A rainfall‐runoff model parameterized from GIS and runoff data

A guiding principle in hydrological modelling should be to keep the number of calibration parameters to a minimum. A reduced number of parameters to be calibrated, while maintaining the accuracy and detail required by modern hydrological models, will reduce parameter and model structure uncertainty and improve model diagnostics. In this study, the dynamics of runoff are derived from the distribution of distances from points in the catchments to the nearest stream. This distribution is unique for each catchment and can be determined from a geographical information system. The distribution of distances, will, when a celerity of (subsurface) flow is introduced, provide a distribution of travel times, or a unit hydrograph (UH). For spatially varying levels of saturation deficit, we have different celerities and, hence, different UHs. Runoff is derived from the superposition of the different UHs. This study shows how celerities can be estimated if we assume that recession events represent the combined UHs for different levels of saturation deficit. A new soil moisture routine which estimates saturated and unsaturated volumes of subsurface water and with only one parameter to calibrate is included in the new model. The performance of the new model is compared with that of the Swedish HBV model and is found to perform equally well for eight Norwegian catchments although the number of parameters to be calibrated in the module concerning soil moisture and runoff dynamics is reduced from seven in the HBV model to one in the new model. It is also shown that the new model has a more realistic representation of the subsurface hydrology. Copyright © 2013 John Wiley & Sons, Ltd.     

Fuzzy computing based rainfall–runoff model for real time flood forecasting

This paper analyses the skills of fuzzy computing based rainfall–runoff model in real time flood forecasting. The potential of fuzzy computing has been demonstrated by developing a model for forecasting the river flow of Narmada basin in India. This work has demonstrated that fuzzy models can take advantage of their capability to simulate the unknown relationships between a set of relevant hydrological data such as rainfall and river flow. Many combinations of input variables were presented to the model with varying structures as a sensitivity study to verify the conclusions about the coherence between precipitation, upstream runoff and total watershed runoff. The most appropriate set of input variables was determined, and the study suggests that the river flow of Narmada behaves more like an autoregressive process. As the precipitation is weighted only a little by the model, the last time‐steps of measured runoff are dominating the forecast. Thus a forecast based on expected rainfall becomes very inaccurate. Although good results for one‐step‐ahead forecasts are received, the accuracy deteriorates as the lead time increases. Using the one‐step‐ahead forecast model recursively to predict flows at higher lead time, however, produces better results as opposed to different independent fuzzy models to forecast flows at various lead times. Copyright © 2004 John Wiley & Sons, Ltd.     

Enforced self‐organizing map neural networks for river flood forecasting

Self‐organizing maps (SOMs) have been successfully accepted widely in science and engineering problems; not only are their results unbiased, but they can also be visualized. In this study, we propose an enforced SOM (ESOM) coupled with a linear regression output layer for flood forecasting. The ESOM re‐executes a few extra training patterns, e.g. the peak flow, as recycling input data increases the mapping space of peak flow in the topological structure of SOM, and the weighted sum of the extended output layer of the network improves the accuracy of forecasting peak flow. We have investigated an ESOM neural network by using the flood data of the Da‐Chia River, Taiwan, and evaluated its performance based on the results obtained from a commonly used back‐propagation neural network. The results demonstrate that the ESOM neural network has great efficiency for clustering, especially for the peak flow, and super capability of modelling the flood forecast. The topology maps created from the ESOM are interesting and informative. Copyright © 2007 John Wiley & Sons, Ltd.     

