Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Doctorat → Chine → 2019 → Detection of Terminal Lake Wetland and Its Landscape Dynamics in Arid Regions

Université du Xinjiang (2019)

Detection of Terminal Lake Wetland and Its Landscape Dynamics in Arid Regions


Titre : Detection of Terminal Lake Wetland and Its Landscape Dynamics in Arid Regions

Auteur : 王敬哲;

Grade : Doctoral Dissertation 2019

Université : Université du Xinjiang

Wetlands are transitional zones between land and water and are widely distributed throughout the world.Wetlands play an important role in the global ecosystem.The wetlands support a large number of wild animals and plants,and have made great contributions to the protection of biodiversity,water source protection,soil erosion control,and regulation of regional climate.As a bridge between aquatic and terrestrial environments,the boundary effects of wetlands promote rich biodiversity and unique wetland ecosystem structures,processes and functions.Wetlands cover at least 7×10 6km 2 of the Earth’s surface,but wetlands are prone to degradation.In fact,nearly 60%of the wetlands around the world have been transformed or lost.In arid regions,the shrinkage of wetlands is more pronounced,and wetlands are an important carrier of water resources in arid and semi-arid regions,and their ecological functions are irreplaceable.Wetlands in arid areas are more sensitive to the fluctuation of climatic factors and the response of human factors.The environmental changes in the wetland area can be deeply reflected by the circulation,extinction,shrinkage and development of wetlands.The temporal and spatial changes of wetlands are bound to be in the arid areas.The quality of the ecological environment,especially water resources,has a profound impact.Therefore,the study of wetland in arid areas has great scientific significance for the rational allocation of water resources,ecosystem function,ecosystem value,landscape pattern,evolution process and its driving mechanism.On the basis of reviewing previous scholars’results on wetland remote sensing,wetland landscape connectivity and its driving factors,this study takes the Ebinur Lake Wetland National Nature Reserve,one of the important wetlands in the country as the research target area.We collected satellite remote sensing data over the past 30 years(1991,1994,1997,2000,2003,2006,2009,2012,2015 and 2017)in spring,summer and autumn months.In addition,regional elevation data DEM(Digital Elevation Model)and corresponding hydrometeorological data were also collected.Besides,Sentinel-1A remote sensing images were collected to further illustrate the dynamic changes of Ebinur Lake(lake wetland)in 2017.Based on the previous studies,the main wetland classification systems of the Ebinur Lake Wetland were identified as:river wetland,lake wetland,artificial wetland,marsh wetlands,saline marsh,and non-wetlands.In the spatial scale,the RF(Random Forest)algorithm was selected as the main classifier.The optimal variables are selected to dynamically detect the surface area of different types in the Ebinur Lake wetland.With ArcGIS,Geoda,FRAGSTATS,Graphab and other software and other operation platforms,we analyzed the spatial structure change,transfer characteristics,spatial-temporal correlation mode,landscape pattern and spatial connectivity of the wetland.In the time scale,based on the collected meteorological hydrology such as temperature,precipitation,hydrological station runoff and lake runoff,we applied the linear tendency estimation,R/S analysis,mutation analysis,wavelet period analysis and other methods for further research.Based on the multi-scale changes of these major meteorological hydrological elements,explore the response mechanism of wetland changes to climate change.The main conclusions of the study are as follows :(1)Under the background of RF classifier,after repeated trials,the number N of fixed decision trees is 100,the number of characteristic variables x is 25,and it is regarded as the optimal parameter setting scheme in the present study.(2)There were significant time and space changes in various land types throughout the study area from 1991 to 2017.Especially in the spring/summer/autumn three different seasons,there is a significant difference in the area and spatial and temporal distribution of various types of wetlands.The average annual area of river wetlands,lake wetlands,constructed wetlands,marsh wetlands,salt marshes,and non-wetlands in the Ebinur Lake Reserve is 110.00 km 2,604.92 km 2,17.08 km 2,760.41 km 2,322.02km 2,and 1292.02 km 2,respectively.(3)A new SAR water index(modified Sentinel 1A water index,MSWI)was proposed.Based MSWI and Sentinel 1A data.The Otsu method was used to select the dynamic classification threshold.Seasonal changes indicate“rising rise”–“significant decline”–“gradual stabilization”phase during the study period.(4)The overall spatial autocorrelation level of the wetland rate of the Ebinur Lake wetland fluctuates,while the annual autocorrelation level of the severe dry season(such as 2015)is extremely low,indicating that the spatial concentration effect of the Ebinur Lake wetland during this period is higher.In the past 30 years,the spatial autocorrelation of the Ebinur Lake wetland rate has fluctuated continuously from 1991to 2017.

Mots clés : Wetland; Remote sensing; Classification; Graph metrics; Landscape pattern; Dynamics;

Présentation (CNKI)

Page publiée le 2 mai 2020