of coniferous forest and broad-leaved forest in the study area was 0.8421gCMJ-1, 1.8082 gCMJ-1, and the coefficient of determination R2 was 0.7000 and 0.8345, respectively. (4) according to the observation data of the vorticity correlation flux tower, the high-space resolution high-time resolution data obtained by the space-time fusion of the remote sensing, the classification regression tree model is utilized, the machine learning technology is adopted to construct the estimation model of the carbon exchange amount of the high-space resolution net ecosystem, The estimated NEE results are more reasonable. In general, the light energy utilization rate, GPP and NEE related to the land surface carbon revenue and expenditure were studied by the remote sensing data, and the ability of the photochemical reflection index to monitor the light energy utilization rate was evaluated by studying the carbon flux and high-spectrum continuous synchronous observation of the ecosystem. The external (non-physiological) factors that affect the relation between the photochemical reflection index and the light energy utilization rate are analyzed, the surface reflectance data with high temporal and spatial resolution is provided by the image fusion method, and the input data is provided for estimating the carbon and expenditure research. And provides a reference for establishing a high-resolution carbon budget estimation study.
【学位授予单位】：中国矿业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：X171;X87
本文编号：2497086
[Abstract]:The carbon cycle of terrestrial ecosystem is one of the hot issues in the research of global change, and the remote sensing technology has a good application in the process of exploring the carbon cycle because of the wide space coverage and the rapid and non-destructive detection of the surface vegetation. in view of that, the present study first provides a high time-space resolution surface reflectance data set for a remote sensing-based carbon revenue and expenditure estimation model by improving the remote sensing image fusion method, and secondly, using multi-angle observation data at the site observation level (site scale), calculating the arithmetic mean value of the canopy-scale photochemical reflectance vegetation index, evaluating the capability of the photochemical reflectance vegetation index to monitor the light energy utilization rate change, and analyzing the external (non-physiological) factors affecting the relation between the photochemically reflected vegetation index and the light energy utilization ratio; The maximum light energy utilization efficiency of different land cover types of the study area is obtained by using the flux column observation data, and then the data of the photosynthesis effective radiation component (fPAR) of the NCEP (National Centers for Environmental Prediction) meteorological re-analysis data and the MODIS (Modern-Resolution Imaging Spectroradiometer) are adopted, The total primary productivity (GPP) of the site observation is extended to the landscape scale; and finally, the machine learning method (regression tree), the footprint model (Simple Analytical Footprint model on Eulian coorates, SAFE-f) and the image fusion method are utilized to combine the flux tower observation data and the remote sensing data, A high-resolution net ecosystem exchange of CO2 model based on remote sensing data is established. The following main conclusions are obtained: (1) The enhanced spatial-temporal adaptive reflectance fusion model (ESTARFM) is optimized, and the modified Landsat-like surface reflectance of the modified image fusion algorithm has higher accuracy, The predicted high temporal resolution reflectivity data can be used in a carbon flux estimation model based on remote sensing data. (2) using the high-frequency multi-angle reflectivity data observed by the automatic multi-angle spectrometer, the data analysis of the flux tower shows that the saturated water pressure, the shallow soil temperature, the total primary productivity, Photosynthesis of effective radiation has a certain effect on the change of the photochemically reflected vegetation index and the light energy utilization rate. The effect of photosynthetically active radiation on the change of the two groups was 64% and 22%, respectively. There was a good correlation between the photochemically reflected vegetation index and the light energy utilization rate. On different time scales, the average of the half-hour mean determination coefficient (R2) was 0.4084. the daily average determination coefficient R2 is 0.749; the correlation performance of the photochemically reflected vegetation index and the light energy utilization rate under a specific environmental factor condition can be better, and the photochemical reflectance vegetation index has better sensitivity to the change of the light energy utilization rate in the detection stress state, And the sensitivity is optimal under the condition that the saturated water vapor pressure VPD is 20-25 hPa, the shallow soil temperature is 20-25 DEG C, the photosynthetic effective radiation is 300-600 umol. m-2.s-1 and the total primary productivity is 40 umolCO2. m-2.s-1. (3) using a light energy utilization model based on the flux-tower observation data to push the GPP on the site scale to the landscape scale. The maximum light energy utilization ratio of coniferous forest and broad-leaved forest in the study area was 0.8421gCMJ-1, 1.8082 gCMJ-1, and the coefficient of determination R2 was 0.7000 and 0.8345, respectively. (4) according to the observation data of the vorticity correlation flux tower, the high-space resolution high-time resolution data obtained by the space-time fusion of the remote sensing, the classification regression tree model is utilized, the machine learning technology is adopted to construct the estimation model of the carbon exchange amount of the high-space resolution net ecosystem, The estimated NEE results are more reasonable. In general, the light energy utilization rate, GPP and NEE related to the land surface carbon revenue and expenditure were studied by the remote sensing data, and the ability of the photochemical reflection index to monitor the light energy utilization rate was evaluated by studying the carbon flux and high-spectrum continuous synchronous observation of the ecosystem. The external (non-physiological) factors that affect the relation between the photochemical reflection index and the light energy utilization rate are analyzed, the surface reflectance data with high temporal and spatial resolution is provided by the image fusion method, and the input data is provided for estimating the carbon and expenditure research. And provides a reference for establishing a high-resolution carbon budget estimation study.
【学位授予单位】：中国矿业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：X171;X87
【参考文献】
相关期刊论文 前10条
1 于贵瑞;张雷明;孙晓敏;;中国陆地生态系统通量观测研究网络(ChinaFLUX)的主要进展及发展展望[J];地理科学进展;2014年07期
2 郑腾飞;于鑫;包云轩;;多角度高光谱对光化学反射植被指数估算光能利用率的影响探究[J];热带气象学报;2014年03期
3 黄永喜;李晓松;吴炳方;董泰锋;;基于改进的ESTARFM数据融合方法研究[J];遥感技术与应用;2013年05期
4 任小丽;何洪林;刘敏;张黎;周磊;于贵瑞;王辉民;;基于模型数据融合的千烟洲亚热带人工林碳水通量模拟[J];生态学报;2012年23期
5 于贵瑞;方华军;伏玉玲;王秋凤;;区域尺度陆地生态系统碳收支及其循环过程研究进展[J];生态学报;2011年19期
6 于贵瑞;王秋凤;朱先进;;区域尺度陆地生态系统碳收支评估方法及其不确定性[J];地理科学进展;2011年01期
7 王兴昌;王传宽;于贵瑞;;基于全球涡度相关的森林碳交换的时空格局[J];中国科学(D辑:地球科学);2008年09期
8 同小娟;李俊;王玲;;农田光能利用效率研究进展[J];生态学杂志;2008年06期
9 吴朝阳;牛铮;汤泉;;利用光化学植被指数估算叶片的光能利用率[J];兰州大学学报(自然科学版);2008年02期
10 陈晋;唐艳鸿;陈学泓;杨伟;;利用光化学反射植被指数估算光能利用率研究的进展[J];遥感学报;2008年02期
相关硕士学位论文 前1条
1 胡淼淼;北京奥林匹克森林公园植物景观与生态效益初探[D];北京林业大学;2009年
，本文编号：2497086
本文链接：https://www.wllwen.com/shengtaihuanjingbaohulunwen/2497086.html
