【作者】 张静；

【导师】 杜加强； 马妙君；

【作者基本信息】 兰州大学 ， 生态学， 2022， 硕士

【摘要】 不透水面(Impervious surface)是地表覆盖的重要组成部分,是衡量城镇化的关键指标。不透水面的持续增加会导致一系列显在或潜在的负面影响,如耕地的损失、地表热岛效应、空气质量恶化和区域气候水文改变等等。准确、及时地掌握不透水面的空间分布格局、量化动态变化信息对于了解区域城镇化进程、制定城镇开发边界至关重要,生成并存档不透水面时序数据集对于土地利用变化、环境变化和城镇可持续发展等研究具有重要参考意义。遥感观测具有覆盖范围广、信息丰富、连续观测的优势,且陆地卫星数据自2008年以来可免费开放获取,为长时间序列中高分辨率不透水面数据集提取提供基础数据源。黄河流域东西地区城镇化差异显著且生境脆弱,适宜开展长时间序列不透水面扩张变化和生态影响分析。本论文基于Landsat系列数据,选取黄河流域作为研究区域,采用随机森林和时间一致性检验的提取方法,评估提取数据集的精度与可靠性,进而分析不透水面的时空变化,识别扩张主要驱动因素并探讨生态效应,主要研究内容和结论如下:(1)1986-2020年黄河流域不透水面数据集构建与精度评定。基于Google Earth Engine平台对Landsat/C02/T1＿L2数据进行预处理,导入筛查后的分类样本集,利用随机森林分类算法得到初始不透水面解译结果;再通过时间滤波和逻辑合理化等后处理步骤,得到最终不透水面时序数据集。利用总体精度、Kappa系数和均方根误差、决定系数,评估时序数据集精度并与已有数据产品开展对比分析,结果表明:随机森林和时间一致性检验是提取不透水面分布的有效途径,尤其是在不透水面覆盖较高的区域。(2)时序不透水面时空演变分析。首先,通过统计黄河流域和其西部、中部、东部以及域内各省会城市1986-2020年不透水面面积、计算不透水面覆盖率分析不透水面动态变化,结果表明:不透水面面积在1986-2000、2001-2010和2011-2020三个阶段分别呈缓慢增长、快速增长和加速增长,中部地区不透水面面积最大,省会城市中郑州、太原和西安不透水面面积最大;省会城市不透水面覆盖率显著高于非省会城市,尤其是中部地区,体现了城镇发展的区域不平衡性。其次,通过不透水面密度和密度变化探究不透水面空间变化,结果表明:不透水面初始分布、过程变化以及最终分布在城镇核心区、郊区或近郊村、远郊村或偏远村三个圈层呈现“高覆盖-低变化-高覆盖”、“低覆盖-高变化-高覆盖”、“低覆盖-低变化-低覆盖”的分层特征。最后,通过扩张速率指数和景观格局指数分析不透水面扩张趋势,结果发现:不透水面扩张在三个阶段间呈增加趋势、阶段内呈减小趋势;不透水面斑块间逐渐增强连通、填充外扩,在景观格局中的主导地位提升。(3)不透水面扩张驱动因素分析。从定性角度分析1986-2020年黄河流域不透水面扩张的驱动因素,总结1986-2000、2001-2010和2011-2020三个时段的主要驱动因素分别为人口扩张、国家战略和房地产热、政府政策和经济增长。从定量角度分析2012-2020年黄河流域不透水面扩张的驱动因素,借助空间化和POI数据以增强回归树(BRT)为模型方法实现相对影响和边际效应研究。BRT结果显示GDP增长、DEM和服务设施密度变化对不透水面扩张的贡献率分别为47.1%、23.8%和12.8%,除DEM为负向遏制外其他均为正向促进。(4)时序不透水面生态效应分析。以NDVI、WET、LST和NDBSI表征植被覆盖、地表湿度、地表温度和地表干度,同时构建RSEI指数表征生态环境质量,重点探究不透水面覆盖率与上述指数间的相关关系,分析结果发现:不透水面覆盖率与NDVI和WET呈负相关,与LST和NDBSI呈正相关;而不透水面覆盖率与RSEI的拟合曲线表明两者间的关系并非单一正/负相关。城镇中心不透水面的致密化和扩张对生态环境质量带来了负向效应,但在城镇以外不透水面低覆盖区域的致密化和扩张自2010年始逐渐显现出正向效应。更多还原

