Deterministic Insight into ANN Model Performance for Storm Runoff Simulation

The artificial neural network (ANN) theory has been widely applied to practical applications in hydrology. Since watershed rainfall–runoff processes are nonlinear and exhibit spatial and temporal variability, the ANN model, which considers watershed nonlinear characteristics, can usually but not always obtain satisfactory simulation results. The training of an ANN network is based completely on the reliability of the available hydrologic records. The objective of this study was to provide deterministic insight into the limitations of storm runoff simulation when using ANN. Hydrologic records of 42 storm events from two watersheds in Taiwan were adopted for analysis. A deterministic runoff model was used to classify the hydrologic records into “usual” and “unusual” storm events. The analytical results show that the ANN model could provide good simulation results for “usual” storm events; however, its performance was poor when it was applied to “unusual” storm events because no consistent hydrologic characteristics could be extracted from the storm event records using ANN. The success of the ANN model in usual storm discharge simulations may be mainly due to the input vectors including the previous observed discharge. Moreover, the number of past periods of rainfall that were set as the input vectors of the ANN model was found to be highly correlated with the watershed time of concentration. It can be used to efficiently determine the ANN network structure instead of using iterative network training.     

Parameters Optimization using Fuzzy Rule Based Multi-Objective Genetic Algorithm for an Event Based Rainfall-Runoff Model

The calibration of an event based rainfall-runoff model for steam flow forecasting is challenging because, it is difficult to measure the parameters physically on the field for each rainfall event. In the present study, Fuzzy rule based Multi-objective Genetic Algorithm (MGA) is developed to optimize the infiltration and roughness parameters of an event based rainfall-runoff model. Nash Sutcliffe Efficiency (NSE), Coefficient of Determination (R²) and transformed volume difference (f(V)) are used as the objective functions of the MGA and all Pareto optimal solutions are identified using Nondominated Sorting method. As three objective functions are included in the calibration, the number of Pareto optimal solutions are also increases and hence, the optimization problem now becomes a decision making problem. Therefore, to select the best solution from all Pareto optimal solutions, a Fuzzy Rule-Based Model (FRBM) is developed to get alternative values of each Pareto optimal solution. First, the Fuzzy rule based MGA is developed by integrating the FRBM with the MGA. Then the Fuzzy rule based MGA is integrated with an event based runoff model. The developed Fuzzy-MGA based runoff model is tested on three different watersheds and the simulation results of Fuzzy-MGA based runoff model are compared with observed data and previous study results. From the simulated events of three watersheds using Fuzzy-MGA based runoff model, it is observed that the mean percentage error in any criteria (i.e. volume of runoff, peak runoff, and time to peak) of the developed model for a watershed is less than 16.33%. It is also noted that the developed Fuzzy-MGA based runoff model is able to produce hydrographs that are much closer to the measured hydrographs.     

Efficacy of Nakagami-<Emphasis Type="Italic">m</Emphasis> Distribution Function for Deriving Unit Hydrograph

This paper applies the Nakagami-m distribution for the derivation of unit hydrograph (UH). The applicability of this distribution was verified using the data from 13 watersheds and results were compared with other distributions, viz., Gamma (GM), Beta, normal (NL), log-normal (LN), Weibull (WB), logistic (LG), generalized logistic (GLG) and Pearson type 3 (PT3). Based on visual comparison as well as statistical measures, such as root mean square error (RMSE), coefficient of efficiency (CE), mean absolute percentage error (MAPE), and application efficiency (η _dist.), it was found that the Nakagami-m distribution yielded satisfactory UHs and direct runoff hydrographs for watersheds of various sizes.     

FFMS模型与HEC-HMS模型在暴雨洪水预报中的应用比较

针对小流域暴雨洪水预报难的问题,利用模块化小流域暴雨洪水预报FFMS(Flash Flood Modul Simulation System)模型和HEC-HMS(Hydrologic Engineering Center Hydrological Model System)模型,以河南栾川、韩城及辽宁郝家店、梨庇峪4个山丘区小流域为例,对比分析了4个流域的暴雨洪水预报过程,以洪峰相对误差、峰现时间误差以及纳什系数等为评价准则,比较和分析了2个模型的预报精度和适用性。实例验证结果表明:虽然2种模型均能实现小流域暴雨洪水的预报,但从3个评价准则的结果来看,FFMS模型的预报精度优于HEC-HMS模型。研究成果证明了FFMS模型在山丘区小流域暴雨洪水预报中的有效性和可行性,可以在类似山丘区小流域暴雨洪水预报中进行推广应用。     

