Spatial And Temporal Evolution Analysis Of Surface Water In The Yellow River Basin And Ecological Health Of Rivers In The Wei River Basin Based On Multi-Source Remote Sensing Data

Surface water plays an important role in the global water cycle and climate.The Yellow River Basin(YRB)is a significant agricultural and industrial area in China,as well as the cradle of Chinese civilization.Regrettably,numerous cities in the YRB are currently facing critical water resource challenges,including water pollution,soil erosion,and an imbalance between water supply and demand,exacerbated by irregular water resource management practices.These issues are further compounded by the long-term impacts of over-exploitation and climate change,which have only served to exacerbate the situation.This pressing issue requires urgent attention,and there is a need to study the changes and health of surface water resources in the YRB to ensure sustainable socio-economic development and effective water resource management.A comprehensive body of research has been conducted to examine the spatial and temporal evolution of surface water resources and the ecological health of the YRB,including its typical region,the Wei River Basin(WRB).These studies have shed light on the challenges facing the YRB and the potential solutions that can be implemented to address the water resource problems and promote sustainable development.The main contents and results are as follows.(1)The study aimed to analyze the spatiotemporal characteristics of surface water in the YRB based on Landsat satellite remote sensing images.Firstly,all Landsat 5/7/8 images of the Yellow River basin from 2000 to 2020 were pre-processed such as de-clouding,and cross-sensor conversion coefficients were proposed to correct the inter-sensor radiometric bias,and 26,681 high-quality observations were obtained.Secondly,we propose a“multi-indicator surface water extraction rule(MIWER)”by using the percentile synthesis method to collect training samples and combining water bodies and vegetation indices.Finally,the algorithm was used to obtain the spatial and temporal variation of surface water in the study area over 20 years,and the driving mechanism was analyzed by combining the meteorological data of precipitation,temperature and evapotranspiration.Results showed a significant inter-annual variation in the surface water area over the past 20 years in the YRB,with the area of the largest water body showing a significant increasing trend(110 km~2/a,p<0.05).The surface water area also demonstrated distinct seasonal variations with a larger area during the rainy season(June-September)than the dry season(October-May).The minimum surface water extent occurred in January(13,200 km~2),while the maximum occurred in August(15,150 km2).The changes in surface water area were found to be related to climate change and intensive human activities,including precipitation,temperature,evaporation,and glacier melting.Precipitation and human activities were identified as the primary factors influencing the changes in surface water area in the YRB.(2)This study focuses on simulating surface runoff in the WRB,a typical region in the YRB,using both the traditional distributed hydrological model and the deep neural network model for accurate prediction.The Soil and Water Assessment Tool(SWAT)is utilized for the traditional model while the deep neural network model is used to achieve more precise and reliable results.By comparing the performance of these two models,the study aims to provide valuable insights into the feasibility and effectiveness of using deep learning techniques for surface runoff prediction in hydrological modeling.For the traditional distributed hydrological model of SWAT,the impact of dynamic land use changes on runoff prediction was mainly considered.8 input conditions(4 static and 4 dynamic)were set,and the accuracy of runoff prediction results under different input conditions was verified by using the measured runoff data.The results show that the accuracy of runoff simulation under dynamic land use input conditions in both the rate period and the validation period is higher than that of static land use input conditions,and the land use update every five years can achieve higher accuracy of runoff simulation.A CNN-LSTM integrated runoff prediction model is constructed for deep neural network models,in which CNN is used to extract features and LSTM can capture long-term dependencies in time series,which has better applicability and also improves the prediction accuracy.The results show that the SWAT model describes the hydrologic processes in the WRB relatively accurately,and the SWAT model can be used to simulate surface runoff in the geographic context of the WRB;the CNN-LSTM model outperforms independent deep learning models,traditional AI(artificial intelligence)and integrated models;the CNN-LSTM model is compared with the SWAT distributed watershed hydrology model.The daily runoff prediction accuracy of the CNN-LSTM model test set is significantly better than that of the SWAT model,while the monthly runoff prediction accuracy of the SWAT model test set is better than that of the CNN-LSTM model.(3)The evaluation of river ecological health status is crucial for maintaining the well-being of both ecosystems and human society.In this study,we focused on the WRB,and analyzed the spatial and temporal evolution characteristics of river ecological health status based on remote sensing ecological index.We identified multiple influencing factors affecting the semi-arid region,including social,biological,water body,and habitat factors.A river ecological health evaluation index(REHAI)system was constructed to comprehensively present the influence of these multiple factors on ecological health.To obtain the spatial and temporal distribution of river ecological health status during 2015-2022,we established a regression model using the RSEI and river health evaluation index.Our results revealed that the Weihe River system as a whole is in a healthy state,while the Jinghe and Luohe River systems are in a sub-healthy state.Moreover,2019 was a critical year,as the REHAI in the WRB showed a decreasing trend year by year in the previous five years and a rising trend year by year from 2020,putting the basin-wide REHAI in a sub-healthy state in the last two years.The upper reaches of the Wei River were found to be healthier than the middle reaches,and the downstream water environment was the worst.These findings may be attributed to the downstream subsidence,increasing pollution,and rapid urbanization.Our study provides insights into the factors affecting the ecological health of river systems and highlights the need for improved conservation and management strategies to promote sustainable development.