Chandranath Chatterjee

Flood Forecasting Using ANN, Neuro-Fuzzy, and Neuro-GA Models

Flood forecasting at Jamtara gauging site of the Ajay River Basin in Jharkhand, India is carried out using an artificial neural network (ANN) model, an adaptive neuro-fuzzy interference system (ANFIS) model, and an adaptive neuro-GA integrated system (ANGIS) model. Relative performances of these models are also compared. Initially the ANN model is developed and is then integrated with fuzzy logic to develop an ANFIS model. Further, the ANN weights are optimized by genetic algorithm (GA) to develop an ANGIS model. For development of these models, 20 rainfall–runoff events are selected, of which 15 are used for model training and five are used for validation. Various performance measures are used to evaluate and compare the performances of different models. For the same input data set ANGIS model predicts flood events with maximum accuracy. ANFIS and ANN model perform similarly in some cases, but ANFIS model predicts better than the ANN model in most of the cases.

Development of Regional Flood Frequency Relationships Using L-moments for Middle Ganga Plains Subzone 1(f) of India

In this study, screening of the data has been carried out basedon the discordancy measure (Di) in terms of the L-moments. Homogeneity of the region has been tested using the L-moments based heterogeneity measure, H. For computing the heterogeneity measure H, 500 simulations were carried out using the four parameter Kappa distribution. Based on this test, it has been observed that the data of 8 out of 11 bridge sites constitute ahomogeneous region. Hence, the data of these 8 sites have been used in this study. Catchment areas of these 8 sites vary from 32.89 to 447.76 km2 and their mean annual peak floods varyfrom 24.29 to 555.21 m3 s-1. Comparative regional floodfrequency analysis studies have been carried out using the various L-moments based frequency distributions viz. Extreme value (EV1), General extreme value (GEV), Logistic (LOS), Generalized logistic (GLO), Normal (NOR), Generalized normal (GNO), Uniform (UNF), Pearson Type-III (PE3), Exponential (EXP),Generalized Pareto (GPA), Kappa (KAP), and five parameter Wakeby(WAK). Based on the L-moment ratio diagram and ∣ Zidist ∣–statistic criteria, GEV distribution has been identified as the robust distribution for the study area. For estimation of floods of various return periods for gauged catchments of the study area, regional flood frequency relationship has been developed using the L-moments based GEV distribution. Also, for estimation of floods of desiredreturn periods for ungauged catchments, regional flood frequencyrelationship has been developed by coupling the regional flood frequency relationship with the regional relationship between mean annual maximum peak flood and catchment area.

Effect of Utilization of Discrete Wavelet Components on Flood Forecasting Performance of Wavelet Based ANFIS Models

Wavelet based flood forecasting models are known to perform better than conventional models, yet the effect of the way wavelet components are combined to develop a model on the forecasting performance, is inadequately investigated. To demonstrate this, two types of wavelet- adaptive neuro-fuzzy inference system (WANFIS), i.e. WANFIS-split data model (WANFIS-SD) and WANFIS-modified time series model (WANFIS-MS) are developed to forecast river water levels with 1-day lead time. To develop these models, first the original level time series (OLTS) is decomposed into discrete wavelet components (DWCs) by discrete wavelet transform (DWT) upto three resolution levels. In WANFIS-SD, all wavelet components are used as inputs while WANFIS-MS ignores the noise wavelet components and utilizes only the effective wavelet components. The effectiveness of the developed models are evaluated through application to two Indian rivers, Kamla and Kosi, which vary significantly in their catchment area and flow patterns. The proposed models are found to forecast river water levels accurately. On comparison, the WANFIS-SD is found to perform better than WANFIS-MS for high flood levels.

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.

Spatial variability of climate change impacts on yield of rice and wheat in the Indian Ganga Basin

Indian Ganga Basin (IGB), one of the most densely populated areas in the world, is facing a significant threat to food grain production, besides increased yield gap between actual and potential production, due to climate change. We have analyzed the spatial variability of climate change impacts on rice and wheat yields at three different locations representing the upper, middle and lower IGB. The DSSAT model is used to simulate the effects of climate variability and climate change on rice and wheat yields by analyzing: (i) spatial crop yield response to current climate, and (ii) impact of a changing climate as projected by two regional climate models, REMO and HadRM3, based on SRES A1B emission scenarios for the period 2011-2040. Results for current climate demonstrate a significant gap between actual and potential yield for upper, middle and lower IGB stations. The analysis based on RCM projections shows that during 2011-2040, the largest reduction in rice and wheat yields will occur in the upper IGB (reduction of potential rice and wheat yield respectively by 43.2% and 20.9% by REMO, and 24.8% and 17.2% by HadRM3). In the lower IGB, however, contrasting results are obtained, with HadRM3 based projections showing an increase in the potential rice and wheat yields, whereas, REMO based projections show decreased potential yields. We discuss the influence of agro-climatic factors; variation in temperature, length of maturity period and leaf area index which are responsible for modeled spatial variability in crop yield response within the IGB.

