Parameter identification and global sensitivity analysis of Xin'anjiang model using meta-modeling approach

Parameter identification, model calibration, and uncertainty quantification are important steps in the model-building process, and are necessary for obtaining credible results and valuable information. Sensitivity analysis of hydrological model is a key step in model uncertainty quantification, which can identify the dominant parameters, reduce the model calibration uncertainty, and enhance the model optimization efficiency. There are, however, some shortcomings in classical approaches, including the long duration of time and high computation cost required to quantitatively assess the sensitivity of a multiple-parameter hydrological model. For this reason, a two-step statistical evaluation framework using global techniques is presented. It is based on （1） a screening method （Morris） for qualitative ranking of parameters, and （2） a variance-based method integrated with a meta-model for quantitative sensitivity analysis, i.e., the Sobol method integrated with the response surface model （RSMSobol）. First, the Morris screening method was used to qualitatively identify the parameters＇ sensitivity, and then ten parameters were selected to quantify the sensitivity indices. Subsequently, the RSMSobol method was used to quantify the sensitivity, i.e., the first-order and total sensitivity indices based on the response surface model （RSM） were calculated. The RSMSobol method can not only quantify the sensitivity, but also reduce the computational cost, with good accuracy compared to the classical approaches. This approach will be effective and reliable in the global sensitivity analysis of a complex large-scale distributed hydrological model.

Hydrological daily rainfall-runoff simulation with BTOPMC model and comparison with Xin'anjiang model

A grid-based distributed hydrological model, the Block-wise use of TOPMODEL （BTOPMC）, which was developed from the original TOPMODEL, was used for hydrological daily rainfall-runoff simulation. In the BTOPMC model, the runoff is explicitly calculated on a cell-by-cell basis, and the Muskingum-Cunge flow concentration method is used. In order to test the model＇s applicability, the BTOPMC model and the Xin＇anjiang model were applied to the simulation of a humid watershed and a semi-humid to semi-arid watershed in China. The model parameters were optimized with the Shuffle Complex Evolution （SCE-UA） method. Results show that both models can effectively simulate the daily hydrograph in humid watersheds, but that the BTOPMC model performs poorly in semi-humid to semi-arid watersheds. The excess-infiltration mechanism should be incorporated into the BTOPMC model to broaden the model＇s applicability.

Application of TOPMODEL in Buliu River Basin and comparison with Xin’anjiang model

Along with the rapid development of computer and GIS technology, hydrological models have progressed from lumped to distributed models. TOPMODEL, a bridge between lumped and distributed models, is a semi-distributed model in which the predominant factors determining the formation of runoff are derived from the topography of the basin. A test application of TOPMODEL in the Buliu River Basin is presented. For the sake of comprehensively evaluating the TOPMODEL, the Xin’anjiang model, a classic lumped hydrological model, was also applied in the basin. The structural differences and the simulation results of the two models are compared and analyzed.

Test of newly developed conceptual hydrological model for simulation of rain-on-snow events in forested watershed

A conceptual hydrological model that links the Xin＇anjiang hydrological model and a physically based snow energy and mass balance model, described as the XINSNOBAL model, was developed in this study for simulating rain-on-snow events that commonly occur in the Pacific Northwest of the United States. The resultant model was applied to the Lookout Creek Watershed in the H. J. Andrews Experimental Forest in the western Cascade Mountains of Oregon, and its ability to simulate streamflow was evaluated. The simulation was conducted at 24-hour and one-hour time scales for the period of 1996 to 2005. The results indicated that runoffand peak discharge could be underestimated if snowpack accumulation and snowmelt under rain-on-snow conditions were not taken into account. The average deterministic coefficient of the hourly model in streamflow simulation in the calibration stage was 0.837, which was significantly improved over the value of 0.762 when the Xin＇anjiang model was used alone. Good simulation performance of the XINSNOBAL model in the WS 10 catchment, using the calibrated parameter of the Lookout Creek Watershed for proxy-basin testing, demonstrates that transplanting model parameters between similar watersheds can orovide a useful tool for discharge forecastin~, in un^au^ed basins.

