数码相机与MODIS卫星遥感结合的森林物候监测及管理系统初步研建

发布时间：2018-05-08 18:16

本文选题：森林物候 + 数码相机　；参考：《四川农业大学》2015年硕士论文

【摘要】：物候是气候与自然环境变化最直观、敏感的综合指示器和“诊断指纹”,植物的物候信息不仅反映当地当时的环境条件,而且反映过去一段时间环境条件的积累。随着科学技术的发展,植物物候的研究不再局限于传统手段,卫星遥感技术及数码相机延时拍摄技术等被越来越多地运用到植物物候研究中。利用卫星遥感技术,结合数码相机延时拍摄技术记录植被物候变化过程,既节约人力资源和监测成本,又能同时在大区域范围和局部地段内研究植被物候变化。本研究基于MODIS卫星遥感技术和数码相机延时摄影技术,开展四川省森林和都江堰市灵岩山森林公园森林植被物候的研究。本研究采用2014年全年的MODIS地表反射率产品,探测四川省森林植被物候的年度变化。同时,选择四川农业大学教学科研基地——都江堰市灵岩山森林公园为研究样本,以春秋两季颜色变化鲜明的落叶树种银杏和水杉为代表,基于自行设计的森林(树木)物候数据采集系统,实现银杏、水杉冠层数码延时照片的自动获取。采用ENVI/IDL、ArcGIS、TimeSAT等软件,对MODIS卫星遥感时间序列影像和近地面数码相机时间序列数据进行分析,探讨了使用这两种物候监测手段在大区域和景观尺度上提高森林物候监测精细程度的可能性。主要结果如下：(1)利用自主开发的数码相机监测自动管理软件,定时自动获取银杏、水杉植被冠层可见光及红外延时照片。计算数码照片感兴趣区(ROI)、ExG、Gcc、NDVI及EVI等7种植被指数。从各植被指数时间序列曲线看出,只有ExG、Gcc、NDVI能够正确反映森林植被年内生长变化趋势。(2)利用双Logistic模型对数码相机3种植被指数进行平滑去噪处理,得出了较为合理的时间序变化数据。采用动态阈值法提取基于数码相机的植被生长季开始期(SOS)、生长结束期(EOS)、生长季长度(LOS)及生长最大值日期(MOE)。结果表明,3种植被指数中,ExG、Gcc得出的物候关键期较为接近,SOS在130天左右；EOS在天280天左右；LOS在150天左右；MOE在180天左右。表明,基于数码相机监测森林植被冠层物候变化是可行的；基于数码相机的可见光通道植被指数在表征物候变化信息时优于红外通道。(3)对MODIS 8天合成地表反射率产品数据进行预处理,计算NDVI、EVI、ExG及Gcc植被指数,从构建的植被指数时间序列变化情况看出,4种植被指数均能够通过森林冠层物候正确反映森林植被年内生长变化趋势。(4)以EVI植被指数为例,采用不同平滑方法对其进行平滑去噪处理发现,S-G滤波方法在拟合效果及对原始数据保真度方面优于其它方法。采用动态阈值法提取四川森林植被物候关键期,对比4种植被指数发现,EVI植被指数在反映森林植被物候变化方面优于其它3种植被指数。(5)四川森林植被物候关键期空间分布规律：SOS：盆地内部、盆地北部(90-110天)盆地西部、盆地南部、盆地西南部(110-130天)川西亚高山、川西北高原(150-170天)；EOS：盆地西南部(310-330)盆地内部、盆地北部(290-310天)盆地西部、盆地南部(270-290天)川西亚高山、川西北高原(250-290天)；LOS：盆地内部、盆地北部(190-210天)盆地西部、盆地南部、盆地西南部(150-170天)川西亚高山、川西北高原(110-130天)；MOE：各林区中除盆地西南缘林区有超过30%面积大于220天外,其它林区大部分森林生长达最大值日期在180-200天。(6)分析四川森林物候关键期与海拔高度的关系发现,在海拔2000m以下,海拔每上升200m, SOS推迟3.1天,EOS推迟5.1天,LOS延长2天,MOE推迟0.8天；在海拔2000m到4000m之间,海拔每升高200m, SOS提前3.4天,EOS提前8.2天,LOS缩短4.8天,MOE缩短1.5天。海拔4000m以上地区,森林植被物候随海拔变化的规律不明显。(7)比较两种不同监测手段和不同植被指数得到的森林植被物候关键期,发现：卫星遥感和数码相机监测及采用不同植被指数分析得到的结果存在不同程度偏差。总的来说,基于卫星遥感的EVI和基于数码相机的ExG及Gcc较其它植被指数更适合于森林植被物候的研究。(8)利用IDL、C#等初步开发建立了四川森林物候遥感监测系统。该系统包含了数码相机森林物候自动监测分析管理、MODIS卫星森林物候遥感监测分析和数据Web发布3个主要模块,主要实现了森林(树木)物候数码照片的自动获取,数码照片R、G、B和红外通道值的批量提取分析,MODIS卫星数据的预处理,影像分析,模型拟合分析(S-G滤波),森林植被物候参数提取,矢量数据和影像数据叠加显示,基本的地图操作工具和监测结果Wleb发布等。
[Abstract]:Phenology is the most intuitive, sensitive comprehensive indicator and "diagnostic fingerprint" of climate and natural environment changes. The phenological information of plants not only reflects the environment conditions at the time, but also reflects the accumulation of environmental conditions in the past period. With the development of science and technology, the research of plant phenology is no longer limited to traditional means, satellite remote sensing technology. Technology and digital camera delay shooting technology have been used more and more in the study of plant phenology. Using satellite remote sensing technology and digital camera delay shooting technology to record the process of vegetation phenology change, not only save human resources and monitoring cost, but also study the change of vegetation phenology at the same time in large area and part of the area. Based on MODIS satellite remote sensing technology and digital camera delay photography, the forest vegetation phenology in Sichuan forest and Dujiangyan city of Ling Yan Mountain Forest Park was studied. The annual variation of the forest vegetation phenology in Sichuan province was detected by using the surface albedo of MODIS in 2014. At the same time, the teaching and scientific Research of Sichuan Agricultural Uniersity was selected. Base - the Forest Park of Ling Yan Mountain in Dujiangyan city is the research sample, which is represented by the deciduous tree ginkgo and Metasequoia in the two seasons of spring and Autumn period. Based on the self designed forest (Shu Mu) phenological data collection system, the auto acquisition of the digital delayed photo of Ginkgo and Metasequoia canopy is realized. The software of ENVI/IDL, ArcGIS, TimeSAT and so on is used for MODI The S satellite remote sensing time series image and the time series data of the near ground digital camera are analyzed. The possibility of improving the fine degree of forest phenology monitoring in large area and landscape scale using these two kinds of phenological monitoring means is discussed. The main results are as follows: (1) the automatic management software is monitored by the self developed digital camera. Obtain Ginkgo biloba, the visible light and infrared delay photos of the vegetation canopy of Metasequoia metasequoia. Calculate the 7 planting index of digital photo area (ROI), ExG, Gcc, NDVI and EVI. From the curve of each vegetation