Including spatial distribution in a data‐driven rainfall‐runoff model to improve reservoir inflow forecasting in Taiwan

Multi‐step ahead inflow forecasting has a critical role to play in reservoir operation and management in Taiwan during typhoons as statutory legislation requires a minimum of 3‐h warning to be issued before any reservoir releases are made. However, the complex spatial and temporal heterogeneity of typhoon rainfall, coupled with a remote and mountainous physiographic context, makes the development of real‐time rainfall‐runoff models that can accurately predict reservoir inflow several hours ahead of time challenging. Consequently, there is an urgent, operational requirement for models that can enhance reservoir inflow prediction at forecast horizons of more than 3 h. In this paper, we develop a novel semi‐distributed, data‐driven, rainfall‐runoff model for the Shihmen catchment, north Taiwan. A suite of Adaptive Network‐based Fuzzy Inference System solutions is created using various combinations of autoregressive, spatially lumped radar and point‐based rain gauge predictors. Different levels of spatially aggregated radar‐derived rainfall data are used to generate 4, 8 and 12 sub‐catchment input drivers. In general, the semi‐distributed radar rainfall models outperform their less complex counterparts in predictions of reservoir inflow at lead times greater than 3 h. Performance is found to be optimal when spatial aggregation is restricted to four sub‐catchments, with up to 30% improvements in the performance over lumped and point‐based models being evident at 5‐h lead times. The potential benefits of applying semi‐distributed, data‐driven models in reservoir inflow modelling specifically, and hydrological modelling more generally, are thus demonstrated. Copyright © 2012 John Wiley & Sons, Ltd.     

On the estimation of antecedent wetness conditions in rainfall–runoff modelling

The analysis of the physical processes involved in a conceptual model of soil water content balance is addressed with the objective of its application as a component of rainfall–runoff modelling. The model uses routinely measured meteorological variables (rainfall and air temperature) and incorporates a limited number of significant parameters. Its performance in estimating the soil moisture temporal pattern was tested through local measurements of volumetric water content carried out continuously on an experimental plot located in central Italy. The analysis was carried out for different periods in order to test both the representation of infiltration at the short time‐scale and drainage and evapotranspiration processes at the long time‐scale. A robust conceptual model was identified that incorporated the Green–Ampt approach for infiltration and a gravity‐driven approximation for drainage. A sensitivity analysis was performed for the selected model to assess the model robustness and to identify the more significant parameters involved in the principal processes that control the soil moisture temporal pattern. The usefulness of the selected model was tested for the estimation of the initial wetness conditions for rainfall–runoff modelling at the catchment scale. Specifically, the runoff characteristics (runoff depth and peak discharge) were found to be dependent on the pre‐event surface soil moisture. Both observed values and those estimated by the model gave good results. On the contrary, with the antecedent wetness conditions furnished by two versions of the antecedent precipitation index (API), large errors were obtained. Copyright © 2007 John Wiley & Sons, Ltd.     

Storm‐event rainfall–runoff modelling approach for ungauged sites in Taiwan

This work develops a top‐down modelling approach for storm‐event rainfall–runoff model calibration at unmeasured sites in Taiwan. Twenty‐six storm events occurring in seven sub‐catchments in the Kao‐Ping River provided the analytical data set. Regional formulas for three important features of a streamflow hydrograph, i.e. time to peak, peak flow, and total runoff volume, were developed via the characteristics of storm event and catchment using multivariate regression analysis. Validation of the regional formulas demonstrates that they reasonably predict the three features of a streamflow hydrograph at ungauged sites. All of the sub‐catchments in the study area were then adopted as ungauged areas, and the three streamflow hydrograph features were calculated by the regional formulas and substituted into the fuzzy multi‐objective function for rainfall–runoff model calibration. Calibration results show that the proposed approach can effectively simulate the streamflow hydrographs at the ungauged sites. The simulated hydrographs more closely resemble observed hydrographs than hydrographs synthesized using the Soil Conservation Service (SCS) dimensionless unit hydrograph method, a conventional method for hydrograph estimation at ungauged sites in Taiwan. Copyright © 2008 John Wiley & Sons, Ltd.     

