Analysis Of Runoff Variation And Optimal Allocation Of Water Resources In The Amu Darya Basin

Water resources and ecological issues in the Aral Sea have attracted global attention.Since the Amu Darya Basin(ADB)is an essential part of the Aral Sea Basin,the changes in its water resources will significantly affect the water volume of the Aral Sea.The growing influence of increasing human activities and escalating climate change are aggravating the social and eco-environmental problems caused by water shortage in the basin.To guarantee sustainable soc-economic development and ameliorate the eco-environment of the Aral Sea,it is imperative to evaluate the trend of future available water resources and water flow into the Aral Sea under the influences of climate change and human activity and to conduct research on the optimal allocation of water resources in the ADB.Herein,machine learning,the hydrological model,and the system optimization method are adopted to research two perspectives,i.e.,the analysis of runoff variation and optimal allocation of water resources.In detail,(1)A Bayesian least squares support vector machine based downscaling method is developed for addressing the mismatch problem between the global climate model and horological model.It can be used to establish a statistical relationship between the large-scale atmospheric circulation factors and basin-scale meteorological elements.The characteristics of climate change in the upper reaches of the basin from 2021 to 2100 are projected,providing technical support for the hydrological model to analyze runoff variation under climate change.The temperature in the upper reaches of the ADB is found to show an obvious upward trend.The order of temperature rise is maximum temperature>average temperature>minimum temperature.Moreover,the temperature rise in winter and spring is more significant than that in summer and autumn.In terms of precipitation,a downward trend is identified with the largest decrease in spring.(2)A multiple climate scenarios-based HBV model is built to explore the impact of climate change on future runoff in the ADB.It can effectively characterize the complex glacier runoff process.Combined with the ensemble prediction under multiple climate scenarios,reduced uncertainty caused by the heterogeneity of climate models and improved accuracy of runoff prediction can be yielded.The predicted runoff from 2021 to 2100 under the ensemble average of multiple climate scenarios presents a downward trend.The seasonal changes are represented by an increase in spring and a decrease in summer.(3)An integrated Bayesian least squares support vector machine based factorial analysis(BLSVM-FA)method is designed through the integration of techniques of Bayesian inference,least squares support vector machine,and factorial analysis into a general framework.BLSVM-FA can quantitatively identify the key anthropogenic,hydrometeorological,and ecological factors affecting the inflow from the Amu Darya to the Aral Sea as well as their interactions.The future trends in the inflow from the Amu Darya to the Aral Sea for 2020 to 2050 are inferred under ensemble prediction.The major factors identified are upstream runoff,agricultural water use in Uzbekistan,reservoir water storage,and evapotranspiration.Among them,the upstream runoff has the largest contribution,while the interactions between two factors contribute about 3.8%.In addition,162 scenarios based on ensemble prediction are analyzed,indicating that the inflow from the Amu Darya to the Aral Sea will be close to the average level of 1970-1980 if the drip irrigation rate reaches 50%at the end of 2050 and the reservoir water storage level reduces to the average value of 1960-1970.(4)A Monte-Carlo-based interval De Novo programming(MC-IDP)method is proposed through the combination of Monte Carlo simulation,interval parameter programming,and De Novo programming within a general framework.MC-IDP has advantages in handling uncertainty presented as interval number,multiobjective,and decision makers' subjectivity problems.In addition,it can construct optimal system design to realize reasonable resource allocation under the constraints of total resources.MC-IDP is then applied to an example of resource planning involving multiple conflicting objectives.Compared to the traditional interval multiobjective programming,it effectively realizes the full utilization of resource and simultaneous optimization of multiple objectives.(5)Combined with the MC-IDP method above,an integrated simulation and optimization optimal model of water resources allocation is established for the ADB.Also,243 scenarios are designed based on the results of runoff prediction under climate change and the inflow under multi-scenario ensemble simulation.It aims to explore the influences of different combinations of upstream runoff,the inflow from the Amu Darya to the Aral Sea and irrigation efficiency on the optimal water resources allocation.Some alternative schemes of optimal water resources allocation are generated,supporting water resources management decisions for the recovery of the inflow and the adaptation to the impact of climate change.In summary,this thesis realizes the ensemble prediction of future runoff in the ADB by integrating multiple climate scenarios,Bayesian least-squares-support-vector-machine based downscaling method,and HBV hydrological model.BLSVM-FA is developed for revealing the driving factors affecting the inflow change and inferring its trend in the future under ensemble prediction.The MC-IDP-based optimal model of water resources allocation is developed for the ADB.The influences of climate change,eco-environment protection of the Aral Sea and irrigation efficiency on the optimal water resources allocation are then analyzed based on the results of runoff prediction and the simulation of the inflow.It will guide decision-makers to make timely adjustments to water resources allocation schemes according to the actual situation,so as to better cope with the continuously changing water resources problems.