Knowledge Resource Center for Ecological Environment in Arid Area
| 基于温度植被干旱指数的旱情监测研究 | |
| 其他题名 | Study on Spatiotemporal Variation of Drought Conditions Based on Temperature-Vegetation Dryness Index |
| 刘馨 | |
| 出版年 | 2018 |
| 学位类型 | 硕士 |
| 导师 | 宋小宁 |
| 学位授予单位 | 中国科学院大学 |
| 中文摘要 | 干旱是严重的自然灾害之一，有证据表明，自上世纪70年代以来，干旱等极端气候现象频次不断增加。持续的干旱灾害会直接影响农业生产和经济发展，长期干旱甚至会造成植被减少、河水断流、沙漠化现象出现。因此，加强干旱监测，对于农业管理、旱灾防治、水资源管理及保护都有着极强的现实意义。干旱作为一种复杂的现象难以直接观测，因此通常采用干旱指标进行评估。干旱过程与土壤干湿状况以及作物水分亏缺有着紧密联系。本研究采用的温度-植被干旱指数TVDI (Temperature-Vegetation Dryness Index)结合了基于热红外遥感的地表温度和基于可见光-近红外波段的植被指数两种指示因子，改善了单一使用表征区域旱情的不足，同时解决了植物在水分亏缺时表征现象的滞后性。本研究旨在充分利用光学-热红外遥感结合的TVDI指数模型表征区域旱情的优势，减少对地面辅助数据的依赖，选取了丹麦、中国两个研究区，并分别利用了无人机获取的两天(2017年5月26日、2017年6月18日)的可见光-近红外和热红外遥感影像，以及2007-2016年来日序数第193-257天的MODIS 1km分辨率的地表温度（LST）和植被指数(NDVI)产品，分别反演了丹麦和中国两个地区的TVDI (分辨率分别为0.33m和1km)，从而表征不同区域不同尺度下的区域旱情。围绕这一研究目标，本文在以下几个方面开展了研究：（1）首先对TVDI指数模型进行了分析，并对算法进行了一定改进：在原有的LST-NDVI特征空间的基础上进行了分析，同时进行特征空间的构建，改进了算法，有效地大批量构建时间序列上的LST-NDVI特征空间，并逐一计算了特征空间干湿边线性拟合公式，提高了算法的适用性。（2）考虑到高分辨率遥感数据为土壤湿度研究带来新的可能，研究利用了无人机(UAV)遥感时空分辨率高，机动性强，可获取云下地面信息等优势，获取了高分辨率的多光谱可见光和热红外遥感影像，通过对获取的数据进一步处理，从而得到温度和NDVI数据，进而构建LST-NDVI特征空间；同时同步进行的土壤水分实测实验对构建的TVDI指数进行可信度检验和分析，从而分析丹麦Risoe地区的旱情。其中，遥感反演的温度与通量数据反演温度偏差都在0.43℃以内，利用手持便携光谱仪ASD检验NDVI计算所需的波段反射率，近红外NIR波段和红光R波决定系数R2分别为0.95以上和0.92以上。利用实测土壤水分检验与TVDI的相关性时，同时利用了克里金插值法和缓冲区法提高土壤水分与TVDI的相关性，相关系数R最多能提高0.18。（3）为分析黄河源区的区域旱情时空格局变化，本研究基于2007年至2016年日序数第193-257天的MODIS 1km分辨率地表温度（LST）和植被指数(NDVI)产品反演了10年温度-植被干旱指数(TVDI)。LST-NDVI特征空间干湿边的拟合效果较好，R2均为0.9以上，最高的R2达到0.99。2008-2010年实测土壤水分与TVDI线性关系的相关系数达到0.7，说明TVDI能作为干旱指标有效地指示黄河源区的区域旱情。在年际变化趋势上，10年间日序数第193-257天的TVDI均在0.4-0.6之间，属干旱分级中的正常状况，但东南部地区整体处于干旱-重旱。中、低植被覆盖区域的干旱趋势较为相似，10年来旱情均较为严重，TVDI数值大致均在0.6-1.0间浮动，其中低植被覆盖区所有时相的TVDI在10年内数值均在0.6及以上，表明旱情持续。高植被覆盖区域严重干旱的情况较中低覆盖区域缓解，但是干旱依然在10年间持续发生。旱情的空间变化特征显著，严重旱情多集中于东北部和东南部区域，分布趋势基本与中西部整体土壤水分较充足的客观事实吻合，干旱比较严重的区域主要在东北部和东南部。 |
| 英文摘要 | Drought is one of the serious natural disasters. There is evidence to show that the frequency of extreme weather events such as drought has continued increasing since the 1970s. The continuous drought will directly affect agricultural production and economic development. Long-term drought will even result in the phenomenon of vegetation reduction, river flow cutoff and desertification. Therefore, drought monitoring has great practical significance for agricultural management, drought prevention, water resources management and protection. Drought is difficult to directly observe as a complex phenomenon, so it is usually evaluated by drought indicators; and the drought process is closely related to soil moisture and crop water deficit. TVDI (Temperature Vegetation Dryness Index) combines two indicators, namely land surface temperature based on thermal infrared remote sensing and vegetation index based on the visible-near infrared waveband. TVDI improves the deficiencies that indicate soil moisture with only thermal infrared remote sensed data or only optical remote sensed data, and solve the hysteresis of plants in indicating the phenomenon of water deficit. This study aims to fully utilize the advantages of TVDI combined with optical-thermal infrared remote sensing to indicate soil moisture, and to reduce the dependence on ground auxiliary data. Two research areas, Denmark and China, were selected respectively. Two-day (26th of May, 2017 and 18th of June, 2017) visible-NIR (near-infrared) and thermal infrared remote sensing images from UAVs (Unmanned Aerial Vehicles) are utilized to obtain TVDI in Risoe, Denmark. The Moderate Resolution Imaging Spectroradiometer (MODIS) products, including MOD11A2（8-day land