Integrating hydrometeorological information for rainfall‐runoff modelling by artificial neural networks

The major purpose of this study is to effectively construct artificial neural networks‐based multistep ahead flood forecasting by using hydrometeorological and numerical weather prediction (NWP) information. To achieve this goal, we first compare three mean areal precipitation forecasts: radar/NWP multisource‐derived forecasts (P_r), NWP precipitation forecasts (P_n), and improved precipitation forecasts (P_m) by merging P_r and P_n. The analysis shows that the accuracy of P_m is higher than that of P_r and P_n. The analysis also indicates that the NWP precipitation forecasts do provide relative effectiveness to the merging procedure, particularly for forecast lead time of 4–6 h. In sum, the merged products performed well and captured the main tendency of rainfall pattern. Subsequently, a recurrent neural network (RNN)‐based multistep ahead flood forecasting techniques is produced by feeding in the merged precipitation. The evaluation of 1–6‐h flood forecasting schemes strongly shows that the proposed hydrological model provides accurate and stable flood forecasts in comparison with a conventional case, and significantly improves the peak flow forecasts and the time‐lag problem. An important finding is the hydrologic model responses which do not seem to be sensitive to precipitation predictions in lead times of 1–3 h, whereas the runoff forecasts are highly dependent on predicted precipitation information for longer lead times (4–6 h). Overall, the results demonstrate that accurate and consistent multistep ahead flood forecasting can be obtained by integrating predicted precipitation information into ANNs modelling. Copyright © 2009 John Wiley & Sons, Ltd.     

Assessment of rainfall‐runoff models based upon wavelet analysis

A basic hypothesis is proposed: given that wavelet‐based analysis has been used to interpret runoff time‐series, it may be extended to evaluation of rainfall‐runoff model results. Conventional objective functions make certain assumptions about the data series to which they are applied (e.g. uncorrelated error, homoscedasticity). The difficulty that objective functions have in distinguishing between different realizations of the same model, or different models of the same system, is that they may have contributed in part to the occurrence of model equifinality. Of particular concern is the fact that the error present in a rainfall‐runoff model may be time dependent, requiring some form of time localization in both identification of error and derivation of global objective functions. We explore the use of a complex Gaussian (order 2) wavelet to describe: (1) a measured hydrograph; (2) the same hydrograph with different simulated errors introduced; and (3) model predictions of the same hydrograph based upon a modified form of TOPMODEL. The analysis of results was based upon: (a) differences in wavelet power (the wavelet power error) between the measured hydrograph and both the simulated error and modelled hydrographs; and (b) the wavelet phase. Power difference and wavelet phase were used to develop two objective functions, RMSE(power) and RMS(phase), which were shown to distinguish between simulated errors and model predictions with similar values of the commonly adopted Nash‐Sutcliffe efficiency index. These objective functions suffer because they do not retain time, frequency or time‐frequency localization. Consideration of wavelet power spectra and time‐ and frequency‐integrated power spectra shows that the impacts of different types of simulated error can be seen through retention of some localization, especially in relation to when and the scale over which error was manifest. Theoretical objections to the use of wavelet analysis for this type of application are noted, especially in relation to the dependence of findings upon the wavelet chosen. However, it is argued that the benefits of localization and the qualitatively low sensitivity of wavelet power and phase to wavelet choice are sufficient to warrant further exploration of wavelet‐based approaches to rainfall‐runoff model evaluation. Copyright © 2006 John Wiley & Sons, Ltd.     

Assessing rainfall–runoff models for the management of Lake Verbano

Growing human pressure and potential change in precipitation pattern induced by climate change require a more efficient and sustainable use of water resources. Hydrological models can provide a fundamental contribution to this purpose, especially as increasing availability of meteorological data and forecast allows for more accurate runoff predictions. In this article, two models are presented for describing the flow formation process in a sub‐alpine catchment: a distributed parameter, physically based model, and a lumped parameter, empirical model. The scope is to compare the two modelling approaches and to assess the impact of hydrometeorological information, either observations or forecast, on water resources management. This is carried out by simulating the real‐time management of the regulated lake that drains the catchment, using the inflow predictions provided by the two models. Copyright © 2010 John Wiley & Sons, Ltd.     