Complementary system-theoretic modelling approach for enhancing hydrological forecasting

Hydrologic models generally represent the most dominant processes since they are mere simplifications of physical reality and thus are subject to many significant uncertainties. As such, a coupling strategy is proposed. To this end, the coupling of the artificial neural network （ANN） with the Xin＇anjiang conceptual model with a view to enhance the quality of its flow forecast is presented. The approach uses the latest observations and residuals in runoff/discharge forecasts from the Xin＇anjiang model. The two complementary models （Xin＇anjiang ＆ ANN） are used in such a way that residuals of the Xin＇anjiang model are forecasted by a neural network model so that flow forecasts can be improved as new observations come in. For the complementary neural network, the input data were presented in a patterned format to conform to the calibration regime of the Xin＇anjiang conceptual model, using differing variants of the neural network scheme. The results show that there is a substantial improvement in the accuracy of the forecasts when the complementary model was operated on top of the Xin＇anjiang conceptual model as compared with the results of the Xin＇anjiang model alone.

Parameter identification and calibration of the Xin＇anjiang model using the surrogate modeling approach

Practical experience has demonstrated that single objective functions, no matter how carefully chosen, prove to be inadequate in providing proper measurements for all of the characteristics of the observed data. One strategy to circumvent this problem is to define multiple fitting criteria that measure different aspects of system behavior, and to use multi-criteria optimization to identify non-dominated optimal solutions. Unfortunately, these analyses require running original simulation models thousands of times. As such, they demand prohibitively large computational budgets. As a result, surrogate models have been used in combination with a variety of multi- objective optimization algorithms to approximate the true Pareto-front within limited evaluations for the original model. In this study, multi-objective optimization based on surrogate modeling （multivariate adaptive regression splines, MARS） for a conceptual rainfall-runoff model （Xin＇anjiang model, XAJ） was proposed. Taking the Yanduhe basin of Three Gorges in the upper stream of the Yangtze River in China as a case study, three evaluation criteria were selected to quantify the goodness-of-fit of observations against calculated values from the simulation model. The three criteria chosen were the Nash-Sutcliffe efficiency coefficient, the relative error of peak flow, and runoff volume （REPF and RERV）. The efficacy of this method is demonstrated on the calibration of the XAJ model. Compared to the single objective optimization results, it was indicated that the multi-objective optimization method can infer the most probable parameter set. The results also demonstrate that the use of surrogate-modeling enables optimization that is much more efficient; and the total computational cost is reduced by about 92.5%, compared to optimization without using surrogate model- ing. The results obtained with the proposed method support the feasibility of applying parameter optimization to computationally intensive simulation models, via reducing the number of simulation runs required in the numerical model considerably.

微分形式新安江模型

现有新安江模型数学上是代数方程并限于一阶差分方法求解,不可避免存在数值误差,探索数值求解误差控制新途径,对于提高模型计算精度具有重要意义。基于微分系统框架,识别新安江模型的状态变量和通量,推导其控制方程和本构方程,提出微分形式的新安江模型(ODE-XAJ),并采用四阶显式Runge Kutta法求解。数值实验结果显示,以解析解为基准,ODE-XAJ绝对误差处于或小于10^(-4)mm量级,可实现对解析解的高阶近似;以ODE-XAJ结果为基准,按归一化平均绝对误差评价,现有新安江模型的数值求解误差约为8.7%。典型流域应用结果显示,ODE-XAJ确定性系数提升0.02,洪量相对误差降低4.3%。研究表明,ODE-XAJ理论上分离了模型的数学方程与具体解法,可有效控制数值求解误差,提升模型模拟精度。