surface temperature） and MOD13A2（16-day vegetation index） in 1 km resolution are used to obtain TVDI over a 10-year period from 2007 to 2016 in the Source Area of the Yellow River, China. The main content of this study can be summarized as follows: (1) Firstly, the TVDI is analyzed and the algorithm is improved. The analysis is based on the LST-NDVI feature space. The algorithm is improved here, to effectively constructs the LST-NDVI feature space in the time series. Besides, the linear fitting formulas for the dry and wet edges of the feature space are calculated, which improves the applicability of the algorithm. (2) To explore the new possibilities of high-resolution remote sensed data, UAVs were applied to obtain high-resolution multi-spectral visible and thermal infrared remote sensed images. Temperature and NDVI data were obtained afterwards, and then the LST-NDVI feature space is constructed. Simultaneously, soil moisture measurements are performed to analyze the credibility of TVDI. The deviation of remote sensed temperature and temperature derived from flux data is within 0.43 °C. Analytical Spectral Devices (ASD, FiledSpec HandHeld 2TM Spectroradiometer) is applied to validate band reflectance for NDVI calculation. The R2 for NIR and R band are more than 0.95 and 0.92, respectively. The Kriging interpolation method and the buffer zone method are also used to increase the correlation between in-situ soil moisture and TVDI. R can be increased by up to 0.18 afterwards. (3) To determine the drought situation over the Source Area of the Yellow River (SAYR) and to analyze the spatiotemporal patterns, the Moderate Resolution Imaging Spectroradiometer (MODIS) products, including MOD11A2（8-day land surface temperature） and MOD13A2（16-day vegetation index） in 1km resolution are used to obtain the Temperature-Vegetation Dryness Index (TVDI) over a 10-year period from 2007 to 2016. Results show that: 1) Significant correlation coefficient of 0.7 occurs in the linear fitting between TVDI and in-situ soil moisture measurements, indicating that TVDI can be recognized as an effective drought indicator over the SAYR for monitoring drought. 2) Average TVDI value over the whole study area ranges between 0.4 and 0.6 in the study period, showing a normal condition in drought classification standard. However, the southeastern region reveals various drought conditions. Moreover, Grassland in different vegetation coverage in the past 10 years has serious drought conditions. Median and low vegetation covered areas show the similar trend with a TVDI varying from 0.6 to 1. Severe drought in high vegetation regions are alleviated comparing to those with low vegetation coverage. 3) The spatial variation of drought is significant. Severe drought mostly occurs in northeastern and southeastern regions. The distribution trend basically corresponds to the fact that the overall soil water content in central and western regions is more adequate, while the regions with severe drought mainly distributed in the northeastern and southeastern regions. |
| 中文关键词 | 地表温度 ; NDVI ; TVDI ; 无人机 ; MODIS |
| 英文关键词 | Land surface temperature (LST) NDVI TVDI Unmanned Aerial Vehicles (UAVs) MODIS |
| 语种 | 中文 |
| 国家 | 中国 |
| 来源学科分类 | 环境科学 |
| 来源机构 | 中国科学院大学 |
| 资源类型 | 学位论文 |
| 条目标识符 | http://119.78.100.177/qdio/handle/2XILL650/288222 |
| 推荐引用方式 GB/T 7714 | 刘馨. 基于温度植被干旱指数的旱情监测研究[D]. 中国科学院大学,2018. |
| 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 |
| [刘馨]的文章 |
| 百度学术 |
| 百度学术中相似的文章 |
| [刘馨]的文章 |
| 必应学术 |
| 必应学术中相似的文章 |
| [刘馨]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。