Rainfall‐runoff analysis based on competing linear impulse responses: decomposition of rainfall‐runoff processes

Many recent studies have successfully used neural networks for non‐linear rainfall‐runoff modelling. Due to fundamental limitation of linear structures, approaches employing linear models have been generally considered inferior to the neural network approaches in this area. However, the authors believe that with an appropriate extension, the concept of linear impulse responses can be a viable tool since it enables one to understand underlying dynamics of rainfall‐runoff processes. In this paper, the use of competing impulse responses for rainfall‐runoff analysis is proposed. The proposed method is based on the switch over of competing linear impulse‐responses, each of which satisfies the constraints of non‐negativity and uni‐modality. The computational analyses performed for the rainfall‐runoff data in the Seolma‐Chun experimental basin, Korea showed that the proposed method can yield promising results. Considering the basin characteristics as well as the results from this study, it may be concluded that three impulse responses are enough for rainfall‐runoff analysis. Copyright © 2007 John Wiley & Sons, Ltd.     

Effects of conditional parameterization on performance of rainfall‐runoff model regarding hydrologic non‐stationarity

Better parameterization of a hydrological model can lead to improved streamflow prediction. This is particularly important for seasonal streamflow forecasting with the use of hydrological modelling. Considering the possible effects of hydrologic non‐stationarity, this paper examined ten parameterization schemes at 12 catchments located in three different climatic zones in east Australia. These schemes are grouped into four categories according to the period when the data are used for model calibration, i.e. calibration using data: (1) from a fixed period in the historical records; (2) from different lengths of historical records prior to prediction year; (3) from different climatic analogue years in the past; and (4) data from the individual months. Parameterization schemes were evaluated according to model efficiency in both the calibration and verification period. The results show that the calibration skill changes with the different historic periods when data are used at all catchments. Comparison of model performance between the calibration schemes indicates that it is worth calibrating the model with the use of data from each individual month for the purpose of seasonal streamflow forecasting. For the catchments in the winter‐dominant rainfall region of south‐east Australia, a more significant shift in rainfall‐runoff relationships at different periods was found. For those catchments, model calibration with the use of 20 years of data prior to the prediction year leads to a more consistent performance. Copyright © 2011 John Wiley & Sons, Ltd.     

Green‐Ampt Curve‐Number mixed procedure as an empirical tool for rainfall–runoff modelling in small and ungauged basins

The Soil Conservation Service Curve Number (SCS‐CN) method is a popular rainfall–runoff model that is widely used to estimate direct runoff from small and ungauged basins. The SCS‐CN is a simple and valuable approach to quantify the total streamflow volume generated by storm rainfall, but its use is not appropriate for estimating the sub‐daily incremental rainfall excess. To overcome this drawback, we propose to include the Green‐Ampt (GA) infiltration model into a mixed procedure, which is referred to as Curve Number for Green‐Ampt (CN4GA), aiming to distribute in time the information provided by the SCS‐CN method. For a given storm, the computed SCS‐CN total net rainfall amount is employed to calibrate the soil hydraulic conductivity parameter of the GA model. The proposed procedure is evaluated by analysing 100 rainfall–runoff events that were observed in four small catchments of varying size. CN4GA appears to provide encouraging results for predicting the net rainfall peak and duration values and has shown, at least for the test cases considered in this study, better agreement with the observed hydrographs than the classic SCS‐CN method. Copyright © 2012 John Wiley & Sons, Ltd.     

