FLOOD FORECASTING AND EARLY WARNING STUDIES FOR PARTS OF BANGLADESH

Abstract

Bangladesh is a flood prone country and extreme floods inundate more than half of the country's landmass almost every year. The country is in the lowest ridge of Hindu Kush Himalayan region, which makes the country hydrologically very diverse, complex, and unique. Economy, environment, ecology, livelihood, and development are affected by devastating floods every year. Flood Forecasting and Warning Service (FFWS) of Bangladesh was established in 1972 as a permanent entity under Bangladesh Water Development Board (BWDB). Initially co-axial correlations, gauge-to-gauge statistical co-relation relationships, and Muskingum-Cunge routing method were used for forecasting of water levels in advance. Facing different devastating floods, Government of Bangladesh conceptualized that the advancement of the forecasting and warning system can improve the FFWS and accordingly different projects were taken up. Despite the advancement of FFWS, the present flood forecasting and warning system of Bangladesh has the following limitations: • Limitation in updating the morphometric characterstics of river basins which are needed for modeling the flood forecasting system, • The Lead-time of 48 hrs is not sufficient for disseminating the information to effect timely response of flood prone communities, • Improper hydraulic designs in the flood plains due to lack of hydrometric measurement of discharge and stage, • Limited co-ordination between associated organizations, and • Absence of feedback from end users with the system. The lead time in present flood forecasting setup can be improved significantly by introducing the concept of forecasted rainfall in flood forecasting models. Further, the present flood warning system can be improved significantly by people's participation and feedback from the involved communities. The present study has been taken up with the broad objective to develop the methodology for an improved flood forecasting and warning system suitable for Bangladesh using the web resources available in public domain and recent developments in hydrological modeling and GIS technology. The study involved the use of following methods: (i) Extraction of river network and catchment boundary using 90m (3-arc second) SRTM (Shuttle Radar Topography Mission) DEM (Digital Elevation Model) and GIS (Geographical Information System) application, (ii) Evaluation of ECMWF (European Centre for Medium-range Weather Forecasts) and TRMM (Tropical Rainfall Measurement Mission) rainfall data for flood studies, (iii) Development of flood forecasting system using MIKE11 Rainfall-Runoff (NAM: Nedbor-Afstromnings-Model), HD (hydrodynamic), and FF (flood forecasting) modelling and ECMWF data for increasing the lead time and minimizing the forecast errors, (iv)Estimation of design flood discharge and flood levels using L-moments based methods and different modules of MIKE11, (v) Assessment of existing early flood warning dissemination system based on literature review, interaction with associated organizations, and feedback from the end users. The selection of study areas for different objectives of the study was of critical importance as most of the rivers of Bangladesh are international rivers and there are a number of unresolved issues between the associated countries. Hence, the study areas were selected in a way that does not generate any controversy and the departments responsible for maintaining data records are minimum. In addition, the availability of data and particular objective of the study played major role in the selection of study areas. For development of flood forecasting system, Jamuneswari river system has been selected as it has a high density of raingauges and its stage and discharge data and the river cross sections at different locations are available. Also the entire catchment of the river lies in Bangladesh. To delineate the river network and catchment boundary of the Jamuneswari river catchment, a study has been conducted over twelve different catchments at different locations in Bangladesh to find out the limitation of 90m SRTM DEM (Shuttle Radar Topographic Mission; Digital Elevation Module) and D8 method of ArcGIS 9.3 software. Teesta subcatchment in Bangladesh has been used for estimation of design flood discharge and design stage. This approach has been developed using frequency analysis, MIKE11 NAM and HD model and GIS mapping technique. Flood prone areas of Dhobaura and Shibalaya sub-districts in Bangladesh have been selected for feedback from endusers to assess the existing early flood warning dissemination system (EFWDS). • Delineation of river network and catchment boundary using 3-arc second SRTM DEM The 90m SRTM DEM have been used to delineate river networks and to extract Jamuneswari catchments boundary using the D8 method and ArcGIS. While delineating the Jamuneswari catchment, limitation of 90m SRTM DEM in drainage network delineation were observed. This led to an extensive study to determine the limitation of DEM data and the D8 method. Twelve catchments of varying geomorphology were chosen from five hydrological zones of Bangladesh. Basin characteristics such as bifurcation ratio, drainage density and channel slope of the catchments were estimated and analyzed. From this study, it is concluded that, in flat terrains, having a slope flatter then 1:2850, delineation of drainage network must be carried out carefully using the Hydrology tool of