Knowledge Resource Center for Ecological Environment in Arid Area
| 非气候因素引起的中国植被变化遥感诊断 —以林业工程为例 | |
| 其他题名 | Nonclimate Triggered Vegetation Trend in China by Remote Sensing: A Case Study of Forest Construction Projects |
| 田海静 | |
| 出版年 | 2017 |
| 学位类型 | 博士 |
| 导师 | 李小文 ; 曹春香 |
| 学位授予单位 | 中国科学院大学 |
| 中文摘要 | 植被在生态系统中起着主导的作用，植物群落是生态系统的主要组成成分，生态系统中的绿色植物是第一性生产者，它们为其他生物的生存提供了赖以生活的有机物质。植被变化不仅受气候因素的影响，也受非气候因素的制约。与气候因素对植被变化的影响相比，非气候因素由于包含的方面太多，且并没有成空间分布的多年连续观测数据，因此非气候因素对植被变化的影响研究难度较大，这方面研究尚存在较多不足，且目前的大多数研究为定性分析，缺乏定量的研究。我们不能够简单的将中国植被恢复的原因归结为气候因素或非气候因素，而应该用科学的方法剔除掉气候因素对植被变化的影响，从而提取出植被变化受非气候因素影响的区域，并对非气候因素进行进一步验证和分析。定量的研究非气候因素对植被变化的影响是非常艰巨的，然而对我们了解植被变化的原因和过程，以及政府管理和部门决策是至关重要的。本研究发挥时间序列遥感数据的优势，在宏观上及时、快速、有效地对1982年到2013年中国植被变化进行定量和定性分析，深入并定量研究植被变化对非气候因素和气候因素的响应，客观地对我国重大林业建设工程对植被变化的影响做出评估，为国家和区域尺度的生态系统保护、恢复与优化管理提供有效的科学依据和决策支持。本文的主要研究工作和结论如下：（1）中国植被变化遥感监测研究。研究中发现不同NDVI的时间序列一致性好并不代表其变化趋势一致性好，因此时间序列变化NDVI趋势的对比和验证分析是基于NDVI进行植被变化研究的基础。除裸地和稀疏植被区外，在中国其它区域植被生长季GIMMS NDVI变化趋势与MODIS NDVI变化趋势一致性较好，说明在这些区域植被生长季GIMMS NDVI变化可以用于反映中国植被长势的变化信息。研究表明最小二乘的线性估计模型和Mann-Kendall非参数估计模型得到的GIMMS NDVI3g变化化斜率及显著性水平空间分布吻合性好，一致性程度高，且变化斜率和相对变化率的空间分布一致性程度高。中国植被变化监测结果表明1982年到2013年中国呈现整体植被恢复，局部区域出现植被退化的现象。其中NDVI显著性增加比例最高的是农田区域（45%），其次是常绿阔叶林（43.9%）和热带稀树草原（39.3%）。且在1982-1991年间和2000-2013年间，中国呈现较为明显的植被恢复现象，而在1991-2000年间植被变化不明显。（2）气候因素对中国植被变化的影响研究。降水量-蒸散量比降水量对中国NDVI空间分布的影响更大，研究中最终选定植被生长季温度、降水量-蒸散量、云覆盖比例作为影响当年植被变化的气候因素。1982年到2013年中国气温上升显著，而降雨量-蒸散量和太阳辐射强度并没有发生明显变化。增温对植被生长有着正负两方面的效应，在中国的中部和南部地区，降水较为充沛，温度的增加使得该区域的GIMMS NDVI3g呈现上升的趋势，而在中国的北部的部分地区，温度增加导致地表蒸散量增大，而降雨量没有显著增加，因此导致地表含水量下降，植被长势变弱。通过NDVI与多气候因子的归一化回归分析，我们得出结论：温度为主导气候因素的区域所占比例最大，且主要分布在中国的中部和南部的半湿润和湿润区域，降水量-蒸散量为主导气候因素的区域主要分布在中国的北部和西部等干旱和半干旱区域，云覆盖对植被变化影响范围较小。（3）非气候因素对中国植被变化的影响研究。研究中利用的是改进的NDVI残差趋势分析方法，主要改进之处有两个方面，一方面研究中充分考虑了温度、水分、太阳辐射对NDVI的影响，这使得分析方法适用于所有的气候区，而不仅仅是干旱和半干旱区域；另一方面在分析NDVI显著变化的驱动因素时综合考虑了NDVI的时间序列变化趋势、NDVI与多气候因子的相关性及显著水平、以及NDVI的残差变化趋势及显著性，并最终定义了6种类型的植被显著变化驱动因素。基于改进的方法得到了中国1982年到2013年植被显著恢复的驱动因素制图。研究结果表明非气候因素对中国植被恢复起到了不可忽视的作用。非气候因素引起的植被显著恢复主要分布在中国中部平原区域的黄河平原混交林，南部的亚热带常绿森林，以及东北部地区的针叶林和嫩江流域草地。在1982-1991年间和2000-2013年间，非气候因素对植被显著恢复发挥了较大的作用，而在1991-2000年间，非气候因素对植被恢复的影响较小。（4）林业工程对植被变化影响的验证与分析。研究发现第八次森林资源调查（2009-2013年）所得到的各省人工林面积与研究所得的各省非气候因素对植被显著恢复的影响面积比例呈现显著正相关，这说明在全国范围内林业工程为引起植被显著恢复的最主要非气候因素，林业建设工程对中国植被长势的增加发挥了较好的作用。在黄土高原和毛乌素沙地两个典型研究区，通过250m分辨率的MODIS NDVI数据和毛乌素沙地2011年野外考察数据，证明了我们提取的非气候因素对植被显著恢复发挥作用的区域对应的是林业工程的实施区。在三北防护林建设规划区，1982年到2013年林业工程和气候因素共同引起的植被显著恢复区域占规划区的8.4%，林业工程主导的植被显著恢复区域占规划区面积的3.5%，工程实施以来在工程区的东部和西南部区域发挥了较好的效果。在天然林保护工程建设规划区，2000年到2013年林业工程和气候因素共同引起的植被显著恢复区域占天然林保护工程建设规划区的12.03%，林业工程主导的植被显著恢复区域占规划区的4.66%，工程实施以来在黄河流域发挥了较好的作用，而在长江流域和东北内蒙古等地区作用不明显。在全国平原绿化工程建设规划区，1988年到2013年林业工程和气候因素共同引起的植被显著恢复区域占规划区的6.5%，林业工程主导的植被显著恢复区域占规划区的5.3%，工程实施以来在华南片、中南华东片、华北西部片、华北东部片发挥了较好的作用。在环京津地区防沙治沙工程规划区，2000年到2013年林业工程和气候因素共同引起的植被显著恢复区域占规划区的9.9%，林业工程引起的植被显著恢复区域占规划区的3.2%，工程自实施以来在工程区的东南部和北部区域发挥了较好的效果。研究发现在西北部干旱的荒漠和半荒漠地带，各大工程没有发挥较好的作用，原因为这些区域自然环境恶劣，加之近30多年来降雨量没有增加，因此这些区域的人工林成活率较低，且受水分条件的限制，即使树木成活，也会存在根系发达、树冠和树叶面积小的现象，因此NDVI变化不明显。本论文所包含的创新点主要有以下几点：（1）首次对比了中国2000年到2013年GIMMS NDVI3g变化趋势和MODIS NDVI 变化趋势的一致性。并证明了NDVI的时间序列一致性好并不代表时间序列变化趋势一致性好，因此时间序列变化NDVI趋势的对比和验证分析是基于时间序列NDVI进行植被变化研究的基础，且在中国除裸地和稀疏植被区外，植被生长季GIMMS NDVI3g变化可以用于反映中国植被长势的变化信息。（2）研究中利用改进的NDVI残差趋势分析方法来区分气候变化和非气候因素对植被变化的影响，改进的方法普适性较好（适用于不同气候区）。主要改进之处有两个方面，一方面充分考虑了温度、水分、太阳辐射对NDVI的影响，这使得分析方法适用于所有的气候区，而不仅仅是干旱和半干旱区域；另一方面在分析NDVI显著变化的驱动因素时，综合考虑时间序列NDVI的变化趋势，时间序列NDVI与气候数据的归一化回归分析和NDVI残差的变化趋势进行植被显著变化的驱动力判定。（3）对非气候因素引起的植被显著恢复区域，从中国和区域尺度，利用第八次森林资源调查数据中各省人工林面积数据和250m的MODIS NDVI数据，以及典型区域实地考察数据进行了验证分析，证明了林业建设工程为影响中国植被显著变化的最主要非气候因素。（4）选取三北防护林建设规划区、天然林保护工程建设规划区、全国平原绿化工程建设规划区、环京津地区防沙治沙工程规划区为典型，基于论文中提出的改进的NDVI残差趋势分析方法分析了规划区内林业建设对植被变化的影响。 |