Using recurrent neural networks to reconstruct rainfall‐runoff processes

Many novel techniques for reconstructing rainfall‐runoff processes require hydrometeorologic and geomorphologic information for modelling. However, certain information is not always measurable. In this paper, we employ a special recurrent neural network to reconstruct the rainfall‐runoff process by using collected rainfall data. In addition, we propose an indirect system identification to overcome the drawback of a traditional, time‐consuming trial‐and‐error search. The indirect system identification is an efficient method to recognize the structure of a recurrent neural network. The unit hydrograph can be derived directly from the weights of the network due to a state‐space form embedded in the recurrent neural network. This improves the link between the weights of the network and the physical concepts that most neural networks fail to connect. The case studies of 41 events from 1966 to 1997 have been implemented in Taiwan's Wu‐Tu watershed, where the runoff path‐lines are short and steep. Two recurrent neural networks and one state‐space model are utilized to simulate the rainfall‐runoff processes for comparison. The results are validated by four criteria: coefficient of efficiency; peak discharge error; time to peak arrival error; total discharge volume error. The resulting data from the recurrent neural network reveal that the neural network proposed herein is appropriate for hydrological systems. Copyright © 2005 John Wiley & Sons, Ltd.     

Conceptual rainfall–runoff model with a two‐parameter,infinite characteristic time transfer function

A two‐parameter transfer function with an infinite characteristic time is proposed for conceptual rainfall–runoff models. The large time behaviour of the unit response is an inverse power function of time. The infinite characteristic time allows long‐term memory effects to be accounted for. Such effects are observed in mountainous and karst catchments. The governing equation of the model is a fractional differential equation in the limit of long times. Although linear, the proposed transfer function yields discharge signals that can usually be obtained only using non‐linear models. The model is applied successfully to two catchments, the Dud Koshi mountainous catchment in the Himalayas and the Durzon karst catchment in France. It compares favourably to the linear, non‐linear single reservoir models and to the GR4J model. With a single reservoir and a single transfer function, the model is capable of reproducing hysteretic behaviours identified as typical of long‐term memory effects. Computational efficiency is enhanced by approximating the infinite characteristic time transfer function with a sum of simpler, exponential transfer functions. This amounts to partitioning the reservoir into several linear sub‐reservoirs, the output discharges of which are easy to compute. An efficient partitioning strategy is presented to facilitate the practical implementation of the model. Copyright © 2015 John Wiley & Sons, Ltd.     

A semi‐distributed parallel‐type linear reservoir rainfall‐runoff model and its application in Taiwan

The main purpose of this paper is to introduce a semi‐distributed parallel surface rainfall‐runoff conceptual model. In this paper, a general solution of the instantaneous unit hydrograph (IUH) has been derived successfully for N linearly connected reservoirs, each having a different storage constant. The solution is a function of geomorphologic parameters, meteorologic factors and roughness coefficients. The model also takes into account the hydrologic response which is influenced by outflow downstream of a reservoir. For calibration, the shuffled complex evolution (SCE) algorithm is used to search for the global optimal parameters of the model. Because of the parallel structure, the mean roughness parameter of the channel becomes a “conceptual” parameter without a real physical meaning. To evaluate the adaptability of the model adopted, three watersheds around the city of Taipei in Taiwan were chosen to test the effectiveness of the model. The study provides an appropriate rainfall‐runoff model for planning flood mitigation in Taiwan. Copyright © 1999 John Wiley & Sons, Ltd.     

Long‐term runoff forecasting by combining hydrological models and meteorological records

Reservoir operation is generally based on the inflows of the upstream catchment of the reservoir. If the arriving inflows can be forecasted, that can benefit reservoir operation and management. This study attempts to construct a long‐term inflow‐forecasting model by combining a continuous rainfall–runoff model with the long‐term weather outlook from the Central Weather Bureau of Taiwan. The analytical results demonstrate that the continuous rainfall–runoff model has good inflow simulation performance by using 10‐day meteorological and inflow records over a 33‐year period for model calibration and verification. The long‐term inflow forecasting during the dry season was further conducted by combining the continuous rainfall–runoff model and the long‐term weather outlook, which was found to have good performance. Copyright © 2005 John Wiley & Sons, Ltd.     

