A problem commonly encountered in the stormwater design project is the determination of the design flood. Design flood estimation **using** established methodology is relatively simple when records of streamflow or **runoff** and **rainfall** are available for the catchment concerned. The quantity of **runoff** resulting from a given **rainfall** event depends on a number of factors such as initial moisture, land use, and slope of the catchments, as well as intensity, distribution, and duration of the **rainfall**. Knowledge on the characteristics of **rainfall**-**runoff** relationship is essential for risk and reliability analysis of water resources projects. Since the 1930s, numerous **rainfall**-**runoff** models have been developed to forecast streamflow. For example, conceptual models provide daily, monthly, or seasonal estimates of streamflow for long term forecasting on a continuous basis. Sherman (1932) defined the unit graph, linear systems analysis has played an important role in relating input-output components in **rainfall**-**runoff** modelling and in the development of stochastic models of single hydrological sequences (Singh, 1982). The performance of a **rainfall**-**runoff** model heavily depends on choosing suitable model parameters, which are normally calibrated by **using** an objective function (Yu and Yang, 2000). The entire physical process in the hydrologic cycle is mathematically formulated in conceptual models that are composed of a large number of parameters (Tokar and Johnson , 1999).

Show more
35 Read more

NFS learning is usually classified as either offline or online learning. Online learning allows models to dynamically reiterate parameters based on new training data point that are presented to the model while offline learning or batch learning optimizes parameters based on a static dataset. Adaptive **Network**-based Fuzzy Inference System (ANFIS) [1] is a popular model used in R-R **modeling**; however, ANFIS employs offline learning which suffers from increased computational time and requires retraining to capture recent changes of the system. Moreover, the number of rules in ANFIS is predefined by the user. Dynamic Evolving **Neural**-Fuzzy Systems (DENFIS) [2] is a TSK type model which employs online learning capabilities that allows dynamic reiteration of the model parameters. This allows DENFIS to dynamically evolve its rule base to capture changes within the system through continuous updating of the model. To date, there are only few studies on NFS with online learning in R-R **modeling**. This study aims to investigate the application NFS with online learning for event-based **rainfall**-**runoff** **modeling** for a rural tropical catchment of size 23.22km 2 . The results obtained would be compared against an autoregressive model with

Show more
Over the last decade, there has been a tremendous growth in the interest of application of a class of techniques that op- erate in a manner analogous to that of biological neurons sys- tem, i.e. **artificial** **neural** networks (ANNs). While ANNs are capable of capturing non-linearity in the **rainfall**-**runoff** pro- cess compared with other **modeling** approaches (Hsu et al., 1995), ANN models have been applied in hydrology and in the context of **rainfall**-**runoff** **modeling** (Smith and Eli, 1995; Dawson and Wilby, 1998; Tokar and Markus, 2000; Zhang and Govindaraju, 2003; Kumar et al., 2005). From these studies, it has been demonstrated that ANN models can be flexible enough to simulate the **rainfall**-**runoff** processes suc- cessfully.

Show more
11 Read more

Reliable **modeling** for the **rainfall**-**runoff** processes embedded with high complexity and non-linearity can overcome the problems associated with managing a watershed. Physically based **rainfall**-**runoff** models need many realistic physical components and parameters which are sometime missing and hard to be estimated. In last decades the **artificial** intelligence (AI) has gained much popularity for calibrating the nonlinear relationships of **rainfall**–**runoff** processes. The AI models have the ability to provide direct relationship of the input to the desired output without considering any internal processes. This study presents an application of Multilayer Perceptron **neural** **network** (MLPNN) for the continuous and event based **rainfall**-**runoff** **modeling** to evaluate its performance for a tropical catchment of Lui River in Malaysia. Five years (1999-2013) daily and hourly **rainfall** and **runoff** data was used in this study. **Rainfall**-**runoff** processes were also simulated with a traditionally used statistical **modeling** technique known as auto- regressive moving average with exogenous inputs (ARMAX). The study has found that MLPNN model can be used as reliable **rainfall**-**runoff** **modeling** tool in tropical catchments.

