Figure 4 - uploaded by Shu-Kuang Ning
Content may be subject to copyright.
Source publication
Remote sensing, such as from satellite, has been recognized as useful for monitoring the changes in hydrology. In this study, we propose a way that is able to estimate flooding probability based on satellite data from the observation network of the World Meteorological Organization. Through a two-stage probability analysis, we can depict the area w...
Similar publications
The hydraulic capacity of a channel is impacted due to sediment deposition in urban drainage and sewer system. As a result, the self-cleansing mechanism is a widely used phenomena in urban drainage and sewer systems. In this context, the prediction of particle Froude number plays an important role in the design of the sewer system. This study inves...
Citations
... CART operates by a tree building mechanism, where the final output is an interpretable set of decision rules [43]. Since its inception in the 1980s, CART has been productively applied in numerous urban flood studies [35,36,147,187]. Indeed, Bouramtane et al had demonstrated that the prediction ability of CART surpassed support vector, logistic regression, and discriminant analysis models [23]. ...
... The most often ANN where applied for the rainfall forecast [e.g. Chen et al. (2008), Hung et al. (2008) and Schellart et al. (2011)]. In Duncan et al. (2011) ANN where used for simulating a sewer system. ...
Urban flooding is often characterised by short lead times. In combination with the uncertainty in precipitation forecasting, the real-time forecasting of urban flooding is still challenging. Fast physically based models are still too slow for the usage in real-time forecasting. Data driven models are suitable to face this problem. The present study deals with testing an artificial neural network based model for the prediction of water levels with two dimensional spatial distributions at the catchment surface. The model was tested for synthetic rain events in a prior study. In the present study the model is successfully tested for spatially uniform distributed natural rain events.
... For the prediction of rainfall [e.g. Chen et al. (2008), Hung et al. (2008 and Schellart et al. (2011)] and in Duncan et al. (2011) for the computation of sewer systems. The calibration of physically-based sewer system model parameters with ANN is shown in Bruen and Yang (2006). ...
... Over the past two decades, several international scholars have examined the utility of artificial neural networks in supporting urban waterlogging prevention and control. Artificial neural network technologies have been used in several studies on urban flooding, regional flooding, and urban waterlogging prevention (Varoonchotikul 2003;Chen et al. 2008;Savic et al. 2013;Liu et al. 2005;Zhao 2009;Fu and Zhang 2013). Hung et al. (2009) used artificial neural networks to identify several factors that affect rainfall patterns and thereby established a model that generates accurate rain forecasts and predicts trends several hours into the future. ...
Artificial neural network technologies are frequently used in flood disaster simulations to aid regional disaster analyses. However, despite being an important factor that affects urban waterlogging, urban underground pipeline knowledge is seldom coupled with artificial neural networks or applied to urban waterlogging simulations. This article presents a simulation of urban waterlogging that utilises professional knowledge of urban underground drain pipelines coupled with BP artificial neural networks. Using this method, actual input weights are computed to simulate the river stage variations in the Panlong River of Kunming, China, for 35 consecutive hours during a heavy rainstorm that took place on 19 July 2013. The artificial neural network is coupled with drain pipeline knowledge, and river stage variations during this heavy rainfall are successfully simulated. The study results indicate that, in comparison with traditional BP neural network simulation methods, the use of knowledge of urban drain pipelines coupled with artificial neural networks yields more precise forecasting results for the urban river stage, with 85.7 % of all simulated river stage values corresponding closely with observed values. To support decision-making based on urban waterlogging forecasts, a map showing the impact distribution of the maximum river stage of Panlong River on the day of field study is provided. The results of the simulations show that the predicted locations of river water overflow were similar to the observed locations.
