| The Yellow River Delta is located on the shore of the Bohai Sea in Shandong Province,where the Yellow River enters the sea and forms a typical coastal estuarine wetland in the world,with important ecological location advantages and conservation values.However,the degradation of the Yellow River Delta wetlands has intensified in recent years due to global climate change and human activities interference,and its landscape ecology is facing serious threats.For this reason,it is urgent to comprehensively explore the landscape patterns and ecological environment evolution patterns of the Yellow River Delta wetlands from multiple scales.It is necessary to provide scientific and reasonable decisions for the healthy and sustainable development of wetland ecosystems.The multi-scale landscape ecological analysis helps to reveal the complexity and variability of the landscape structure of the Yellow River Delta wetlands,which is of theoretical significance to clarify the scale effects of landscape ecology,and of practical importance to promote the cross-scale hierarchical management of the Yellow River Delta wetland ecosystem.Based on the above background,we use six periods of Landsat images from 2000-2022 as the basis for fine-scale mapping of wetlands in the Yellow River Delta.Based on the thematic map of classification,21 landscape indices are selected to explore the response of landscape index granularity at different scales in the whole study area,counties and townships,to determine the optimal granularity at each scale,and to analyze the dynamic changes of multi-scale landscape patterns based on this.Finally,the multi-scale ecological environment evaluation of the Yellow River Delta wetlands is carried out in terms of ecosystem service value and ecological risk evaluation,as well as exploring the spatial correlation distribution between them.The main research works are as follows:(1)A classification model combining hierarchical classification and feature preference is proposed.The Yellow River Delta wetland types are complex and the boundaries are difficult to distinguish.This thesis performs inheritance-based multi-scale segmentation of remote sensing images to determine the best segmentation scale for different land classes.After that,different hierarchical classification frameworks are developed according to the features of remote sensing images in different years,firstly,the major categories are distinguished by thresholding,then the optimal features of different land categories are selected by using recursive feature elimination and cross-validation(RFECV)method,and finally the wetland landscape is classified by random forest.The classification accuracy is stable at 89%-92% in each year,and the applicability of this thesis’ s classification method in the Yellow River Delta wetlands is demonstrated by comparison experiments.(2)The suitable grain size thresholds for each scale are derived based on the grain size response of the landscape indices at different scales,and we analyze the changes of multi-scale landscape patterns in the Yellow River Delta wetlands based on this.A total of 21 landscape indices under area-edge,shape,dispersion and diversity indices are selected to explore the granularity response of the landscape indices at different magnitudes in the vertical hierarchy of the whole,counties and townships.Screening the landscape index with high sensitivity,the optimal particle size under the global,county and township magnitudes are determined to be100 m,90m and 60 m respectively by qualitatively observing the first particle size domain and quantitatively calculating the maximum curvature point.The range and rate of change of landscape indices vary at different scales,and the indicators of aggregation and diversity tend to decrease when the scale becomes smaller.Based on this,the dynamic change analysis of the multi-scale landscape pattern of the Yellow River Delta wetlands is carried out.The results show that the landscape fragmentation and diversity are high in the north and low in the south,high in the east and low in the west,and the natural wetlands decrease significantly in 22 years,and the landscape spatial pattern tends to be dispersed,while the area of artificial wetlands increases and the landscape pattern tends to be aggregated.The landscape fragmentation and diversity of coastal counties and townships are significantly higher than those of inland areas.(3)Based on the grid,county and township scales,the multi-scale ecological environmental changes in the Yellow River Delta wetlands from 2000 to 2022 are accounted for in terms of ecosystem service values and ecological risks,and the spatial correlations between the two are analyzed at the grid and township scales.The results show that the spatial distribution patterns of ecosystem service values and ecological risks at different scales are similar and different,and the spatial distribution patterns at the grid scale and township scale are more similar,which provides a reference basis for the formulation of ecological development plans at each scale.The temporal changes of ecosystem service value and ecological risk structure are also quantified and analyzed,and it is found that the total value of ecosystem services in the Yellow River Delta wetlands has increased,mostly from the benefits of artificial wetlands,but the value of natural wetlands is on a decreasing trend.The ecological protection measures are beginning to be effective,and the high ecological risk areas have been reduced,but the lower risk areas have shifted to the medium and higher risk areas significantly.The high value-high risk areas are mostly distributed in the coastal zone area,where natural and artificial wetlands are staggered,and are both the main supply areas of service value in the study area and also the high-risk areas that should be focused on prevention. |