基于等容瞬时单位线的水文模型研究

流域下垫面空间分布的不均匀性对径流的形成有较大影响,流域蓄水容量曲线解决了流域产流空间异质性的问题,但在采用瞬时单位线进行地表汇流计算时,又将其作为空间分布均匀的净雨处理,局部产流并没有实现局部的汇流。考虑汇流时间与蓄水容量的空间相关性,通过Copula函数构造二者的联合分布,以条件概率的形式反映不同蓄水容量处汇流时间的空间分布特征,进而以等蓄水容量为计算单元,提出了等容瞬时单位线的概念,相当于采用变单位线的方式计算单元产流量的地表径流,从而可以实现局部产汇流计算过程的统一。由此,创建了基于等容瞬时单位线的水文模型IIUHH,并根据Copula函数的类型构建了多种形式的IIUHH,最后选择清江上游作为案例进行验证。结果表明:Clayton型的IIUHH在研究区域表现更优,相比于新安江模型,Nash效率系数可以提高4%~12%,Kling-Gupta效率系数可以提高13%~27%,洪峰相对误差则可降低26%~36%,较大程度上提高了洪峰和洪水过程的模拟精度。研究成果可为流域高精度径流预报提供技术支撑。

基于SWAT与新安江模型场次洪水模拟研究及评估

中小河流洪水大多位于资料短缺的山丘区,具有突发性强,汇流时间短的特点,已成为当前防洪工作的重点和难点。新安江模型和SWAT(Soil and Water Assessment Tool)模型都是应用广泛的水文模型,但其在中小河流洪水模拟中的对比分析及效果评估方面还鲜有研究。以北潦南河为例,建立了日尺度和小时尺度的新安江模型和SWAT模型,评估2种模型在径流深、洪峰流量和峰现时间等关键要素方面的计算效果,分析各自优缺点。结果表明,在率定期间,新安江模型相对径流误差(RRE)、洪峰流量相对误差(RPE)和峰现时间误差(PTE)平均值分别为-2.6%、-4.3%、-0.3 h, SWAT模型RRE、RPE和PTE平均值分别为-4.3%、-3.3%、-0.1 h,新安江模型在RRE方面优于SWAT模型,但SWAT模型在RPE、PTE和多峰洪水模拟方面则优于新安江模型;在验证期得出了与率定期相同的结论。

概念性水文模型与智能模型在中小河流洪水模拟中的比较研究

相较于大江大河的流域水文预报研究,中小流域的研究相对匮乏。以沅江河溪水文站以上流域为例,研究LSTM(Long Short-Term Memory)模型和新安江模型在场次洪水中的模拟效果。通过对比,新安江模型的整体模拟精度较高,洪量、洪峰、峰现时间的平均相对误差分别为9.39%、9.55%、1.6h,确定性系数为0.73,综合合格率为100%,达到甲级精度标准;LSTM模型的模拟精度较低,洪量、洪峰、峰现时间的平均相对误差分别为11.76%、12.33%、2.3 h,确定性系数为0.60,综合合格率为75%,达到乙级精度标准。结果表明,新安江模型和LSTM模型是中小河流洪水预报的有效方法,均可用于河溪流域的正式预报,且对于河溪流域,新安江模型的模拟精度比LSTM模型更高。

考虑融雪及土壤冻融的新安江模型及其应用——以雅砻江上游径流模拟为例

寒区为中国大江大河的主要产流区,其径流预报精度直接影响下游水利工程调度和水资源科学管理。针对传统概念性水文模型较少考虑积雪融雪及土壤冻融过程对径流的叠加影响这一问题,以新安江模型为基础,以降水量和气温为驱动数据,提出了基于“等效土壤温度”和“土壤冻融特征曲线”的概念性土壤冻融模块来刻画积雪融雪及土壤冻融共同影响下的流域产流过程,构建了考虑融雪产流和土壤冻融的新安江模型。以雅砻江甘孜水文站以上流域为典型研究区,开展了流域2017—2022年冰雪消融期逐小时的径流模拟分析。结果表明,相比于HBV模型,考虑融雪产流和土壤冻融的模型可以显著提升流域冰雪消融期的水文模拟精度,各年份消融期径流的纳什效率系数中位数由0.69提升至0.83,相关系数中位数由0.88上升至0.92。本研究所提出的考虑融雪及土壤冻融的新安江模型可以为寒区的水文业务预报提供科学支撑。