【Abstract】 Impervious surface(IS)is one of the important components of land cover and a key index to measure the urbanization.The continuous increase of IS will lead to a series of visible or potential negative impacts,such as the loss of cultivated land,urban heat island effect,deterioration of air quality,and regional climate and hydrological changes.Accurately and timely grasping spatial distribution pattern and quantifying dynamic change information of IS are crucial for understanding the process of regional urbanization and planning urban growth boundary.Mapping and archiving the time series data sets of IS are of great referenced significance for the research of land use change and environmental change,and the exploration of sustainable urban development.Remote sensing observation have the advantages of wide coverage,rich information and continuous monitoring.since 2008,the Landsat imagery have been freely and openly available,providing a basic data source for the extraction of mediumhigh spatial resolution IS datasets in long time series.In the Yellow River Basin(YRB),the urbanization differs in the eastern versus the western.And the regional ecological environment is fragile,so it is suitable to analyze the spatio-temporal differences and ecological effect of impervious surface.In this paper,YRB were chosen as the study object based on Landsat data,the classification method combining the spectral and textural features based on Random Forest was applied to distinguish IS pixels.We focused on the assessment of the extraction accuracy and reliability,the analysis of spatiotemporal changes and main driving factors,and the identification of ecological effect.The main research contents and conclusions are as follows:(1)Construction and accuracy assessment of impervious surface dataset in the Yellow River Basin from 1986 to 2020.Based on the Google Earth Engine platform,the Landsat/C02/T1＿L2 data are preprocessed.The initial IS interpretation results were obtained by random forest classification algorithm though the classified samples.Then post-processing steps such as time filtering and logical rationalization were performed to get the final IS time series.The overall accuracy and Kappa coefficient,root mean square and error determination coefficient were used to evaluate the accuracy of time series dataset,conduct comparative analysis with existing data products.The results showed that random forest and temporal consistency check were an effective way to extract the distribution of IS pixels,especially in areas with high IS coverage.(2)Temporal and spatial variation analysis of impervious surface.Firstly,the dynamic changes of IS were analyzed by calculating the IS areas of the YRB and the western,central,eastern and provincial capital cities,and calculating the IS coverage rate from 1986 to 2020.The IS areas increased slowly,rapidly and acceleratedly during 1986-2000,2001-2010 and 2011-2020,respectively.The central region had the largest IS areas,and the provincial capital cities had the largest IS areas in Zhengzhou,Taiyuan and Xi’an.The IS coverage rate in provincial capital cities was significantly higher than that in non-provincial capital cities especially in the central region,which reflected the regional imbalance of urban development.Secondly,the spatial changes of the IS were explored through the density and its changes.The results showed that the initial distribution,the process change and the final distribution in the urban core areas,the suburban or rural areas,and the exurb rural or depopulated areas were concluded as “high cover-low change-high coverage”,“low cover-high changehigh coverage” and “low cover-low change-low cover-low coverage” respectively.Finally,the expansion trend was analyzed by using the expansion rate indices and landscape pattern indices.The results showed that the expansion trend increased between the three stages and decreased within the three stages.And the connectivity between IS patches was gradually enhanced,and filling and expanding,and the dominant position in landscape pattern was enhanced.(3)Analysis about driving factors of impervious surface expansion.From the qualitative perspective,the driving factors of IS expansion in the YRB from 1986 to 2020 were analyzed,and the main driving factors in three periods of 1986-2000,2001-2010 and 2011-2020 were summarized.The driving factors of IS from 2012 to 2020 were quantitatively analyzed,and the relative impacts and marginal effects were studied with the help of spatialization and POI data using boosted regression tree(BRT)method.The BRT results indicated that the contribution rates of GDP growth,DEM and service facility density changes to impervious surface expansion were 47.1 %,23.8 %,and 12.8 %,respectively.Except for DEM,which was a negative containment,all others were positive promotions.(4)Ecological effect analysis of impervious surface.NDVI,WET,LST and NDBSI were used to represent vegetation coverage,surface humidity,surface temperature and surface dryness.Meanwhile,RSEI index was constructed to represent ecological environmental quality.We focused on the correlation between impervious coverage rate and the above indexes.The results showed that IS coverage was negatively correlated with NDVI and WET,but positively correlated with LST and NDBSI.However,the fitting curve of IS coverage rate and RSEI implied that the relationship was not a single positive/negative correlation.The densification and expansion of IS in urban core areas had a negative effect on ecological environment quality,but in low cover areas outside the urban areas gradually showed a positive effect.更多还原