Development of Geomorphology Based Regional Nash Model for Data Scares Central India Region

The development of rainfall runoff relationship for ungauged watersheds using topography, geomorphology and other regional information remains the most active area of research in the field of hydrology. In the developing countries, some thumb rules and very old equations are in practice for designing water resources structures which sometimes provide erroneous results. In the proposed study, regional relationships have been developed for computation of peak velocity and scale parameters of Nash model using geomorphological and fluvial characteristics of 41 watersheds of varying characteristics in Central India region. The regional relationships developed to determine scale parameter (k) of Nash model from a morpho-fluvial factor, has facilitated derivation of at-site regional and regional only instantaneous unit hydrograph (IUH), unit hydrograph (UH) and direct surface runoff (DSRO). The performance of proposed regional model has been evaluated using spatial correlation coefficient, integral square error, relative mean absolute error, root mean square error, relative error in peak, coefficient of residual mass and model efficiency. The response of proposed regional model have been found comparable with the observed values as the Nash-Sutcliffe efficiency of proposed model during calibration varies from 69.7 % to 95.2 % for site specific approach, 60.6 % to 97.7 % for at-site regional and 67.1 % to 98.7 % for regional only approach. Similarly, the performance of proposed model have been found satisfactorily during validation as the efficiency varies from 81.3 % to 99.9 % for site specific approach, 83.5 % to 99.9 % for at-site regional and 82.7 % to 99.9 % for regional only approach. The simple regional relationships developed in the study can be used for event based rainfall-runoff modeling and estimation of design flood in ungauged catchments of central Indian region.     

Flow Forecasting in a Watershed using Autoregressive Updating Model

A real-time autoregressive updating model is proposed in this study to forecast the flow in a watershed. The model has two components: (1) Finite Element-Event based distributed rainfall runoff model for runoff simulation and (2) Autoregressive model for updating the error forecast. The efficiency of the runoff updating model depends on the accuracy of the rainfall. Forecasting plays a major role in view of the lead time. In the present study, forecasting is carried out with a lead period of 1 to 3 h. The performance of the integrated model is tested using Nash Sutcliffe efficiency (E) and correlation coefficient (r). The integrated model is applied for Banha, Harsul and Khadakohol watersheds in India. From the results, it can be concluded that the developed model is efficient in flow forecasting on real-time basis in the watersheds.     

Case study evaluation of the geomorphologic instantaneous unit hydrograph

Several issues related to the probabilistic and hydraulic structure of the geomorphologic instantaneous unit hydrograph (GIUH) are addressed. These issues are: (1) accuracy of the geomorphologic expressions of the probabilities of surface runoff movement in a watershed, (2) identifying, for a given storm, a representative time-average velocity for surface runoff, (3) estimation of this velocity for the ungauged watersheds and effect of velocity estimation errors on the GIUH predictability, and (4) suitability of incorporating a linear expression for infiltration in the GIUH as well as the estimation problem of the infiltration coefficient and its effect on the reliability of predicted hydrographs. These issues are analyzed through application of the GIUH for two gauged watersheds in the southwestern region of Saudi Arabia. Twelve storm events are used in the analysis and the results are presented.Irrigation and Hydraulics Department, Faculty of Engineering, Cairo University, Giza Egypt.     

A regional suspended load yield estimation model for ungauged watersheds

Developing regional models using physiographic, climatic, and hydrologic variables is an approach to estimating suspended load yield (SLY) in ungauged watersheds. However, using all the variables might reduce the applicability of these models. Therefore, data reduction techniques (DRTs), e.g., principal component analysis (PCA), Gamma test (GT), and stepwise regression (SR), have been used to select the most effective variables. The artificial neural network (ANN) and multiple linear regression (MLR) are also common tools for SLY modeling. We conducted this study (1) to obtain the most effective variables influencing SLY through DRTs including PCA, GT, and SR, and then, to use them as input data for ANN and MLR; and (2) to provide the best SLY models. Accordingly, we used 14 physiographic, climatic, and hydrologic parameters from 42 watersheds in the Hyrcanian forest region (in northern Iran). The most effective variables as determined through DRTs as well as the original data sets were used as the input data for ANN and MLR in order to provide an SLY model. The results indicated that the SLY models provided by ANN performed much better than the MLR models, and the GT-ANN model was the best. The determination of coefficient, relative error, root mean square error, and bias were 99.9%, 26%, 323 t/year, and 6 t/year in the calibration period, and 70%, 43%, 456 t/year, and 407 t/year in the validation period, respectively. Overall, selecting the main factors that influence SLY and using artificial intelligence tools can be useful for water resources managers to quickly determine the behavior of SLY in ungauged watersheds.     