ArcGIS software that uses the D8 method for delineation of drainage pattern and catchments. It is also recommended that other techniques excluding D8 method as implemented in ArcGIS, should be experimented with before a general conclusion about the use of SRTM data in flat terrains could be drawn. Evaluation of ECMWF and TRMM rainfall data for flood studies Under this study, ECMWF and TRMM daily rainfall data for three locations of Ganges, Brahmaputra, and Meghna (GBM) basins in Bangladesh have been analyzed by statistical visual verification, yes/no-dichotomous verification, and continuous variables verification methods. Bangladesh Meteorological Department (BMD) rainfall data for the years 2004 and 2006 are used as the reference data. The results of the analysis indicate the potential for use of both ECMWF and TRMM in flood studies. Quantitative precipitation estimates from ECMWF and TRMM may be used for areas where rainfall data are not available or where number of rainfall stations are inadequate. The supremacy of either of the methods of rainfall estimation over the other method could not be established. The ECMWF provides the rainfall data in advance and hence can be used in flood forecasting studies to increase the forecast lead time. The ECMWF forecasted rainfall data have been used in Jamuneswari Flood Forecasting System (JFFS) for augmentation of lead time. Development of Jamuneswari Flood Forecasting and Warning System (JFFS) In this study, a Jamuneswari Flood Forecasting System (JFFS) has been developed using MIKE11 NAM, HD and FF model in a study area Jamuneswari catchment in the northwestern part of Bangladesh. The used real time hydrometorological data of the catchment have been analyzed to reduce uncertainties. The effect of uncertainties in flood iii forecasting has been assessed by comparing efficiency index, coefficient of correlation, volume error, peak error, and peak time error. MIKE11 FF module has been applied to minimize error in the forecasted result. The 24-, 48-, and 72-hour ECMWF forecasted rainfall data for 2006 have been used in the JFFS for augmentation of lead time of flood forecast in the Jamuneswari catchment area. The results show that with increase in forecast lead-time, the accuracy decreases. For increasing the accuracy of flood forecasted result, the JFFS has then been updated using MIKE11 FF module with observed data. The updated JFFS has produced reliable and satisfactory results. The steps for developing this flood forecasting system are generic and can be used in any geographic condition in the world. Development of approach for estimation of design flow and stage A study was conducted in the Teesta subcatchment in Bangladesh for determining design flood flows and corresponding flood stages for different return periods using frequency analysis and MIKE11 model. Different distribution functions of frequency analysis were tested for their goodness of fit. The observed discharge data at Kaunia on the river Teesta was used for estimation of design flood. The Pearson Type III distribution was found best fitted by Kolmogorov-Smirnov, D-index, and L-Moment Diagram Ratio tests and accordingly 25-, 50-, and 100-year return periods design floods were computed. The river network of Teesta River was extracted from SRTM 90m DEM. The river network of Teesta subcatchment was then simulated by MIKE 11 NAM and HD model. The resultant time series of river stage was then compared with corresponding observed values. From the model, a stage-discharge relationship (Q-h) curve and respective equation were developed for Kaunia station on the river Teesta. The developed equation determines the corresponding flood stage of estimated flood flow of 25-, 50-, and 100-year return periods. The resulting flows and stages will be useful to design hydraulic structures, prepare flood extent maps, to assess vulnerability of flood damage for different return periods, and provide flood forecasting for early warning of floods. The approach presented would be applicable to similar river basin systems where data are limited and scarce. Assessment for improvement of existing EFWDS of Bangladesh Flood Forecasting and Warning Centre (FFWC) was established in 1972. The FFWC developed a comprehensive system for collection, processing, and transmission of data, preparation of flood forecasts and warnings on a daily basis and dissemination of forecasts and warnings to various government and non-government organizations, media groups and iv a ^ other concerned parties over the year. From establishment through present, EFWDS of Bangladesh has been improved under several projects. Despite several advances in flood forecasting system in Bangladesh, the existing system often underperforms because the warning dissemination and response of the end users are unsatisfactory. The present study has been taken up with the objective to critically assess the existing EFWDS of Bangladesh and suggest suitable improvements in this system based on review of literature, interaction with officials of various organizations involved in flood forecasting and dissemination, and interaction with the flood affected people of Dhobaura and Shibalaya sub-districts in Bangladesh. The recommendation for active participation by all related organizations has been made in this study. Two studies have been conducted by surveying the opinion of flood vulnerable communities so that all elements of the EFWDS would provide useful flood warnings to all potential users.