Assessment Of Spatio-temporal Variability Of Hydrometeorology And Uncertainty Of Hydrological Modeling In The Upper Reaches Of The Han River

Hydrological model is an important tool for simulating hydrological processes in the basin.Input data of the model is a key factor in building models.Measurement errors in data such as precipitation and evaporation lead to uncertainty in the model's input data.At the same time,the physical assumptions and generalizations in different hydrological models are different,which also brings uncertainty.Therefore,the use o.f scientific methods to separate model input,structure and their interaction on runoff simulation has important theoretical significance and application value for understanding model structure,mproving simulation accuracy and scientifically evaluating hydrological variation characteristics.This paper takes the area above Ankang in the upper reaches of the Han River as the research obj ect.Firstly,the temporal and spatial variabi,lity of the factors such as runoff,precipitation and temperature in the basin are identified.Then,based on the results of the monthly scale ABCD hydrological model,the impact of climate change and human activities on runoff in different characteristic periods are quantitatively assessed by the seasonal scale Budyko hypothesis.After,three hydrological models(HyMod,TOPMODEL,and Xin'anjiang)with different concepts of runoff generation and concentration are constructed to simulate hydrological processes in the study area.In order to clarify the influence of uncertainty factor(meteorological factor and model structure)on runoff simulation at annual scale.,a factorial analysis variance decomposition method based on sub-sampling is used to quantitatively evaluate the impact of meteorological factors,model structure and the interaction between them on runoff simulation.The main achievements are as follows:(1)Using mathematical statistics methods,the change characteristics and trends of runoff,precipitation,temperature,relative humidity,sunshine hours and wind speed sequences in the basin are clarified.The results showed that runoff,precipitation,temperature,relative humidity,sunshine hours and wind speed are all changed.Runoff,precipitation,sunshine hours and wind speed show a downward trend,while temperature and relative humidity both show an upward trend.(2)Using the global and local moran's I method,the spatial autocorrelation of different meteorological factors at the interannual and annual scales is quantified,and the spatial and temporal variability characteristics of climate factors is revealed in the basin.The results show that the spatial autocorrelation of precipitation,temperature and relative humidity factors in the upper reaches of the Han River is a slight downward trend on the interannual scale,while the sunshine hours and wind speed factors show a slight upward trend.On the annual scale,the spatial autocorrelation of the relative humidity in the upper reaches of the Han River is not obvious in 2006 and 2011.The relative humidity generally shows a low correlation in the eastern part of the basin and a high correlation in the central part,which means that the northern part of the basin is humid and the east is arid.The spatial autocorrelation of precipitation in different months is significantly different,especially in April and September,which means that the spatial variation of precipitation in this month is large.The spatial correlation of summer temperature is strong,followed by autumn and worse in spring.(3)Using monthly ABCD model and seasonal Budyko hypothesis,the impacts of climate change and human activities on runoff in different periods are quantitatively assessed.The indicators of NSE,RMSE,R2 and WBE in the calibration period are 0.89,0.89,0.95,0.05%respectively.The indicators of NSE,RMSE,R2 and WBE in the validation period are 0.77,0.89,0.90 and-13.30%respectively.For the interannual scale,climate change is the main reason for the decrease of runoff,accounting for about 75%,followed by human activities,only about 25%.For the scale of the year,the contribution of different seasonal climatic factors and human activity factors to runoff reduction is obvious.In spring,summer and autumn,climatic factors contribute a lot to the reduction of runoff,accounting for 89%,66%and 63%respectively.Correspondingly,the contribution rate of human activity factors accounte for 11%,34%and 37%respectively.While,human activities have a significant impact on runoff,accoun-ting for 84%,and the contribution of climatic factors is only 16%in winter.It mainly due to the fact that there is less precipitation in winter and the regulation of runoff by water conservancy projects(reservoirs and hydropower stations)is significant.In addition,climate change contributes a lot during the dry season and the wet season.And climate change in the dry season is greater than the wet season.(4)Runoff simulation is carried out in calibration period of 1984-1986 and validation periods of 1987-1989.The results show that the hydrological simulation process in the two periods is similar to the actual hydrological process,but the difference between different hydrological models under the same meteorological input data.At the same time,the simulation results by the same hydrological model are also significantly different under different meteorological input data.In general,the combinations of different meteorological input data and hydrological model can obtain good fitting accuracy,but the difference between different time periods(the flood season and the non-flood period)is significant.(5)In order to separate the influence of input data,model structure and their interaction on the runoff simulation,a factorial analysis variance decomposition method based on sub-sampling is used to quantitatively evaluate the impact of climate input data and model structure on runoff in different periods.The results show that with the increase of the number of meteorological stations,the simulation accuracy of different hydrological models increases first and then decreases,and the variation characteristics of confidence intervals are similar.In flood season,precipitation is concentrated,and the uncertainty of meteorological input data is prominent,but it is second to the uncertainty of hydrological model structure.At the same time,with the increase of precipitation,the interaction between meteorological input data and model structure uncertainty is gradually enhanced.In short,the contribution of meteorological input data,model structure and interaction between them is different in different months.Among them,the uncertainty of hydrological model structure is the dominant factor affecting the accuracy of runoff simulation,followed by the uncertainty of meteorological input data,and the interaction between them is the smallest.