Uncertainties in real‐time flood forecasting with neural networks

Although artificial neural networks (ANNs) have been applied in rainfall runoff modelling for many years, there are still many important issues unsolved that have prevented this powerful non‐linear tool from wide applications in operational flood forecasting activities. This paper describes three ANN configurations and it is found that a dedicated ANN for each lead‐time step has the best performance and a multiple output form has the worst result. The most popular form with multiple inputs and single output has the average performance. In comparison with a linear transfer function (TF) model, it is found that ANN models are uncompetitive against the TF model in short‐range predictions and should not be used in operational flood forecasting owing to their complicated calibration process. For longer range predictions, ANN models have an improved chance to perform better than the TF model; however, this is highly dependent on the training data arrangement and there are undesirable uncertainties involved, as demonstrated by bootstrap analysis in the study. To tackle the uncertainty issue, two novel approaches are proposed: distance analysis and response analysis. Instead of discarding the training data after the model's calibration, the data should be retained as an integral part of the model during its prediction stage and the uncertainty for each prediction could be judged in real time by measuring the distances against the training data. The response analysis is based on an extension of the traditional unit hydrograph concept and has a very useful potential to reveal the hydrological characteristics of ANN models, hence improving user confidence in using them in real time. Copyright © 2006 John Wiley & Sons, Ltd.     

Comparing theoretically supported rainfall‐runoff erosivity factors at the Sparacia (South Italy) experimental site

Interpreting rainfall‐runoff erosivity by a process‐oriented scheme allows to conjugate the physical approach to soil loss estimate with the empirical one. Including the effect of runoff in the model permits to distinguish between detachment and transport in the soil erosion process. In this paper, at first, a general definition of the rainfall‐runoff erosivity factor REF_e including the power of both event runoff coefficient Q_R and event rainfall erosivity index EI₃₀ of the Universal Soil Loss Equation (USLE) is proposed. The REF_e factor is applicable to all USLE‐based models (USLE, Modified USLE [USLE‐M] and Modified USLE‐M [USLE‐MM]) and it allows to distinguish between purely empirical models (e.g., Modified USLE‐M [USLE‐MM]) and those supported by applying theoretical dimensional analysis and self‐similarity to Wischmeier and Smith scheme. This last model category includes USLE, USLE‐M, and a new model, named USLE‐M based (USLE‐MB), that uses a rainfall‐runoff erosivity factor in which a power of runoff coefficient multiplies EI₃₀. Using the database of Sparacia experimental site, the USLE‐MB is parameterized and a comparison with soil loss data is carried out. The developed analysis shows that USLE‐MB (characterized by a Nash–Sutcliffe Efficiency Index NSEI equal to 0.73 and a root mean square error RMSE = 11.7 Mg ha^?1) has very similar soil loss estimate performances as compared with the USLE‐M (NSEI = 0.72 and RMSE = 12.0 Mg ha^?1). However, the USLE‐MB yields a maximum discrepancy factor between predicted and measured soil loss values (176) that is much lower than that of USLE‐M (291). In conclusion, the USLE‐MB should be preferred in the context of theoretically supported USLE type models.     

Application of wavelet‐based multi‐model Kalman filters to real‐time flood forecasting

This paper presents the application of a multimodel method using a wavelet‐based Kalman filter (WKF) bank to simultaneously estimate decomposed state variables and unknown parameters for real‐time flood forecasting. Applying the Haar wavelet transform alters the state vector and input vector of the state space. In this way, an overall detail plus approximation describes each new state vector and input vector, which allows the WKF to simultaneously estimate and decompose state variables. The wavelet‐based multimodel Kalman filter (WMKF) is a multimodel Kalman filter (MKF), in which the Kalman filter has been substituted for a WKF. The WMKF then obtains M estimated state vectors. Next, the M state‐estimates, each of which is weighted by its possibility that is also determined on‐line, are combined to form an optimal estimate. Validations conducted for the Wu‐Tu watershed, a small watershed in Taiwan, have demonstrated that the method is effective because of the decomposition of wavelet transform, the adaptation of the time‐varying Kalman filter and the characteristics of the multimodel method. Validation results also reveal that the resulting method enhances the accuracy of the runoff prediction of the rainfall–runoff process in the Wu‐Tu watershed. Copyright © 2004 John Wiley & Sons, Ltd.