index time series, only ExG, Gcc, NDVI can correctly reflect the growth trend of forest vegetation in year. (2) using double Logistic model to digital camera 3 A more reasonable time sequence change data was obtained by the exponential smoothing. The dynamic threshold method was used to extract the vegetative growth season (SOS), the growth end period (EOS), the growth Ji Changdu (LOS) and the maximum growth date (MOE). The results showed that the key phase of phenology obtained by the 3 planting index, ExG, Gcc, was obtained. It is close, SOS is about 130 days, EOS is around 280 days, LOS is about 150 days, and MOE is around 180 days. It shows that monitoring forest vegetation canopy phenology based on digital camera is feasible, and the visible light channel vegetation index based on digital camera is superior to infrared channel in characterizing phenological change information. (3) synthesis of MODIS for 8 days on the surface of MODIS The data of reflectivity products were pretreated and the vegetation index of NDVI, EVI, ExG and Gcc was calculated. From the change of the time series of the vegetation index, the 4 planting index can correctly reflect the growth trend of the forest vegetation through the forest canopy phenology. (4) take the EVI vegetation index as an example and use different smoothing methods to smooth it. The denoising process shows that the S-G filtering method is superior to other methods in the fitting effect and the original data fidelity. The dynamic threshold method is used to extract the key phase of the Sichuan forest vegetation phenology. Compared with the 4 planting index, the EVI vegetation index is better than the other 3 planting indices in reflecting the phenological changes of forest vegetation. (5) the forest vegetation in Sichuan. The spatial distribution law of the key period: SOS: the basin, the north of the basin (90-110 days), the south of the basin, the south of the basin, the southwest of the basin (110-130 days), the Western Sichuan subalpine, the Northwest Sichuan Plateau (150-170 days), the southwest of the basin (310-330), the northern basin (290-310 days), the west of the basin and the southern Sichuan subalpine (270-290 days). The Northwest Plateau (250-290 days); LOS: the basin, the north of the basin (190-210 days), the south of the basin, the south of the basin, the southwest of the basin (150-170 days), the Western Sichuan subalpine, the Northwest Sichuan Plateau (110-130 days); the forest area of the southwest margin of the basin is more than 220 days more than 30% areas, and the most of the most forest growth dates in the other forest areas are at the maximum value. 180-200 days. (6) the analysis of the relationship between the critical period of forest phenology and altitude in Sichuan found that below 2000m, elevation of 200m, SOS delayed 3.1 days, EOS postponed 5.1 days, LOS extension 2 days, MOE postponed 0.8 days; at elevation 2000m to 4000m, each elevation of SOS was 3.4 days ahead of advance, EOS 8.2 days, LOS 4.8 days shortened, 1.5 days shortened. The regularity of forest vegetation phenology changes with altitude is not obvious. (7) compare the key period of forest vegetation phenology of two different monitoring means and different vegetation index. It is found that the results of satellite remote sensing and digital camera monitoring and the use of different vegetation indices have different degrees of deviation. Satellite remote sensing EVI and digital camera based ExG and Gcc are more suitable for the study of forest vegetation phenology than other vegetation indices. (8) using IDL, C# and other preliminary development to establish the Sichuan forest phenology remote sensing monitoring system. The system includes the automatic monitoring and analysis of forest phenology of digital cameras, and the analysis of the forest phenology of MODIS satellite remote sensing monitoring and analysis. And data Web issued 3 main modules, which mainly realized the automatic acquisition of forest (Shu Mu) phenological digital photos, the batch extraction and analysis of digital photos R, G, B and infrared channels, the preprocessing of MODIS satellite data, image analysis, model fitting analysis (S-G filtering), the extraction of phenological parameters of the forest vegetation, the superposition of vector data and image data, Basic map operation tools and monitoring results Wleb release.

【学位授予单位】：四川农业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：S716

【相似文献】