Simple Approach for Flood Forecasting

Bangladesh is a deltaic country located at the lower part of the basins of the three great rivers of the world- the Ganges, the Brahmaputra and the Meghna. Flood is one of the worst natural disasters, the country suffers from. The country is surrounded by hills on its three sides. The rainfall-runoff from this vast hilly area coupled with the snow melt in the Himalayas brings a huge inflow of water to Bangladesh during the wet monsoon (June–September) period. To mitigate the disaster due to flood a long term prediction is a must. Flood Forecasting and Warning Center (FF&WC) is responsible for flood forecasting in the country. FF&WC forecasts flood for 16 locations of the country for 24-hour 48-hour and 72-hour periods. Forecasting of flood only prior to 48 hours is even too short for the residents of the possible affected areas to take necessary precaution. One of the major agricultural crops in Bangladesh is Aman, which is planted in June-July and harvested in September – October which in turn is also within the flood period. Thus, a long term forecast of flood is necessary for the farmers also. This paper aims at developing simple means for forecasting the worst flood in any particular year from the observed flow at the beginning of the monsoon. The forecast is in terms of time when the peak can occur and its magnitude at the Hardinge bridge of the Ganges river. The location of Hardinge bridge is selected for the simplicity of analysis (at the initial stage). This is so because this location is far away from its confluence with the Jamuna. The Ganges has an upstream control structure beyond the national boundary of Bangladesh. The lean period flow in this river is, thus, not representative of the natural flow condition in the months of December to April. In the monsoon heavy rainfall dominates the flow in this river. The line of separation, in terms of time, is determined at first when the flow can be assumed to be dependent on natural phenomenon. Stage and discharge data of the river at the Hardinge bridge location is collected for the past 64 years. Ten day average discharge values are utilized for this purpose. The correlation of the peak discharges is tested with the ten day values of river flow at the beginning of the monsoon at the same location. The correlation is observed for different values of time lag. This provides the time difference between the real flood and the time when the forecast is made in relation to the degree of accuracy that can be achieved. Either of the two- degree of accuracy or the necessary time in hand has to be traded off depending on the reality and the need.