Development of a Flood Forecasting System on Upper Indus Catchment Using Ifas (PAP014410)
Sugiura Ai, Fujioka Susumu, Nabesaka Seishi, Tsuda Morimasa, Iwami Yoichi
Flood forecasting and early warning systems
A flood forecasting system based on hydrological modeling is already covering main Indus but the upperreaches and Kabul river basin where most of the 2010 flood victims were located are not. This is thetarget for this research, providing a calibrated hydrological model to be the base of a flood forecastingmodel for Upper Indus river basin. Upper Indus river basin up to Taunsa, covering 577,000 km2 has beenmodeled with IFAS (Integrated Flood Analysis System) based on a 5 km gridded, spatially distributed 3layered tank model. The model building was performed in order to account as much as possible for localdata available. In particular, soil hydraulic data newly surveyed by Pakistan Council of Research in WaterResources have been integrated in this model. However, a part from the great area to be covered, thelack of sufficient local hydrometeorological data had also to be overcome. Indeed for the upper reaches ofthe basin, there are only 24 rain gauges covering 133,300km2 and only nine discharges measurementspoints at river stations, barrages and dams are available along the 1,650 km of Upper and Mid Indus. Themodel was calibrated on three flood events including 2010 floods and validated on three other eventsincluding 2012 recent floods. The simulation results suggested the uncertainty of rainfall data was great.For this reason, upstream discharges were input as boundary conditions and as a result, Nash-Sutcliffefficiencies reached a satisfactory average. Moreover, as an alternative to raingauges data, GSMaP-NRT(Global Satellite Mapping of Precipitation, Near-Real Time product) was calibrated and considered inrunoff analysis. Input rainfall slightly improved but not enough to explain all runoff and therefore, in orderto achieve acceptable performance, it is recommended to rely on upstream discharges as boundaryconditions.