Risk-based evaluation of performance in water management for the Ganges delta within Bangladesh

The Ganges delta area in Bangladesh plays a dominant role in the economy of the country which is primarily agrarian. The water resources of the region shape the economic activities, and therefore, a dependable appraisal of these resources is of vital importance. In this study, assessment of water availability from sources such as river inflow, precipitation, groundwater, residual soil moisture, standing water, and such others are made from existing data. Current and potential future water demands are estimated for water supply, irrigation, fishery, forestry, and other environmental needs from available information. A risk-based evaluation is then carried out in meeting future water demand of the Ganges Delta within Bangladesh (GDB). This evaluation is based on reliability, resiliency and vulnerability performance indicators, which have been newly defined in this thesis. The analysis included the impacts of climate change on both demands and resources, and generation of synthetic flows of the Ganges river.

For generation of the monthly flows of the Ganges river, different univariate stochastic models have been developed. Split-sample experiments with one-step-ahead forecasts revealed that the Periodic Autoregressive (PAR) model forecasts are better than the Seasonal Autoregressive Integrated Moving Average model and deseasonalized Autoregressive Moving Average model forecasts. A general algorithm has been developed for exact simulation using a PAR model. The model validation and verification using synthetic data indicate that the fitted PAR model is capable of preserving both short- and long-term important historic statistics of the Ganges flow. It also preserves the theoretical statistics. For disaggregation of seasonal stream-flow, generated with a PAR model, into sub-seasonal components, a periodic disaggregation scheme has been proposed and tested for monthly to decadal (10-day) flow generation of the Ganges river. To take the advantage of the power of periodic models, a new bivariate model has been developed by extending the popular Box-Jenkins' Transfer Function-Noise (TFN) model. The new model is named, 'Periodic Transfer Function-Noise (PTFN)' model. The development of the new model is an original contribution of this study. The well-known stages of model formulation, identification, parameter estimation, diagnostic checking, and forecasting are elaborated in the context of the PTFN model. The application of both models to monthly forecasting of the Ganges flow demonstrates that the proposed PTFN m odel p erforms consistently better than the conventional TFN model.

The need to redefine the risk-based performance indicators arose because the commonly used steady-state definitions of reliability, resiliency and vulnerability are not particularly appropriate when there is the presence of strong seasonality. Stochastic simulations carried out for evaluation of the indicators using synthetic data demonstrated that these modified indicators are more useful than their conventional counterparts.

Application of the models developed with the performance indicators redefined revealed that if surface water and groundwater can be used conjunctively, there will be no major problem in meeting future water demands of the GDB under the 'no change' and 'most likely' climatic scenarios. However, under the 'possible adverse' scenario, groundwater development may have to be increased by about 40% from its current use. The first decade of May and the second decade of April are particularly vulnerable to failures due to upstream withdrawal in India under the GWT of 1996. With minimum flow requirement in the Ganges river of 100 m3/s for in-stream uses, another 100 m3/s can be set aside for in-stream uses under the no change climatic scenario with the present level of groundwater use. The additional quantity will vanish under the most likely and possible adverse scenarios. At least 4000, 5000 and 8500 Mm³ of groundwater would be necessary to maintain 80% reliability of supply under the no change, most likely and possible adverse scenarios, respectively. Using surface water during early part of the dry season and groundwater during the later part, may be a good strategy to alleviate the water scarcity problem as well as to reduce the dependence on arsenic contaminated groundwater use. The proposed Ganges barrage will provide the opportunity to divert the surface water though its storage capacity would be small.