Another Look at Z-transform Technique for Deriving Unit Impulse Response Function

This paper presents a technique to derive the unit impulse response functions (UIRF) used for determination of unit hydrograph by employing the Z-transform technique to the response function derived from the Auto Regressive Moving Average (ARMA) process of order (p, q). The proposed approach was applied to reproduce direct surface runoff for single storm event data registered over four watersheds of area ranging from 0.42 to 295 km². It is observed that the UIRF based on ARMA (1, 2) and ARMA (2, 2) provides a better representation of the watershed response. Further, to test the superiority of the developed impulse response function form ARMA process, the direct runoff hydrographs were computed using the simple ARMA process and optimized Nash’s two parameter model and compared with the results obtained from UIRF’s of ARMA model. The performance of the models based on the graphical presentation as well as from the test statistics viz. RMSE and MAPE indicates that UIRF-ARMA (p, q) performs better than optimized Nash Model and mostly similar to simple ARMA (p,q) model. Further more, the ARMA process of order p ≤ 2 and q ≤ 2 is generally sufficient and less cumbersome than the Argand diagram based approach for UIRF derivation.     

ANN、ANFIS和AR模型在日径流时间序列预测中的应用比较

水文预测是水文学为经济和社会服务的重要方面。其预报结果不仅能为水库优化调度提供决策支持,而且对水电系统的经济运行、航运以及防洪等方面具有重大意义。自回归模型(AR模型)、人工神经网络(ANN)和自适应神经模糊推理系统(ANFIS)在日径流时间序列中应用广泛。将这三种模型应用于桐子林的日径流时间序列预测中,不仅采用纳什系数(NS系数)、均方根误差(RMSE)和平均相对误差(MARE)为评价指标,对三种模型的综合性能进行了比较。而且,在对三种模型预测结果的平均相对误差的阈值统计基础上,分析了三种模型的预测误差分布。同时,通过研究模型性能指标随预见期的变化过程评价了三种模型不同预见期下的预测能力。结果表明ANFIS相对于ANN和AR模型不仅具有更好的模拟能力、泛化能力,而且在相同的预见期下具有更优的模型性能,可以作为日径流时间序列预测的推荐模型。     

A modified NRCS-CN method for eliminating abrupt runoff changes induced by the categorical antecedent moisture conditions

The popular Natural Resources Conservation Service Curve Number (NRCS-CN) (earlier known as Soil Conservation Service Curve Number (SCS-CN) method of rainfall-runoff modeling has often faced the criticism of exhibiting quantum jumps in runoff computations because of the sudden jumps appearing in CN-values derived from NEH-4 tables for three antecedent moisture conditions (AMC), viz., AMC-I, AMC-II, and AMC-III valid for dry, normal, and wet conditions, respectively. The variability of antecedent soil moisture within an AMC category is responsible for the abrupt jump and other deficiencies in the CN method for runoff estimation. This paper suggests a novel procedure to account for the antecedent moisture (M), preventing quantum jumps and eliminating deficiencies in determination of CN and, in turn, estimation of direct runoff. Its validity was verified utilizing the observed rainfall (P)-runoff (Q) events from 36 US watersheds, four sub-catchments of the Godavari basin, and small agricultural plots at Roorkee, India. The performance of the proposed model (M5) for runoff prediction was compared with the existing NRCS-CN (M1), Mishra and Singh (2002) (M2), Singh et al. (2015) (M3), and Verma et al. (2021) (M4) model using various performance indices. Using the CNs derived from observed events, model M5 was seen to have performed better than M1-M4 in terms of Nash Sutcliffe Efficiency (NSE), Root Mean Square Error (RMSE), and Percent Bias (PBIAS) for the data of US watersheds, and CN-P correlation improved as the coefficient of determination (R²) enhanced. Similarly, using the RS & GIS-based CNs on natural watersheds of the Godavari basin and considering AMC-I, the performance of M5 was again better than M1-M4 in terms of RMSE, Mean Bias Error (mBIAS), Mean Absolute Error (MAE), and Normalized-Nash Sutcliffe Efficiency (NNSE). Interestingly, there existed a significant (p < 0.05) relationship between the in-situ water content (w) measured for the experimental plots of Roorkee and the model input variable antecedent moisture (M), offering a physical touch to the conceptual model.     