机器学习模型在岩溶地区水文预报中的适用性分析

现有岩溶地区的水文预报主要采用基于物理机制的水文模型,而机器学习模型的应用较为罕见。为了探索机器学习模型在岩溶地区水文预报的适用性,以云南省沙甸河流域为研究区,采用了LSTM模型与随机森林模型对倘甸水文站的逐日径流量与场次洪水进行了模拟,并以针对岩溶地区的改进型新安江模型做参照。研究表明:机器学习模型与改进型新安江模型在日径流过程模拟方面都取得较好的效果,LSTM模型模拟效果更优;场次洪水模拟方面,改进型新安江模型达到了甲级预报精度,机器学习模型6 h预见期预报结果整体优于改进型新安江模型,但24 h预见期的预报结果不能满足预报业务的精度需求。对2种机器学习模型和水文模型的特点及预报精度进行的研究,为岩溶地区水文预报工作提供了参考。

基于并行自适应遗传算法的水文模型率定研究

【目的】参数率定是影响水文模型预报精度的重要因素,采用人工智能算法可以有效提高水文模型参数的率定效果。【方法】采用基于种群离散程度的自适应算子,对GA算法的交叉、变异和迁移过程进行自适应优化,并利用粗粒度并行计算模型提高种群进化效率,综合以上手段研究了一种基于自适应策略的并行遗传算法。将传统遗传算法(GA),串行自适应遗传算法(AGA)和并行自适应遗传算法(PAGA),应用于屯溪流域新安江模型的参数率定,从率定效率、率定收敛性、率定稳定性和率定效果四个方面,验证PAGA算法的综合性能。【结果】结果表明:PAGA算法的计算加速效果显著,在10核环境下相对于AGA算法计算时间减少了87.9%;在进化后期,PAGA算法能够更加稳定的收敛于最优解,收敛后的目标函数值具有更好的稳定性;在验证期的场次洪水模拟中,采用PAGA算法率定的模型模拟效果最优,总体洪水合格率大于90%,确定性系数均值为0.85。【结论】PAGA算法能够明显降低模型参数寻优耗时,改善模型率定效果和收敛性能,为水文模型参数的率定提供了新思路。

基于PSO算法的新安江模型在永宁河洪水预报中的应用研究

根据嘉陵江一级支流永宁河流域的气候特征,利用永宁镇水文站2002-2021年连续20年的逐日水文观测资料,基于粒子群算法(PSO)对新安江模型进行参数率定,基于此对永宁镇水文站日流量进行模拟,探讨该模型在永宁镇水文站洪水预报中的应用。结果表明:新安江模型在永宁镇水文站径流模拟中的成果较好,率定期确定性系数平均值为0.53,合格率为71%,验证期验证期确定性系数平均值为0.59,合格率为100%,模拟成果合格率达到乙级,确定性系数达到丙级预报精度,符合规范要求,可用于该站的洪水预报,为该站洪水预报提出了新的方向,并为新安江模型在嘉陵江上游应用积累了经验。

基于逻辑回归模型的流量计算方法优选

在推求水位流量关系时,不同水流状况适用的计算方法往往不同。为自动优选流量计算方法,以新安江水库库区街口水文站为研究对象,提出了基于逻辑回归模型的流量计算方法优选模型。分别采用二线能坡法和落差指数法进行流量计算,借用逻辑回归模型的大数据分析,依据实时水流状况进行流量计算结果优选,并根据率定结果自我学习,不断优化参数,提升优选准确率。结果表明:新安江水库库区回流状况的准确判定对流量计算结果精度影响较大,模型能够依据实时的上下游水位差、代表线流速等水流状况判断当前水流是否出现回流,从而智能优选流量计算方法。该方法推流结果合理,精度高于单一算法。

