Fractal-based Analysis And Simulation Of Hydrological Processes

In a backdrop of the changing climate,hydrologic cycle is enhancing and extreme hydrologic events is more and more frequent.Meanwhile,rapid development of global economy has brought along the industry and agriculture in Tibet,where the topography and hydrologic processes are very complex.Therefore,the needs of comprehensive and practicable study of hydrological processes in Tibet have become not only important but also urgent.To understand the complex hydrological processes in Tibet,new research approaches should be taken into consideration,one of which is the utilization of fractal theory in hydrology.Fractality is an inherent feature of hydrologic data,but the potential of fractal-based methods in hydrology has not been fully unearthed.In this dissertation,a typical small basin named Dong Gu Gou is selected as the main case study.Several necessary hydrological,meteorological and soil monitoring stations were set for the collection of data.The major works and conclusions are as follows:(1)The hydrological,meteorological and soil data are analyzed with runoff index analysis,correlation analysis and joint multifractal spectrum analysis(JMS).Results of runoff index analysis illustrate that,in Dong Gu Gou basin,the accumulation and thawing of snowpack and frozen soil are important components of hydrological processes.Results of JMS analysis on soil moisture-soil temperature-precipitation(SM-ST-P)show the ability of JMS to distinguish the areas and periods with and without frozen soil.Further inference is that JMS is an efficient approach to analyze the most complicated interplay of hydrological components.(2)A hydrological model of Dong Gu Gou basin based on DHS VM is built and calibrated.Objective functions used in sensitivity analysis and calibration of parameters include Nash-Sutcliffe efficiency coefficient(E)and two of its variations(relative form Erel and inversive form Ein).The overall results show that DHSVM performs generally well while the simulation is not very satisfactory for dry periods.Results of sensitivity analysis indicate that Ein is most sensitive to temperature lapse rate and soil water storage related parameters;E and Erel are most sensitive to rain LAI multiplier and evapotranspiration related parameters.Results of multi-objective calibration indicate that models with the best Ein perform the best in dry periods.(3)A novel calibration strategy combining new fractal-theory-based objective ratio of dimensions(RD)and Nash-Sutcliffe efficiency coefficient(E)is proposed and tested.In all three cases,this strategy significantly improves the simulation performance of runoff components(fast flow and baseflow).Besides,the introduction of RD does not impair the value of E.By selecting the model with the best E among those with perfect RD,a single optimal solution of E-RD strategy is selected,making sure that advantages of both objectives are taken into account while no additional work of analysis and management is needed.(4)JMS is utilized to analyze the simulation results of hydrological model of several cases.The universal existence of multifractality in simulated and observed runoff series is firstly proved,which is the precondition of using JMS in hydrological performance evaluation.Analysis of 6 sets of artificial data and 9 calibrated simulations shows the following merits of JMS:JMS gives consideration to every part of hydrograph;JMS strictly presents the parts which are simulated badly;by selecting values of qs and qo,JMS visually displays the performance of all parts.Furthermore,results of JMS suggest that,although the E of simulation of DHSVM and HBV is close,the DHSVM model is better at detailed simulation and is more suitable for rugged mountainous catchments.