Wavelet Bootstrap Multiple Linear Regression Based Hybrid Modeling for Daily River Discharge Forecasting

A new hybrid model, the wavelet–bootstrap–multiple linear regression (WBMLR) is proposed to explore the potential of wavelet analysis and bootstrap resampling techniques for daily discharge forecasting. The performance of the developed WBMLR model is also compared with five more models: multiple linear regression (MLR), artificial neural network (ANN), wavelet-based MLR (WMLR), wavelet-based ANN (WANN) and wavelet–bootstrap–ANN (WBANN) models. Seven years of discharge data from seven gauging stations in the middle reaches of Mahanadi river basin in India are applied in this study. Significant input vectors are decomposed into discrete wavelet components (DWCs) using discrete wavelet transformation (DWT) to generate wavelet sub time series that are used as inputs to the MLR and ANN models to develop the WMLR and WANN models, respectively. Effective wavelets are selected by considering several types of wavelets with different vanishing moments. WBMLR and WBANN models are developed as ensemble of different WMLR and WANN models, respectively, developed using different realizations of the training dataset generated using bootstrap resampling technique. The results show that the wavelet bootstrap hybrid models (i.e. WBMLR and WBANN) produce significantly better results than the traditional MLR and ANN models. Hybrid models based on MLR (WMLR, WBMLR) perform better than the ANN based hybrid models (WBANN, WANN). The WBMLR and WMLR models simulate the peak discharges better than the WBANN, WANN, MLR and ANN models, whereas the overall performance of WBMLR model is found to be more accurate and reliable than the remaining five models.

Comparison of multi-objective evolutionary neural network, adaptive neuro-fuzzy inference system and bootstrap-based neural network for flood forecasting

Accurate flood forecasting is of utmost importance in mitigating flood disasters. Flood causes severe public and economic loss especially in large river basins. In this study, multi-objective evolutionary neural network (MOENN) model is developed for accurate and reliable hourly water level forecasting at Naraj gauging site in Mahanadi river basin, India. The performance of the developed model is compared with adaptive neuro-fuzzy inference system (ANFIS) and bootstrap-based neural network (BNN) models. The performance of the models is compared in terms of Nash–Sutcliffe efficiency, root mean square error, mean absolute error and percentage deviation in peak (D). The performance of the models in forecasting floods is also evaluated using existing performance evaluation criterion of Central Water Commission, India as well as a multiple linear regression model. A partitioning analysis in conjunction with threshold statistics is carried out to evaluate the performance of the developed models in forecasting floods for low, medium and high water levels. It is found that the performance of MOENN and BNN models is more stable and consistent compared to ANFIS model. For longer lead times, the performance of MOENN model is found to be the best, with its performance in forecasting higher water levels being significantly better compared to ANFIS and BNN models. Overall, it is found that MOENN model has great potential to be applied in flood forecasting.

River-Flow Forecasting Using Higher-Order Neural Networks

In this paper, we propose a novel neural modeling methodology for forecasting daily river discharge that makes use of neural units with higher-order synaptic operations (NU-HSOs). For hydrologic forecasting, conventional rainfall-runoff models based on mechanistic approaches in the literature have shown limitations attributable to their overparameterization and complexity. With the use of neural units with quadratic synaptic operation (NU-QSO) and cubic synaptic operation (NU-CSO), as suggested in this paper, the refined neural modeling methodology can overcome the intricacy and inefficiency of conventional models. In this paper, neural network (NN) models with NU-HSO are compared with conventional NNs with neural units with linear synaptic operation (NU-LSO) for forecasting river discharge. This study was conducted using 1- to 5-day lead time forecasting in the Mahanadi River basin at the Naraj gauging site to evaluate the effectiveness of the higher-order neural networks (HO-NNs). Performance indices for the prediction of daily discharge forecasting indicated that NNs with NU-CSO and NNs with NU-QSO achieved better performance than NNs with NU-LSO even with a lower number of hidden neurons. Thus, this study shows that HO-NNs can be effective in hydrologic forecasting. DOI: 10.1061/(ASCE)HE.1943-5584.0000486.

Erosion study of a part of Majuli River-Island using remote sensing data

Majuli, the world’s largest river island, is situated in mid of river Brahmaputra in Assam. River Brahmaputra flows in highly braided channels most of them are transient in nature, being submerged during high monsoon flows and changing drastically their geometry and location. Majuli island, home of about 1.3 million people is endangered because of the erratic behavior of the river. In this study, an attempt has been made to observe the trends of erosion in a small part of Majuli island, the area near Kaniajan village in south Majuli- a stretch of about 11 km, using satellite data of 1991, 1997 and 1998. Image processing of digital data has been done in ILWIS software. Supervised for delineation of river from land and then change detection analysis has been done to find out changes in river course from 1991 to 1997 and further from 1997 to 1998. Erosion and deposition maps of the area have been prepared and the erosion of island is measured at various sections at 1 km interval. Erosion of 1900 ha has been observed during the period of six years from 1991 to 1997 and 845 ha during the period of one year from 1997 to 1998.

Modeling urban floods and drainage using SWMM and MIKE URBAN: a case study

To avoid the nuisance of frequent flooding during rainy season, designing an efficient stormwater drainage system has become the need of the hour for present world engineers and urban planners. The present case study deals with providing a solution to stormwater management problem in an urbanized area. Mann–Kendall and Sen’s slope tests are used to perform the trend analysis of rainfall events using daily rainfall data (1956–2012), while the L-moments-based frequency analysis method is employed to estimate the design storm for a small urbanized area in West Bengal, India, using daily annual maximum rainfall (1975–2013). SWMM (Storm Water Management Model) and MIKE URBAN models are used to design an efficient drainage system for the study area. Two-dimensional (2D) MIKE URBAN model is primarily used to overcome the limitation of one-dimensional (1D) SWMM in simulating flood extent and flood inundation. Model simulation results from MIKE URBAN are shown for an extreme rainfall event of July 29, 2013. A multi-purpose detention pond is also designed for groundwater recharge and attenuating the peak of outflow hydrograph at the downstream end during high-intensity rainfall. This study provides an insight into the importance of 2D model to deal with location-specific flooding problems.