Come monsoon, floods cause devastating disasters in one part or other of India. They are known to affect almost anyone at almost anytime. Hazards owing to floods can be: primary hazards that occur due to contact with water; secondary hazards such as disruption of services and health impacts like famine and disease; and tertiary hazards like changes in the position of river channels. In terms of both damage to properties and human casualties, floods are seen as costly disasters.

Knowledge of flood levels in advance will however help people and government to take precautionary steps to manage floods and control its aftereffects so as to minimize the woes to the people. Prediction of floods is mainly attempted in three ways: one, statistical studies can be undertaken to attempt to determine the probability and frequency of high discharges of streams that cause flooding; two, floods can be modeled and maps can be made to determine the extent of possible flooding when it occurs in the future; and three, as the main causes of flooding being abnormal amounts of rainfall over a short time span, rainfall should be monitored to provide short-term flood prediction.

The first paper of this issue, “Prediction of One-Day Ahead Flood-Levels of Kosi River Using Neural Networks”, deals with prediction of flood levels of the river Kosi in Bihar, one day ahead using neural networks. Its author, Sahay Rajeev Ranjan, constructed several neural networks—that have the ability to model a system merely based on past inputs and outputs without asking for physical, meteorological and hydrological aspects of the river system—with different number of neurons in the input. Using hidden layers, the model which yielded the lowest root mean square error and highest coefficient of correlation was selected for prediction of floods. The results indicated that the feed forward back propagation network with input set consisting of past two days’ river flood levels and four neurons in the hidden layers is the best performing model for forecasting the current flood level. The same data sets were used for prediction of flood levels from five developed auto regression models. Owing to the fact that the predicted results by neural networks being satisfactory and comparable to auto regression models, neural networks appear to be a good alternative for forecasting floods.

Moving away from floods to prediction of friction factor which is effected by the characteristics of the fluid, flow, geometry of the channel and the characteristics of the channel boundary, we have the authors, Bhoomi Andharia and B K Samtani of the paper, “A Mathematical Model for Predicting Friction Factor and Conveyance at Mahuwa Gauging Station Using Purna River Data, India”, who attempted to compute the friction factor based on the field data of Purna river and study the variation of conveyance with friction factor. Using the average value of the friction factor obtained by various methods and approaches and the multiple regression analysis, the authors have developed a model to predict the friction factor.

Cultivation of crops under irrigation system is found to consume more water and to be less productive and pretty expensive. Thus it raises the need for efficient water management under irrigation systems developed with huge capital outlay. To sustain the irrigation system, the concept of participatory irrigation management has indeed emerged as an efficient tool all over the world. Against this backdrop, the authors, T J Deepak, M S M Amin, Rashid Shariff, Rahman Ramli, M K Rowshon and Anusuiya Subramaniam of the next paper, “Web-Based Participatory Irrigation Management for Sungai Bernam River Basin”, have studied the 3-tier architecture framework and implemented it to create the WebPIM model that allows public participation in resolving disputes between stakeholders of Tanjung Karang Rice Irrigation Project in the Sungai Bernam river basin. Results obtained from the operation of the model reveal that there is no significant difference between the actual TWU and the WebPIM generated TWU, which means model values are at an acceptable level. Similarly, they have found no significant difference between the actual yield and the WebPIM generated yield, which again means model values are acceptable. But there is a difference between actual WPI and the WebPIM, pointing that the model values are not acceptable. Nevertheless, the authors have opined that participatory irrigation management is a right tool to manage irrigation systems and improve efficiency.

In the last paper, “Civil Engineering Aspects of Tsunami Resistant Buildings: A Forensic Approach”, the authors, Y S Prabhakar, M Potha Raju, and K Manjulavani, have presented a critical review of civil engineering aspects of location, planning, design, foundation and geotechnical practices that are capable of minimizing the effects of Tsunami as also the combined effect of earthquake and Tsunami.

-- GRK Murty
Consulting Editor