| 英文摘要 | Vegetation plays a leading role in ecosystems. Plant communities are the main components of ecosystems. Green plants in ecosystems are the primary producers, and they provide the living organic matter for the survival of other organisms. The dynamics of most landscapes are driven by both natural processes and non-climate factors. We can not simply attribute the vegetation restoration in China to climatic factors or non-climate factors, but should use scientific methods to analyze which vegetation changes are mainly affected by climatic factors, and which vegetation changes are mainly affected by non-climate factors. Discerning the different impacts of these two general types of drivers is often formidable, but crucial for both understanding and managing landscapes. In particular, the impacts of non-climate factors on vegetation dynamics can be extremely difficult to be separated out because the influences of climate change and human activity on vegetation are always mixed. Distinguishing vegetation changes induced by non-climate factors from those by climatic variations is crucial to correctly identifying the underlying causes and designing appropriate land use policies.In order to distinguish the influence of climate change and non-climate factors on vegetation change, we must first eliminate the influence of climatic factors, then analyze whether the time series NDVI residual is noise or contains some significant trend caused by non-climate factors. While previous studies did not quantitatively seperate the effects of climate change and non-climate factors on vegetation changes, and no validation was performed. In this study, we try to take advantage of long-term remote sensing data to solve this scientific problem and do some validation.Firstly, the consistency of NDVI trends for different satellite datasets was analyzed in China region. The reliability of GIMMS NDVI and MODIS NDVI in vegetation growing season was firstly verified by the sample plots distributed in Muus Sandland gathered in summer of 2013. Result showed that the growing season averaged MODIS NDVI and GIMMS NDVI can be used to indicate the vegetation growth state. Then the correlation analysis between MODIS NDVI and GIMMS NDVI montly data and annual data from 2000 to 2013 were carried out. Result showed that there was significant positive correlation between MODIS NDVI and GIMMS NDVI data in most regions of China and the growing season averaged NDVI is more stable as compared with annual mean NDVI. Then the GIMMS and MODIS NDVI trends were compared in China region and different landtypes. Result showed that except the barren and sparsely vagetaed area, the NDVI trends for the other landscapes were consistend. So in the next study, we will exclude the barren and sparsely vegetation area and select the growing season averaged GIMMS NDVI3g as the annual vegetation growth state.Secondly, the vegetation change was analyzed by using the growing season averaged GIMMS NDVI3g from 1982 to 2013. At first, the Mann-kendall time series non-parametric estimation model, least squares regression method and relative change rate were compared. Result showed that these three indicators show almost the same spatial pattern and strong positive correlation. Then we use the Mann-Kendall nonparametric estimation method to extract the areas where GIMMS NDVI3g shows significant change from 1982 to 2013. From 1982 to 2013, the vegetation in China shows overall greening and partical degradation, and the percentage of vegetation restoration is highest in croplands (45%), followed by evergreen broadleaf forest (43.9%) and savanna (39.3%). Finally, the trend of GIMMS NDVI in different time periods (1982-1991, 1991-2000, 2000-2013) was analyzed. Result showed