Threshold storm approach for locating phosphorus problem areas: An application in three agricultural watersheds in the Canadian Lake Erie basin

Hot-spots and hot-moments of phosphorus loads in an agricultural watershed depend not only on the watershed characteristics but also on the type and intensity of storms. Not all storms will generate phosphorus that can be considered problematic. A threshold storm is thus proposed and defined as the maximum storm intensity in which the phosphorus generated in a watershed is below seasonal phosphorus tolerance limit. To evaluate the threshold storm approach, separate Agricultural Non-point Source (AGNPS) models for three diverse small agricultural watersheds in southern Ontario, Canada were calibrated for runoff volume, sediment yield, and total phosphorus and run for representative storms with increasing return periods (2-year through 100-year). Results showed that in an upland watershed (Holtby), a 4.8-year early spring storm tend to generate phosphorus load above the threshold limit for the season. The same for low-land watersheds (Wigle and Jeannette) were, respectively, 14.9-year and 12.4-year. In all three watersheds, summer storms up to 100-year will fail to reach the seasonal tolerance limit for phosphorus. The critical source areas, identified based on the threshold storms, were distributed uniformly across the watersheds. As a phosphorus problem is essentially a source problem, such a simple yet robust approach to identify critical source areas of phosphorus can be useful in designing cost-effective best management practices.     

时空变源混合产流模型及其在小流域洪水模拟中的应用——以四川省、甘肃省的4个小流域为例

近年来山洪灾害造成的人员伤亡和财产损失不断增加,基于GIS和遥感数据的水文模型为山洪预报提供了新的思路。利用DEM和遥感数据提取小流域基础属性并划分地貌水文响应单元,利用垂向混合产流计算产流量,采用包气带非饱和土壤计算下渗,建立垂向上的时空变源混合产流模型。利用时空变源混合产流模型,模拟四川省和甘肃省4个流域多场次洪水过程,通过SCE算法进行参数率定和产流分析。结果表明:模型模拟与实测数据基本吻合,纳什效率系数均超过0.8;四川和甘肃两省小流域受到下垫面信息影响,主导产流成分不同。模型能够提供较为精确的山丘区流域洪水预报。     

基于SWAT模型不同产流及蒸发模式的径流模拟

针对四川省彭州市湔江流域,成功构建SWAT分布式水文模型,基于模型提供的两种计算地表径流方法和三种计算蒸散发方法的组合,采用2008—2009年月径流资料进行率定,通过Nash系数、相关系数、相对误差以及持续系数分析比较了6种组合模式下模型的适用性,并用2010—2011年月径流资料进行验证。结果表明,引入作为SWAT适用性评价标准可取得良好的效果;SCS径流曲线数法分别与Penman-Monteith法、Hargreaves法的组合模拟精度较好,Nash系数均达到0.8,相对误差小于20%,持续系数达到0.7以上。这为模型模拟方法的选择提供了依据和参考。     

A Modeling Method to Evaluate the Management Strategy of Urban Storm Runoff

Design of treatment strategy considering the behavior of pollutant with respect to runoff hydrograph is important for effective management of storm runoff. In this study, a mathematical model for storm runoff management was developed. The model determines the storm runoff management strategy and the associated probability under the constraint of regulatory discharge limit, and can quantify the volume or fraction of a storm runoff that should be treated. The model application was demonstrated for total suspended solids (TSS) and benzo(a)pyrene (BaP) in urban storm runoff. The samples were collected from 15 storm runoff events in six impervious sites in Beijing urban area, and. For TSS, the probability that the whole runoff can be discharged without treatment was 46 %, and the whole runoff should be intercepted for treatment was 31 %. While under the constraint of the discharge limit for BaP, the dominant strategy was that the whole runoff should be intercepted for treatment, with a probability of 88 %. Compared with the treatment strategies, it was noted that the need for runoff treatment was more for BaP than TSS. This was because the pollution level of TSS was lower than BaP from the aspect of compliance with discharge limit. Because the most seriously contaminated and toxic pollutant should be taken as the primary indicator for runoff treatment, the treatment option for the study area should follow the strategies for BaP. This methodology may be applied for other pollutants in different watersheds.     