Show more
The **Artificial** **Neural** **Network** (ANN) approach has been successfully used in many hydrological studies especially the **rainfall**-**runoff** **modeling** **using** continuous data. The present study examines its applicability to model the event-based **rainfall**-**runoff** process. A case study has been done for Ajay river basin to develop event-based **rainfall**-**runoff** model for the basin to simulate the hourly **runoff** at Sarath gauging site. The results demonstrate that ANN models are able to provide a good representation of an event-based **rainfall**-**runoff** process. The two important parameters, when predicting a flood hydrograph, are the magnitude of the peak discharge and the time to peak discharge. The developed ANN mod- els have been able to predict this information with great accuracy. This shows that ANNs can be very efficient in mod- eling an event-based **rainfall**-**runoff** process for determining the peak discharge and time to the peak discharge very ac- curately. This is important in water resources design and management applications, where peak discharge and time to peak discharge are important input variables.

Show more
The relationship of **rainfall**-**runoff** is known to be highly non-linear and complex. The **rainfall**- **runoff** relationship is one of the most complex hydrologic phenomena to comprehend due to the tremendous spatial and temporal variability of watershed characteristics and precipitation patterns, and the number of variables involved in the **modeling** of the physical processes. Hydrologists are often confronted with problems of prediction and estimation of **runoff**, precipitation, contaminant concentrations, water stages, and so on. Although many watersheds have been gauged to provide continuous records of stream flow, hydrologists are often faced with situations where little or no information is available. In such instances, simulation models are often used to generate synthetic flows. The available **rainfall**-**runoff** models are HEC-HMS, MIKE-11, SWMM, etc. These models are useful for the hydrologic and hydraulic engineering planning and design as well as water resources management; e.g., hydropower generation, flood protection and irrigation. The existing popular model is considered as not flexible and they require many parameters. Obviously, the models have their own weaknesses. Therefore, in view of the importance of the relationship between **rainfall**-**runoff**, the present study was undertaken in order to develop **rainfall**-**runoff** models that can be used to provide reliable and accurate estimates of **runoff**.

Show more
64 Read more

Monsoon **rainfall** in Gujrat is dependent on the weather systems travelling through central India towards Gujrat and Rajasthan. In Gujrat, amount of **rainfall** in district and talukas are varies. In Gujrat, southern part like Surat, Valsad, Navsari, **rainfall** is fall in large amount. Kutch, central Gujrat, north Gujrat have received very less amount of **rainfall** so it is impact on national economy. Valsad is located in Gujrat and it receives highest **rainfall** 2357mm. [3]

Various types of methods have been used in **runoff** estimation including conceptual and statistical models [1]. However, none of these methods can be considered as a single superior model [2]. Owing to the complexity of the hydrological process, the accurate **runoff** is difficult to be predicted **using** the linear recurrence relations or physically based watershed. The linear recurrence relation model does not attempt to take into account the nonlinear dynamic of hydrological process. The physically based watershed model also ignores the stochastic behavior underlying any hydrosystem. Besides, despite the application of deterministic models include all physical and chemical processes, the successful employment is restricted by a need for catchment-specific data and simplifications involved in solving the governing equations

Show more
10 Read more

The Hydrological Predictions for the Environment (HYPE) is a hydrological model for small-scale and large-scale assessments of water resources and water quality developed by the Swedish Meteorological and Hydrological Institute in between 2005–2007. This is a recently developed semi-distributed, conceptual model which imitates multi-basins, covering broad variations in sediment types, land and shape. HYPE integrates landscape elements and hydrological spaces along with nutrient transport along the flow path. The model divides a river watershed into number of small watersheds and each small watershed is further divided into sub-category depending upon soil type and vegetative cover. For each sub-category, the model imitates snowmelt, **runoff**, soil erosion, drainage, groundwater escape from different soil layers, nutrient in soil and transport to rivers and lakes. Calculations are made on a daily time step in linked small watersheds. The model parameters are related with land use, soil type. Due to this linking of parameters, it is best suited for imitations in ungauged watersheds also. This model takes input of maximum of ten data files independent of size and domain and all the input and output files are in ASCII format. There are some different HYPE models which are developed for the individual countries like S-HYPE model (S for Sweden) was employed for the country of Sweden to imitate daily **rainfall** **runoff** and nutrient concentrations. Similarly, E- HYPE model (E for Europe) for the continent of Europe, BALT-HYPE model was used for the whole Baltic Sea basin. LPB-HYPE model was applied on the La Plata Basin, Niger-HYPE model for imitating on Niger River in Africa and Arctic-HYPE model for imitating hydrological variables for the entire Arctic region. Similarly, development of In-HYPE model for simulating hydrological variables for the Indian region is in progress. 3.4 WinSRM