that the NDVI increase is obvious during 1982 to 1991 and 2000 to 2013. No significant change was found during 1991 to 2000.Thirdly, the influence of climatic factors on vegetation change from 1982 to 2013 was analyzed. At first, we compare the influence of precipitation and minue between precipitation and evapotranspiration on the spatial distribution of NDVI in China. Result showed that the minue between precipitation and evapotranspiration is more sensitive to NDVI spatial distribution compared with precipitation. And the influence of minus between precipitation and evaporation on NDVI distribution is even strong for the arid regions. Then the climate change (temperature, minues between precipitation and evaporation and cloud cover) from 1982 to 2013 was analyzed. Result showed that from 1982 to 2013 temperature increased significantly, while and the water balance and solar radiation is relatively stable. According to the NDVI-climate multiple drivers regression analysis, vegetation growth in 51.33% of China region is significantly influenced by climatic factors, 25.61% is dominated by temperature, mainly distributed in south and central China, 14.04% is dominated by P-PET, mainly distributed in north China, and 11.68% is dominated by cloud cover, distributed in some part of northeast China. This indicates that temperature is the most important factor that affects vegetation growth in China region and the multiple climate drivers analysis can better simulate the combined influence of climatic factors on NDVI trends and compare their relative effects.Fourthly, imporved NDVI-RESTREND method was used to extract human triggered NDVI trend and 6 kinds of causes for significant NDVI trend was defined. Results showed that climatic factors were the most important factors affecting the vegetation change in China from 1982 to 2013. Non-climate factors, especially positive non-climate factors, also play an important role in vegetation restoration in China. Then the percent of provincial artificial forest area from the 8th national forest resource survey conduct by the national forestry administration was use the make the national validation. Result showed that the percent of provincial artificial forest area shows good correlation with the percent of areas where NDVI significant NDVI increase is correlated with non-climate factors, indicating that the large scale forest construction engineerings are the most important non-climate factors influencing vegetation restoration in China. Besides, we select Loess Plateau and Muus Sandland as the two specific study areas and use the relative high spatial resolution MODIS NDVI data to make validation. Result show that the fractional vegetation cove in the regions we defined as vegetation restoration correlated with non-climate factors change increased a lot from 2000 to 2013. Results showed that the NDVI-RESTREND method is reliable in detecting human-induced vegetation change.Finally, we select the planning regions of the three-north shelterbelt construction project, national plain greening project and Beijing-Tianjin sandstorm source control project as the typical forest construction projects, to see how these projects influence the NDVI trends. Results showed that in the planning areas of the three-north forest construction project, regions where forest construction project is the dominant driver for vegetation restoration accountes for 8.4%, and regions where forest construction project and climate change are the dominant drivers for 3.5% from 1982 to 2013. In the planning areas of national plain greening project, regions where forest construction project is the dominant driver for vegetation