基于AVHRR和TM的时间序列较高分辨率NDVI数据集重构方法
文献类型：学位论文
| 作者 | 郭文静 |
| 学位类别 | 硕士 |
| 答辩日期 | 2014 |
| 授予单位 | 中国科学院研究生院 |
| 授予地点 | 北京 |
| 导师 | 李爱农 |
| 关键词 | NDVI 数据融合 高时空分辨率 时间序列 ESTARFM |
| 其他题名 | A Study on constructing the time-series NDVI dataset with a high spatial and temporal resolution through fusing AVHRR with TM data |
| 学位专业 | 地图学与地理信息系统 |
| 中文摘要 | 陆地生态系统是人类赖于生存与可持续发展的生命支持系统，全球环境变化对陆地生态系统产生了严重影响，全球气候变化已经引起了各国政府与社会各界的极大关注。植被是陆地生态系统的主体，影响着能量平衡、气候、水文和生化循环，同时又受到气候、水文以及生化等因素的制约，因此，植被活动可以作为气候和人文因素对环境影响的敏感指标。植被指数是对地表植被活动的简单、有效和经验的度量，可以通过植被指数来指示陆地植被覆盖的变化。在较多的植被指数中，NDVI能够准确地反映植被绿度、光合作用强度，植被代谢强度及其季节和年际变化，有较好的时相和空间适应性，应用最为广泛。 由于技术条件的限制，一个传感器很难同时具有高空间分辨率和高时间分辨率。然而，在高分辨率尺度上监测地表景观季节性变化的能力是全球的迫切需要，融合周期短、覆盖范围大与分辨率高、周期长的遥感数据是一种较好的方法。本文基于AVHRR时间分辨率高和TM空间分辨率高及其数据积累时间长的特点，选择若尔盖高原为研究区域，在改进ESTARFM方法的基础上，对TM NDVI和AVHRR NDVI进行融合，构建高时空分辨率的NDVI数据集。 论文的主要内容包括以下几个方面：（1）高原上长期有云雪，每半月（16天）的合成不能完全去除云雪的影响，因此采用S-G滤波对AVHRR NDVI数据和MODIS NDVI数据进行去噪声处理，提高数据源的质量。（2）本研究选取ESTARFM 方法，并根据研究的实际状况对其进行改进；采用改进后的ESTARFM方法，将高重访率低空间分辨率AVHRR GIMMS NDVI数据与重访周期较长的高空间分辨率TM NDVI数据进行融合，获取同时较高时间分辨率和较高空间分辨率的数据，构建高时空分辨率的数据集。（3）采用MODIS NDVI作为真值对所获取的高分辨率数据进行验证和分析，空间上的分析主要从从视觉、散点图和统计直方图三个方面判断其空间信息和时间信息是否提高，时间上的分析主要是在实验区内选择典型的植被类型，分析其NDVI时间过程与植被生长规律的一致性。 通过研究，本论文的如下几点主要结论：（1）本研究基于ESTARFM算法改进，实验区全年每半月（每月前15天和后14~16天）缺失的1km 高分辨率数据的预测，预测结果能在1km尺度上反应空间分布差异的细节，在时间上表现出明显的韵律性，能够反映植被的季节性变化。（2）与MODIS NDVI产品对比分析发现，预测高分辨率NDVI与MODIS NDVI产品相关系数较好，相关系数达到了0.8以上，且两者之间的差值平均值、标准偏差均较低，预测NDVI值域分布集中，成功包含了高分辨率数据的空间信息；研究区内典型地物样点的NDVI时间过程与试验区内相应土地覆被类型的植被绿度变化相一致，所构建数据集的时间过程合理。 综上所述，该方法能够在时间上保留了高时间分辨率数据的时间变化信息，在空间上反应了高空间分辨率数据的空间差异信息，从而为有效构建相对高分辨率时间序列NDVI数据集提供了有效可能的方法工具。 |
| 英文摘要 | Terrestrial ecosystem is a life support system which human depend on to surviving and sustainable development. Global environmental change had a serious impact on the terrestrial ecosystems, and global climate change has caused great concern among governments and the community. Vegetation is the main part of terrestrial ecosystems, affecting the energy balance, climate, hydrology and biogeochemical cycles, meanwhile constrained by climate, hydrology and biochemical factors. Thus, vegetation activity can be used as a sensitive indicator of influence that climatic and human factors impact on the environment. Vegetation index is a simple, effective and empirical measure of vegetation activity, and can be used to indicate changes in terrestrial vegetation. Among the vegetation indexes, NDVI has been widely used, which can accurately reflect the greenness of vegetation, photosynthetic intensity, vegetation metabolic strength and seasonal and interannual variations, and have a better adaptation of phase and space. Due to technical and budget limitations, Remotely sensed data, with high spatial and temporal resolutions, can hardly be provided by only one sensor. However, the ability to monitor seasonal landscape changes at fine resolution is urgently needed for global change science. One approach is to ”blend” the data from coarse-resolution sensors with frequent coverage (e.g. AVHRR) with data from high-resolution sensors with less frequent coverage (e.g. Landsat).To combine the high spatial resolution of Landsat and high temporal resolution of AVHRR data, We selected a study area in Zoige. A method for blending NDVI of different spatial and temporal resolution to produce high temporal-spatial resolution NDVI data set has been developed based on ESTARFM (Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model). The main contents of this paper include the following aspects: (1)due to the long-standing presence of clouds and snow on the plateau, the synthesis of half mounth (16 days) can not completely remove the effects of clouds and snow, so the S-G filtering should be used on AVHRR NDVI data and MODIS NDVI data to remove the noise, and improve the quality of the data source. (2) This study selected the ESTARFM method and improve it based on the actual situation of the study; By adopting the improved ESTARFM method, we fused the low spatial resolution AVHRR GIMMS NDVI data (long-periodic) and high spatial resolution TM (short-periodic) NDVI data to obtain data with high temporal resolution and high spatial resolution, and build data sets with high spatial-temporal resolution. (3) In this paper, we use the MODIS NDVI as the true value to verify and analyze the reconstruction high-resolution data. On the spatial analysis, the study mainly use vision, scatter plots, histograms and statistical information to determine that whether the spatial information and time information are improved. On temporal analysis, the study selecte the typical vegetation types in the experimental area, and analyzed that whether the NDVI time course is consistent with the growth pattern of vegetation. The main conclusions of this paper are as follows: (1) This study realized the prediction of the every half mounth missing 1km resolution data in the experimental area during 2001-2003. The predicted results can reflect the detailed differences of spatial distribution in the 1km scale, and show obvious rhythmicity in time that can reflect the seasonal variations of vegetation.(2) By comparison with MODIS NDVI, we found that the correlation coefficients between predicted high-resolution NDVI and MODIS NDVI product are better, which are large than 0.8. The average difference and the standard deviation between predicted NDVI and MODIS NDVI are lower, and the range of predicted NDVI distribute concentratedly, that successfully containing the high-resolution spatial information data; The NDVI time course of typical feature samples is consistent with greennes |
| 语种 | 中文 |
| 源URL | [http://ir.imde.ac.cn/handle/131551/8041]  |
| 专题 | 成都山地灾害与环境研究所_数字山地与遥感应用中心 |
| 推荐引用方式 GB/T 7714 | 郭文静. 基于AVHRR和TM的时间序列较高分辨率NDVI数据集重构方法[D]. 北京. 中国科学院研究生院. 2014. |
入库方式： OAI收割
来源：成都山地灾害与环境研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。