Sensitivity Analysis of the GIUH based Clark Model for a Catchment

For estimation of runoff response of an ungauged catchment resulting from a rainfall event, geomorphologicalinstantaneous unit hydrograph (GIUH) approach is getting popularbecause of its direct application to an ungauged catchment. Itavoids adoption of tedious methods of regionalization of unithydrograph; wherein, the historical rainfall-runoff data of anumber of gauged catchments are required to be analyzed. In thisstudy, the GIUH derived from geomorphological characteristics ofa catchment has been related to the parameters of Clark IUH modelfor deriving its complete shape. The DSRO hydrographs estimatedby the GIUH based Clark model have been compared with the DSROhydrographs computed by the Clark IUH model option of the HEC-1package and the Nash IUH model by employing some of the commonlyused error functions. Sensitivity analysis of the GIUH basedClark model has been conducted with the objective to identify thegeomorphological and other model parameters which are moresensitive in estimation of peak of unit hydrographs computed bythe GIUH based Clark model. So that these parameters may beevaluated with more precision for accurate estimation of floodhydrographs for the ungauged catchments.     

基于分布式水文模型的“尼伯特”台风暴雨洪水反演

台风暴雨具有强度大、降雨集中的特点,容易在沿海地区中小流域形成局地破坏性洪水灾害,基于高精度地形地貌数据构建分布式水文模型,是实现中小流域洪水模拟和预报的重要手段。以福建闽清县梅溪流域为研究区,基于高精度DEM、土壤和土地利用数据,采用中国山洪水文模型(CNFF-HM),构建了梅溪流域分布式水文模型,以56场实测水文气象资料对模型进行率定和验证,并对2016年7月9日发生的"尼伯特"台风带来的暴雨洪水进行反演。结果显示,构建的分布式水文模型验证和率定效果良好,对"尼伯特"台风暴雨洪水模拟的洪峰流量误差小于20%,确定性系数达到0.96,因此模型能够很好地反映流域短历时强降雨引发的洪水特征,为流域防洪减灾提供可靠的技术支撑。     

森林植被变化对径流及洪水的影响分析

采用流域平行对比和自身对比两种方法，利用实测水文资料，从理论上分析了森林植被变化对小流域暴雨径流及洪水流量过程的影响。结果表明：森林覆盖率高的流域拦水能力较高，能显著地削减洪峰、延缓洪水过程，降低径流系数，使次暴雨径流量的变差系数c，增大。     

Modeling of Sediment Yield Prediction Using M5 Model Tree Algorithm and Wavelet Regression

The forecast of the sediment yield generated within a watershed is an important input in the water resources planning and management. The methods for the estimation of sediment yield based on the properties of flow and sediment have limitations attributed to the simplification of important parameters and boundary conditions. Under such circumstances, soft computing approaches have proven to be an efficient tool in modelling the sediment yield. The focus of present study is to deal with the development of decision tree based M5 Model Tree and wavelet regression models for modeling sediment yield in Nagwa watershed in India. A comparison is also performed with the artificial neural network (ANN) model for streamflow forecasting. The root mean square errors (RMSE), Nash-Sutcliff efficiency index (N-S Index), and correlation coefficient (R) statistics are used for the statistical criteria. A comparative evaluation of the performance of M5 Model Tree and wavelet regression versus ANN clearly shows that M5 Model Tree and wavelet regression can prove more useful than ANN models in estimation of sediment yield. Further, M5 model tree offers explicit expressions for use by design engineers.     

Potential implications of pre-storm soil moisture on hydrological prediction

Numerous hydrological models with various complexities, strengths, and weaknesses are available. Despite technological development, the association of runoff accuracy with the underlying model's parameters in watersheds with limited data remains elusive. Evaluating the soil moisture impacts at the watershed scale is often a difficult task, but it can be vital to optimally managing water resources. Incorporating pre-storm soil moisture accounting (PSMA) procedures into hydrologic models affects the watershed response to generate runoff from storm rainfall. This study demonstrated the impact of pre-storm and post-storm soil moisture in order to circumvent major obstacles in accurate runoff estimation from watersheds employing the conventional curve number (CN) model. The proposed hydrological lumped model was tested on a data set (1,804 rainfall-runoff events) from 39 watersheds in South Korea. Its superior performance indicates that the reconciliation of pre- and post-storm conceptualization has the potential to be a solution for efficient hydrological predictions and to demonstrate the complex and dynamic nature of tractable hydrological processes. The statistically significant results reveal that the proposed model can more effectively predict runoff from watersheds in the study area than the conventional CN model and its previously proposed modifications.