新安江-海河模型的构建与应用

为研究海河流域下垫面变化对洪水的影响,对新安江模型进行修订,得到新安江-海河模型。选择海河6个典型流域作为研究对象,分别用新安江模型与新安江-海河模型进行产汇流模拟计算,结果表明:洪水分年代、分量级模拟精度较高,但此时新安江模型参数取值不合理,SM值偏大;新安江-海河模型与新安江模型精度相当或稍有提高,参数取值更加合理。

新安江模型的起源及对其进一步发展的建议

回顾新安江模型产生的历史背景,缅怀以赵人俊教授为代表的中国水文学界的先辈们对模型的创建所做的杰出贡献。深入剖析新安江(三水源)模型不同于世界上其他流域水文模型的特点。从模型发展的世界趋势、模型发展中必须引起注意的一些问题和重视数字流域技术开发三方面提出进一步发展新安江流域水文模型的建议。

栅格型新安江模型的参数估计及应用

根据新安江模型与一维扩散波模型理论,构建了基于栅格的新安江(Grid-Xin’anjiang)模型.该模型以DEM栅格为计算单元,将每个单元的产流量划分为地表径流、壤中流以及地下径流3种水源,最后再根据栅格间的汇流演算次序,利用扩散波汇流方法依次将各种水源演算至流域出口.模型在进行产汇流计算时,考虑了栅格之间的水量交换以及河道排水网络的影响.以流域地貌特征、土壤及植被类型等下垫面条件为基础,研究了模型参数的估计方法,并对其进行了验证.将模型用于安徽省屯溪流域的洪水模拟,取得了良好的应用效果.

气候变化下淮河流域极端洪水情景预估

利用IPCC第4次评估公开发布的22个全球气候模式在A1B、A2和B1三种典型排放情景下的未来气温和降水预测结果,结合新安江月分布式水文模型,在对模型验证效果良好的基础上,参照集合预报方法,对未来90年(2010～2099年)气候变化下淮河流域的极端洪水进行预估。研究结果表明,从出现概率来看,淮河流域未来可能发生极端洪水年份的密集程度从大到小依次为A2情景、A1B情景、B1情景。A1B情景下,21世纪下半叶出现极端洪水的可能性增大,A2情景在2035～2065年以及2085年以后是极端洪水发生较为集中的时期。B1情景在21世纪70年代左右发生极端洪水的可能性较大。综合各种极端事件的定义方法,将极端洪水划定3个洪水量级。A2情景预估极端洪水的平均洪量在3种情景中最大,B1情景最小。3种情景未来一级极端洪水发生比例都比历史上偏大,A2情景下增加最多。二级极端洪水都较历史略有减少,三级极端洪水减少最显著。3种情景下各个量级极端洪水所占比例各不相同,A1B和A2情景二级以上极端洪水出现比例较大,B1情景下极端洪水量级多为三级,超1954年的一级极端洪水所占比例较小。

基于代理模型的水文模型参数多目标优化!

针对传统多目标优化算法存在计算复杂且效率偏低的问题,提出了一种基于代理模型的多目标优化方案。以淮河大坡岭水文站以上流域为例,采用多元自适应回归样条方法构建新安江模型参数与不同目标的响应曲面关系,进而估计参数的近似Pareto解集。采用4种目标函数(总水量误差系数、均方根误差、高水流量误差系数和低水流量误差系数)和4种模型精度评价指标(Nash-Sutcliffe效率系数、洪峰流量相对误差、径流深相对误差和峰现时间误差)评定模型优化结果,选择10场洪水过程和4种不确定性评价指标估计Pareto解集的模型预测区间特征。结果表明,代理模型可有效降低模型评估与优化过程中的计算消耗,为实现多目标优化的高效性奠定了基础。此外,不确定性分析结果也进一步验证了方法的有效性和结果的可靠性,为复杂模型参数优化与不确定性评估提供了参考。