Show more
The modelling work was carried out **using** five years period of daily data and ten years period of hourly data consisting the **rainfall** and **runoff** records from selected catchments in Peninsular of Malaysia. There are four catchments being selected for analysis and modelling. Those stations have sufficient length of records and fairly good quality of data. Those are Sungai Bekok (Johor, Malaysia), Sungai Ketil (Kedah, Malaysia), Sungai Klang (Kuala Lumpur, Malaysia), and Sungai Slim (Perak, Malaysia) catchments. Those sites were selected to demonstrate the development and application of ANN, multiple linear regression (MLR), XP-SWMM and HEC-HMS models. It is emphasized that the MLR model is only applied to model the daily **rainfall**-**runoff** for those catchments. The data required to carry out this study are catchment physical data, **rainfall** and river (at catchments outlet). The data of all these gauges is recorded and maintained by Department of Drainage and Irrigation (DID) Malaysia.

Show more
54 Read more

Prediction; rainfall and runoff; artificial neural network; modular model; singular.. 27 spectrum analysis 28 29 1.[r]

65 Read more

In the present study, multilayer perceptron (MLP) based **neural** **network**, which is one of the efficient **artificial** **neural** **network** (ANN) was applied for **modeling** daily **rainfall**-**runoff** in a Himalayan watershed called Bino watershed in Almora and Pauri Garhwal districts of Uttarakhand, India **using** the time series monsoon data of ten years (2000-2009) of **rainfall** and **runoff**. Gamma test (GT) technique was applied for selection of the best input combinations. Performance of model developed was evaluated qualitatively as well as quantitatively **using** indices viz. correlation coefficient (r), root mean square error (RMSE) and coefficient of efficiency (CE). The results of the study showed that model (MLP7) with 5 inputs and one hidden layer with 8 neurons was found to be the best followed by model (MLP 19) having 10 hidden neurons for first hidden layer and 11 for second hidden layer. The r, RMSE and CE values for MLP 7 during testing were determined to be 0.9207, 0.9644 (mm) and 0.7974, respectively. The result of study revealed that ANN can be successfully applied for **rainfall**-**runoff** **modeling** in the study area with good accuracy.

Show more
14 Read more

An ANN can be defined as ‘a data processing system consisting of a large number of simple, highly interconnected processing elements (**artificial** neurons) in an archi- tecture inspired by the structure of the cerebral cortex of the brain’ [1]. The ANN models have been used successfully to model complex non-linear input-output rela- tionships in an extremely interdisciplinary field. It behaves as a black-box model. The natural behaviour of hydrological processes is appropriate for the application of ANN **method**. The ANN **method** has been proven to be potentially useful tools in hydro- logical modelling such as for **rainfall**-**runoff** **modeling** processes [2, 3, 4, 5]; flow pre- diction [6, 7]; water quality predictions [8]; operation of reservoir system [9, 10]; and groundwater reclamation problems, [11]. In Malaysia, the application of **neural** net- work **method** is widely used in the field of mechanics, robotics, electrical, etc. In the hydrology field, it is still in nascent stages. There are only few of research have imple- mented the **neural** **network** approach in the hydrological study. For example, [10] has applied the **neural** **network** **method** to forecast of net inflows for reservoir operation.

Show more
12 Read more

Arti ﬁcial **Neural** **Network** has been inspired from billions of interconnected neurons in human brain based on a mathematical con ﬁguration. It has been proven that ANN could be a better efﬁcient alternative for traditional methods for **modeling** the nonlinear time series. **Neural** **network** must be able to greatly map the data set of numeric inputs to the set of numeric targets. The applied **network** is a two-layer Feed-forward with sigmoid hidden and linear output neurons. This **network** was trained with Levenberg –Marquardt backpropagation (LMBP) algorithm as the fastest **method** for training moderate sized Feed-forward **Neural** Networks (up to several hundred weights). This algorithm uses the approximate Hessian matrix in the weight update procedure as follows:

Show more
11 Read more

Indonesia is a tropical country with two seasons (wet and dry) which play the main role in water cycle process. Occurrence of rain continues into the flow of the discharge in the river with a huge energy potential that can be exploited for the life of the surrounding community. The occurrence and intensity of rain is random and difficult to predict in a certain period of time so that discharge is also difficult to be estimated although it is measured in the field in time of **rainfall** occurrence. The amount of **runoff** produced by the same depth of precipitation in a watershed will result a different magnitude with another watershed because it is influenced by land use in the watershed. This paper discusses the **modeling** of **rainfall**-**runoff** in the Watershed of Bolon in Simalungun district of North Sumatra Province **using** **Artificial** **Neural** **Network** (ANN) to determine the potential of the available discharge in the long term for the purpose of Micro Hydro Power (MHP). The software/program is developed with Scilab mathematical open source software (www.scilab.org) based on ANN algorithm. The data are record of monthly **rainfall** and discharge for 12 years (2001 to 2012). The models developed are 12 monthly neurons, 4 year neurons and series neuron (48 neurons) for input (**rainfall**) - output (**runoff**) neurons. The result shows that reliability the 12 monthly neurons is 99% (the best) followed by series neuron with 78% and 4 year neuron 77%. The chosen model (12 monthly neurons) then to be used for predicting the monthly discharge availability at Bah Bolon Site. Dependable discharges predicted with this software for year 2013 to 2016 consecutively are as follows: 0.678246 m 3 /s, 0. 655288 m 3 /s, 0.678475 m 3 /s and 0.678135 m 3 /s.

Show more
ANNs are highly simplified mathematical models and computing techniques inspired by biological **neural** net- works. It can be categorized as interconnected groups of simple neurons that function as a combined system for processing information and model complex relationships between inputs and outputs by finding patterns in data. The FF-ANN trained with the back propagation algo- rithm is perhaps the most popular **network** for hydrologic modelling [13,14]. This **network** topology which acts an adaptive system consists of simple **artificial** nodes (neu- rons) connected together by links to form a **network** of nodes usually organized in a number of layers hence the term **artificial** **neural** **network**. Weighted input from pre- vious layer is received and processed output is transmit- ted to following layer through links. Mostly ANNs have three or more layers: an input layer for presenting data to **network**, an output layer for producing an appropriate response and intermediate (hidden) layer for collecting feature detectors. Present study highlights on a model back propagation algorithm for training, and the number of hidden neurons is optimized by a trial and error proc- ess. The basic structure of the ANN model is shown in Figure 1.

Show more
In order to further refine the number of inputs for the final RBF NN model to be developed, sensitivity analysis is carried out **using** RBF NN-1. In this **method**, first the mean and standard deviation is computed for each input parameter. The input is varied from (mean - standard deviation) to (mean + standard deviation) and the corresponding output is computed **using** RBF NN-1. These computed outputs are then used to evaluate the sensitivity coefficient between each input and output parameters. In order to select the input parameters which affect the output parameters most the mean of the sensitivity matrix is computed, which is 0.1335. The input parameters, which have sensitivity coefficient greater than mean value, are considered for analysis otherwise dropped. In this analysis the important parameters discussed in Section II are accounted irrespective of the value of sensitivity coefficient. Thus, dropped input parameters are I7-I12, I26, I27, I29, I30, I32, I33, I38, I49 and I50. In the present work a parameter in considered for further analysis if it is affecting more than 30% of total outputs, which is 8 outputs from 24. The number of input and output parameters after sensitivity analysis reduces to 25 and 24, respectively.

Show more
The steps taken in the identification of a nonlinear model of a system are selection of input- output data suitable for Testing and Validation i.e. 80-20%, 70-30 % and 60-40%; selection of a model structure and estimation of its parameters; and validation of the identified models. **Method** Adapted for **Rainfall** prediction: **ARTIFICIAL** **NEURAL** **NETWORK** (ANN).

An ANN is composed of many non- linear and densely interconnected processing elements or neurons. In an ANN architecture, neurons are arranged in groups called layers. Each neuron in a layer operates in logical parallelism. Information is transmitted from one layer to another in serial operations (Hecht- Nielsen, 1991). A **network** can have one or several layers. The basic structure of a **network** usually consists of three layers- the input layer, where the data are introduced to the **network**, the hidden layer(s), where the data are processed, and the output layer, where the results for the given input are produced. The neurons in the hidden layer(s) are connected to the neurons of a neighbouring layer by weighing factors that can be adjusted during the model training process. The networks are organized according to training methods for specific applications. Fig. 1 illustrates a three layer **artificial** **neural** **network** The most distinctive characteristic of an ANN is its ability to learn from examples. Learning or training of an ANN model is a procedure by which ANN repeatedly processes a set of test data (input – output data pairs) , changing the values of its weights. In the training or learning process, the target output at each output node is compared with the **network** output, and the difference or error is minimized by adjusting the weights and biases through some training algorithm. In the present study, the training of ANNs was accomplished by Levenberg- Marquardt algorithm with back- propagation (LMBP). Back- propagation is the most commonly used supervised training algorithm in the multilayer feed forward networks.

Show more