... Application of ANN in hydrologic modeling was inspired by the work on forecasting mapping (predictors) for chaotic dynamic systems (Farmer and Sidorowich 1987). Since early 90s, ANNs have been applied to many hydrologic problems such as rainfall-runoff modeling, streamflow forecasting, groundwater modeling, water quality, rainfall forecasting, hydrologic time series modeling, and reservoir operations (e.g., Govindaraju 2000; Luk et al. 2001;Dawson and Wilby 2001;Zhang and Govindaraju 2003;Abrahart et al. 2004;Abrahart et al. 2007;Chen et al. 2008;Chokmani et al. 2008;Wu et al. 2008;Turan and Yurdusev 2009;Besaw et al. 2010;Tiwari and Chatterjee 2010;Gao et al. 2010;Hou et al. 2012). ...
This paper focuses on the development and testing of the genetic algorithm (GA)-based regional flood frequency analysis (RFFA) models for eastern parts of Australia. The GA-based techniques do not impose a fixed model structure on the data and can better deal with nonlinearity of the input and output relationship. These nonlinear techniques have been applied successfully in many hydrologic problems; however, there have been only limited applications of these techniques in RFFA problems, particularly in Australia. A data set comprising of 452 stations is used to test the GA for artificial neural networks (ANN) optimization known as GAANN. The results from GAANN were compared with the results from back-propagation for ANN optimization known as BPANN. An independent testing shows that both the GAANN and BPANN methods are quite successful in RFFA and can be used as alternative methods to check the validity of the traditional linear models such as quantile regression technique.
... (Venkatesan et al. 1997;Silverman & Dracup 2000;Toth et al. 2000;Pongracz et al. 2001;Sivapragasam et al. 2001;Brath et al. 2002;Bray & Han 2004;Hettiarachchi et al. 2005;Han et al. 2007;Chattopadhyay & Chattopadhyay 2007Chen et al. 2008;Guhathakurta 2008). For example, Venkatesan et al. (1997) employed the ANN to predict the all-India summer monsoon rainfall with different meteorological parameters as model inputs. ...
A hybrid model integrating artificial neural networks and support vector regression was developed for daily rainfall prediction. In the modeling process, singular spectrum analysis was first adopted to decompose the raw rainfall data. Fuzzy C-means clustering was then used to split the training set into three crisp subsets which may be associated with low-, medium- and high-intensity rainfall. Two local artificial neural network models were involved in training and predicting low- and medium-intensity subsets whereas a local support vector regression model was applied to the high-intensity subset. A conventional artificial neural network model was selected as the benchmark. The artificial neural network with the singular spectrum analysis was developed for the purpose of examining the singular spectrum analysis technique. The models were applied to two daily rainfall series from China at 1-day-, 2-day- and 3-day-ahead forecasting horizons. Results showed that the hybrid support vector regression model performed the best. The singular spectrum analysis model also exhibited considerable accuracy in rainfall forecasting. Also, two methods to filter reconstructed components of singular spectrum analysis, supervised and unsupervised approaches, were compared. The unsupervised method appeared more effective where nonlinear dependence between model inputs and output can be considered.
Key words | artificial neural network, daily rainfall prediction, fuzzy C-means, hybrid models, singular spectral analysis, support vector regression
Prediction of air-water two-phase flow frictional pressure loss in pressurized tunnels and pipelines is essentially in the design of proper hydraulic structures and pump systems. In the present study artificial neural networks (ANN) and adaptive neuro-fuzzy inference system (ANFIS) are employed to predict pressure loss in air-water two-phase slug flow. Adaptive particle swarm optimization (APSO) is also applied to optimize the results of the ANN and ANFIS models. To predict the pressure loss in two-phase flow, the frictional pressure loss coefficient needs to be determined with respect to the effective dimensionless parameters including two-phase flow Froude and Weber numbers and the air concentration. Laboratory test results are used to determine and validate the findings of this study. The performances of the ANN-APSO and ANFIS-APSO models are compared with those of the ANN and ANFIS models. Different comparison criteria are used to evaluate the performances of developed models, suggesting that all the models successfully determine the air-water two-phase slug flow pressure loss coefficient. However, the ANFIS-APSO performs better than other models. Good agreement is obtained between estimated and measured values, indicating that the APSO with a conjugated ANFIS model successfully estimates the air-water two-phase slug flow pressure loss coefficient as a complex hydraulic problem. Results suggest that the proposed models are more accurate compared to former empirical correlations in the literature.