restoration accountes for 6.5%, and regions where forest construction project and climate change are the dominant drivers for 5.3% from 1988 to 2013. In the planning areas of Beijing-Tianjin sandstorm source control project, regions where forest construction project is the dominant driver for vegetation restoration accountes for 9.9%, and regions where forest construction project and climate change are the dominant drivers for 3.2% from 2000 to 2013. The study found that in the arid desert and semi-desert areas of the northwest, large scale forest projects did not work obviously. That is because of the harsh natural environment and the low rainfall over the last 30 years, the tree survival rates in these areas is low, while the NDVI information of shrubs and grassland is weak, so that the change is not easy to be detected by the corse resolution mages.In summary, this study confirmed overall greening and partial degradation in China during 1982 to 2013. The NDVI greening in China was attributable to several new land use policies geared toward vegetation conservation and restoration. Through validation with high spatial images, our study has also demonstrated that the NDVI-RESTREND method is a useful tool to help identify human-induced vegetation changes in different landscapes. To effectively use the method, however, one needs check with the high spatial resolution data in combination with the local non-climate factors.The innovations of this thesis are mainly several points as follows:(1) Consistency of NDVI trends for MODIS NDVI and GIMMS NDVI3g was compared in China region for the first time. The reliability of GIMMS NDVI and MODIS NDVI in vegetation growing season was firstly verified by the sample plots distributed in Muus Sandland gathered in summer of 2013. And the monthly NDVI, annual NDVI and NDVI trends for MODIS and GIMMS datasets were compared.(2) Imporved NDVI residual trend method was used to separate the influence of climate change and non-climate factors on vegetation change. In this study, we use the NDVI-multiple climatic factors regeression method to analyze the influence of influence of climatic facors on time series NDVI, and to extract the NDVI residual. Besides, we use not only the NDVI residual trend, but also the results from time series NDVI trends and NDVI-climate multiple drivers regression analysis to define where vegetation significant restoration is related with non-climate factors.(3) As to the validation, we used the provincial 8th forest resource survey to make a validation at the national scale. And use the MODIS NDVI data (250m) at two typical regions to make validation and analysis.(4) Based on the improved NDVI-residual trend analysis method proposed in this paper, we select the planning regions of the three-north shelterbelt construction project, national plain greening project and Beijing-Tianjin sandstorm source control project as the typical forest construction projects, to see how these projects influence the NDVI trends. |
| 中文关键词 | 植被变化 ; 非气候因素 ; 林业工程 ; NDVI残差趋势分析 ; 遥感诊断 |
| 英文关键词 | vegetation trend non-climate factors forest construction projects NDVI residual trend analysis remote sensing diagnostic |
| 语种 | 中文 |
| 国家 | 中国 |
| 来源学科分类 | 地图学与地理信息系统 |
| 来源机构 | 中国科学院遥感与数字地球研究所 |
| 资源类型 | 学位论文 |
| 条目标识符 | http://119.78.100.177/qdio/handle/2XILL650/287846 |
| 推荐引用方式 GB/T 7714 | 田海静. 非气候因素引起的中国植被变化遥感诊断 —以林业工程为例[D]. 中国科学院大学,2017. |
| 条目包含的文件 | 条目无相关文件。 | |||||
| 个性服务 |
| 推荐该条目 |
| 保存到收藏夹 |
| 导出为Endnote文件 |
| 谷歌学术 |
| 谷歌学术中相似的文章 |
| [田海静]的文章 |
| 百度学术 |
| 百度学术中相似的文章 |
| [田海静]的文章 |
| 必应学术 |
| 必应学术中相似的文章 |
| [田海静]的文章 |
| 相关权益政策 |
| 暂无数据 |
| 收藏/分享 |
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。
