Research On Multi-source Remote Sensing Monitoring And Spatiotemporal Change Of Alpine Wetland In Western Sichuan

The western Sichuan plateau(WSCP)has the largest alpine wetland in China.The unique geographical environment and climate system have bred a rich flora and fauna community,and it is also the final habitat of endangered animal-plants and alpine fish species in this area.Due to WSCP in a sensitive and ecologically fragile area of global climate change,coupled with unreasonable human development activities,it has caused ecological and environmental problems such as drought,reverse succession and desertification of alpine wetlands.Therefore,the renewal investigation of alpine wetland resources,the monitoring of wetland ecological environment change,and the analysis of wetland spatiotemporal trends in WSCP are of great significance to regional biodiversity conservation,geochemical cycles and climate regulation,biological resource development,and economic development in ethnic regions.Alpine wetlands are mainly distributed in high-altitude areas in WSCP,and the topography,geomorphology and weather system have a negative impact on the spectral characteristics of satellite remote sensing imaging;In addition,the alpine wetland distribution area has significant temporal and spatial differences in water and heat,and the growing season of the alpine wetland vegetation ecosystem is short,resulting in a narrow remote sensing window for wetland resource investigation and ecological environment monitoring;Multi-source remote sensing is an important data source for wetland resource surveys and wetland ecological environment monitoring.Traditional computing models and methods are used to conduct the multi-source remote sensing integration,long-term series of alpine wetland information extraction and spatial-temporal change monitoring is very difficult.Considering the severe ecological and environmental problems of the alpine wetlands in WSCP and the limitations of the multi-source remote sensing integration-analysis model at the provincial scale.This research provides theoretical reference and technical support for the ecological protection of alpine wetland and the scientific development of alpine wetland resources in WSCP.The main research results obtained in this doctoral dissertation are as follows:(1)A classification scheme of alpine swamp wetland based on remote sensing cloud computing-machine learning algorithm is proposed.Raw remote sensing images have the large volume and coarse resolution to result in poor timeliness of data processing by traditional processing methods in WSCP.To solve this problem,we use the GEE platform to preprocess the Sentinel-2 and Landsat-8 images,and the Landsat-8 MSI band and PAN band were perform the Gram-Schmidt Pan Sharpening.Aiming at the problem of low accuracy of a single classification method for alpine swamps,a multi-classification algorithm based on pixel classification(PBIA)and object-oriented classification(OBIA)and deep learning classification(DL)as a supplement is integrated.Combined with a large sample data set,the extraction of alpine swamp wetland was conducted based on two image-sources of Landsat-8(15m)and Sentinel-2(10m).Significantly improved the accuracy and efficiency of the classification.During 2018-2020,the total resources of the alpine marsh wetland(10m scale)in western Sichuan was 1745736.34 hm²,an increase of 3741.94 hm²compared with the second survey of wetland resources.(2)An alpine river-lake wetland extraction scheme based on cloud computing with composite spectral index is proposed.In view of the fact that the traditional single water index is affected by terrain and shadows,it is difficult to extract the whole area of rivers and lakes and wetlands in western Sichuan,and a wetland partition-extraction scheme based on NDWI-2,MNDWI and EWI composite water index is designed.Aiming at the problem of low operating speed of traditional software-based methods for extracting water index.Using the cloud computing platform to write the code of the multi-index,rapidly extract the alpine river-lake wetlands and accurately estimate wetland resources in western Sichuan.The extraction results show that from 2018 to 2020,the alpine lakes and river wetlands are 31560.11hm²and 159418.00hm²,respectively.Compared with the second Sichuan wetland resource survey under the same standard,lake and river wetlands increased by 650.20 hm²and 410.45 hm²respectively.(3)An alpine wetland ecological environment monitoring program based on big-data sets and the LRM-MKM-SSM coupling models is proposed.In response to the problems of complex wetland monitoring indicators and low availability of relevant survey data,we integrated international open-source satellite-driven products to establish a spatial big data set for monitoring the ecological environment of the alpine wetland in western Sichuan,including wetland landscape,wetland hydrology,wetland biomass,wetland climate and external disturbances.This data collection makes the monitoring index system more comprehensive and data acquisition more efficient and feasible.The traditional monitoring scale is too large or too small,making the monitoring results unable to reflect the overall trend and local characteristics of the wetland ecological environment changes.Design a multi-scale wetland ecological environment monitoring system,which specifically includes:sample scale(?5²km),plot scale(60km×70km)and region scale.Aiming at the problem of low efficiency of time series analysis of ecological big data indicators and no cross-validation,an LRM-MKM-SSM coupling model based on cloud computing is established.It realizes the change trend detection and accuracy verification of the ecological environment of the alpine wetland,and improves the efficiency and accuracy of the traditional single model.(4)An alpine wetland drive mechanism and change simulation plan based on multi-source drive and combined evaluation is proposed.Aiming at the problem of insufficient understanding of the driving forces of temporal and spatial changes in alpine wetlands.We integrate wetland thematic data,satellite-driven products,and socio-economic data to establish a set of 20 driving factors,covering climate factors,geographic landscape,water supply,ecological functions,geological disasters,and social economy.Through grid processing,collinearity diagnosis,optimal scale determination and logistic analysis,the analysis of the driving mechanism of the spatiotemporal changes of alpine wetland is completed.The results reveal that geological conditions and climate systems are the causes of the spatial distribution of alpine wetlands,while climate systems and human activities are the main factors driving changes in wetlands.Aiming at the problem that a single model is difficult to verify and compare in the simulation of the spatio-temporal changes of alpine wetlands.We use BIOCLIM,DOMAIN,MAXENT,and GARP model,integrate the driving force factor and World Climate model data to simulate the distribution status of the western alpine wetland.The results show that with the increase of CO₂ emissions and the intensity of human activities,the alpine wetlands in West Sichuan are gradually regressing to high latitudes and high-altitude areas.