【作者】 罗杰；

【导师】 刘绥华；

【作者基本信息】 贵州师范大学 ， 地图学与地理信息系统， 2022， 硕士

【摘要】 石漠化问题——中国西南区域影响最大的生态环境问题之一,它对人们的生存环境和地区的可持续发展都有较大的影响。石漠化信息获取越准确,石漠化治理的效果就越好,进而为深入开展石漠化相关研究提供数据支撑。石漠化己成为我国西南喀斯特地区长期存在的主要生态环境问题。成像空间和时间连续的遥感影像可以获取地表要素多年变化特征,在区域生态环境监测中扮演了不可替代的角色。基于遥感数据建立高效精准的石漠化信息提取模型,定量评价其演变过程,对于喀斯特地区石漠化预测,生态评估与修复具有重要意义。本文选取贵州省威宁自治县的斗古乡作为研究区域,收集了2001,2007,2014和2020年共四年的Landsat TM/OLI遥感影像数据。首先同时基于传统常规方法和NDVI-Albedo特征空间两种方法提取研究区四期石漠化信息,随后利用研究区土地利用类型数据掩膜了不可能发生石漠化区域,并运用2020年Landsat8 OLI影像与同时期的GF-2数据对两种方法的提取结果进行精度验证。在精度验证结果达到研究要求的基础上,最后通过石漠化转移变化矩阵、演变速率、演变模式和演变频率等四个方面来分析研究区石漠化时空变化特征。论文主要结论如下:（1）基岩裸露率、植被覆盖率是石漠化信息提取的重要指标。选取以上两个指标作为传统常规方法进行石漠化信息提取的总精度是69.33%,Kappa系数为0.60。其中中度石漠化的制图精度最低,为45.45%。基岩裸露率值会伴随着石漠化程度的加重而出现增大趋势,即重度石漠化的基岩裸露率最高,轻度石漠化的基岩裸露率值较低。植被覆盖度的图像与基岩裸露率的图像成相反的对应关系。（2）石漠化差值指数模型（RSDDI）能够准确监测和识别不同程度的石漠化。所选的NDVI-Albedo特征空间指标简单、易获取,对研究区石漠化信息提取效果良好,总精度达到了83%,Kappa系数为0.78。因此,NDVI-Albedo特征空间用于提取石漠化是一种很有应用潜力的喀斯特石漠化信息提取方法,在与研究区地质具有相似的区域具有重要的应用潜力。（3）研究区石漠化现象比较严重,尤其是轻度和中度石漠化区域表现尤为凸出,两者面积之和2020年仍占研究区石漠化总面积的53.86%。在空间上的分布格局为西北部、东南部石漠化程度重,东部和北部石漠化程度轻。研究区石漠化长期变化趋势表现为大部分研究区域石漠化得到改善,石漠化得到有效遏制,但局部地区石漠化的演变具有波动性、易反弹。（4）根据石漠化变化转移矩阵、演变速率、演变模式和演变频率等能够反映石漠化演变状况的指标进行时空变化分析,2001-2020年研究区不同程度石漠化之间发生转移变化面积64.05 km2。不同程度石漠化有着不一样的变化速率。石漠化演变模式以续变模式为主,占比为57.24%;重度石漠化演变频率最高,为-3.00%?a^-1。综上,石漠化区域中植被与基岩裸露率、地表反照率的相关关系为判别石漠化提供了坚实的理论依据,本文构建的石漠化差值指数的石漠化信息提取模型,为准确提取石漠化信息提供新的思路,并剖析了2001-2020年石漠化时空变化特征,为石漠化信息的提取与遥感监测分析提供了一定的科学依据。更多还原

【Abstract】 Rocky desertification is one of the most influential ecological and environmental problems in Southwest China,which has great influence on people’s living environment and regional sustainable development.The more accurate the information is obtained,the better the effect of rocky desertification control will be,thus providing data support for further research on rocky desertification.Rocky desertification has become a long-standing major ecological and environmental problem in karst areas of southwest China.Spatial and temporal continuous remote sensing images can obtain the multi-year variation characteristics of surface elements,and play an irreplaceable role in regional ecological environment monitoring.Establishing an efficient and accurate information extraction model of rocky desertification based on remote sensing data and quantitatively evaluating its evolution process is of great significance for rocky desertification prediction,ecological assessment and restoration in karst areas.In this paper,Dougu Township,Weining Autonomous County,Guizhou Province is selected as the research area,and Landsat TM/OLI remote sensing image data in 2001,2007,2014 and 2020 are collected.At the same time,firstly,the information of the fourth stage rocky desertification in the study area was extracted based on the traditional conventional method and NDVI-Albedo feature space.Then,the land use type data of the study area was used to mask the areas where rocky desertification was impossible.Finally,the accuracy of the two methods was verified by using Landsat8 OLI images in 2020 and GF-2 data of the same period.On the basis of the accuracy verification results meeting the research requirements,finally,the temporal and spatial change characteristics of rocky desertification in the study area are analyzed from four aspects:rocky desertification transfer change matrix,evolution rate,evolution mode and evolution frequency.The main conclusions of this paper are as follows:（1）The exposed rate of bedrock and vegetation coverage rate are important indexes for information extraction of rocky desertification.Selecting the above two indexes as the traditional conventional methods to extract rocky desertification information,the total accuracy is 69.33%,and Kappa coefficient is 0.60.Among them,the mapping accuracy of moderate rocky desertification is the lowest,which is 45.45%.The exposed rate of bedrock will increase with the aggravation of rocky desertification,that is,the exposed rate of bedrock in severe rocky desertification is the highest,while that in mild rocky desertification is the lower.The image of vegetation coverage is inversely related to the image of bedrock exposure rate.（2）The difference index model of rocky desertification（RSDDI）can accurately monitor and identify different degrees of rocky desertification.The selected NDVI-Albedo feature space index is simple and easy to obtain,and it has a good effect on extracting information of rocky desertification in the study area,with the total accuracy of 83%and Kappa coefficient of 0.78.Therefore,NDVI-Albedo feature space is an information extraction method of karst rocky desertification with great application potential,and it has important application potential in areas with similar geology to the study area.（3）The phenomenon of rocky desertification in the study area is serious,especially in mild and moderate rocky desertification areas,and the sum of the two areas will still account for 53.86%of the total rocky desertification area in the study area in 2020.The spatial distribution pattern is that the degree of rocky desertification is heavy in the northwest and southeast,and light in the east and north.The long-term trend of rocky desertification in the study area shows that the rocky desertification in most study areas has been improved and effectively curbed,but the evolution of rocky desertification in some areas is fluctuating and easy to rebound.（4）According to the indexes that can reflect the evolution of rocky desertification,such as the transformation matrix,the evolution rate,the evolution mode and the evolution frequency,the spatial and temporal changes were analyzed.From 2001 to 2020,the area of transformation and change between different degrees of rocky desertification in the study area was 64.05 km2.Different degrees of rocky desertification have different change rates.The evolution mode of rocky desertification is mainly continuous,accounting for 57.24%;The evolution frequency of severe rocky desertification is the highest,which is-3.00%?a^-1.To sum up,the correlation between vegetation,exposed rate of bedrock and surface albedo in rocky desertification areas provides a solid theoretical basis for distinguishing rocky desertification.The rocky desertification information extraction model constructed in this paper provides a new idea for accurately extracting rocky desertification information,and analyzes the temporal and spatial variation characteristics of rocky desertification from 2001 to 2020,which provides a certain scientific basis for extracting rocky desertification information and remote sensing monitoring and analysis.更多还原