Rainfall-runoff models are used in the field of hydrology and runoff estimation for many
years, but despite existing numerous models, the regular release of new models shows that there is still not a model that can provide sophisticated estimations with high accuracy and performance. In order to achieve the best results, modeling and identification of factors affecting the output of the model is necessary. In this regard, in present study, it has been tried to identify the factors and estimating the amount of runoff using a variety of methods of artificial intelligence and multiple regression. Then, to evaluate the efficiency of the implemented models and choose the best model, some performance criteria including the correlation coefficient (R), Nash-Sutcliffe coefficient (NSE), the root mean square error (RMSE) and the mean absolute error (MAE) were used . The data used in this study were 9 rainfall events data measured in time period of 2011- 2015 taken from the Khakh watershed of Gonabad. Artificial intelligence models used in this study were: normal feedforward neural networks, feedforward Cascade neural networks, feedbackward Elman neural networks, Adaptive Neuro Fuzzy Inference System (ANFIS) and regression decision tree model (Regerssion Tree) that were implemented in MATLAB software environment and also step multiple regression as statistical methods which was implemented in Minitab software. The results of this study showed that the used statistical and artificial intelligence methods are considered acceptable with almost similar performance and with relatively appropriate accuracy and low error they are able to estimate the amount of runoff. In the meantime, Cascade and normal feedforward neural models with 5 input parameters, presented better performance comparing to the other models, as the performance criteria of R, RMSE, NSE and MAE in these models were the similar values of 0.88 , 0.76, 2 and 1.5, respectively. Overall, the findings indicate better estimations of the artificial intelligence models comparing to the regression model.
Background and Objectives: Discharge estimation in watersheds with limited statistical data is
of interest of researchers especially in developing countries. In many cases, recorded discharges
are not available or insufficient in terms of quality and quantity that lead to problems for water
resources management plans. Therefore, methods to predict the river discharges in ungauged or
with limited recorded data have considerable importance. Various methods including statistical
models, time series and expert models have been developed for discharge estimation. The decision
tree model is one of expert models that model the nonlinear behavior of data using simple rules
production. The objective of this study is evaluation of multivariate regression and decision tree
model (M5) to estimate monthly discharge in ungauged watersheds in Golestan Province.
Materials and Methods: In this study, the Golestan province that includes various watersheds
with different characteristics was considered. The physiographic characteristics of studied
watershed were extracted and rainfall and temperature climatic parameters were estimated in
monthly time scale in GIS environment for the period 1981-2011. The climatic and
physiographic parameters were considered as input to multivariate regression and decision trees
M5 models and root mean square error (RMSE) and correlation coefficient (R) applied for
models evaluation.
Results: According to the results of multiple regression and decision tree models, discharge
estimations in wet months were more accurate than dry months. In application of decision tree
model the best prediction belonging to March with R=0.93 and RMSE=1.002 while worst
prediction was for August with R=0.571 and RMSE=0.635. Moreover, multivariate regression
model led to best results in March with R=0.723 and RMSE=2.043 and low accurate prediction
in August with R=0.322 and RMSE=1.979. The results of decision tree model were better than
multivariate regression model in all months based in calculations.
Conclusion: The results of this study showed that the discharge estimation using multivariate
regression and decision tree M5 models in wet months are applicable but in dry months'
predictions are not accurate because of high rainfall variations, storm patterns and error in
rainfall zoning and interpolation. The results indicated that the decision tree model had more
accurate results than multivariate regression model considering higher precision and lower error.
The decision tree model had higher correlation coefficient with respect to model evaluation
criteria.
