Analysis of Rainfall –Runoff and Inundation Simulation of Floods in the Jhelum and Chenab

Abstract

Floods have been deemed as the most devastating of all natural calamities in recorded history, claiming millions of lives and causing wide spread damage to life and property. Pakistan has been affected by several devastating floods since its independence till 2010, mainly due to heavy monsoon rainfall in the months of June – September, totaled economical loss of around US$ 30 billion (Baig 2008). More recently, four consecutive years of severe monsoon flooding from 2010 onward, have affected 2.47 million people and caused 283 deaths mainly in the Punjab province (NDMA, 2014). The riverine flood of 2014 in the Jhelum and the Chenab river basins, caused large scale damages in the northeast Pakistani districts of Sialkot, Lahore, Narowal, Gujrat, Mandi Bahauddin, Gujranwala, Hafizabad, Sheikhupura, adversely affecting more than half a million people with about 250 causalities (IR&D, 2014). As flood modeling is one of the non-structural measures to combat floods, newly developed two- dimensional Rainfall Runoff Inundation (RRI) model is employed to simulate the four consecutive flood events of 2013, 2014, 2015 and 2016 over the 100,940 km2 catchment of the both rivers. As an input data, the model requires, 60 arc-second HydroSHEDS Digital Elevation Model (DEM), 30m Global Landcover, Food and Agriculture Organization of the United Nations (FAO) soil infiltration model. For near-real-time simulation corrected satellite-based, hourly rainfall data (GSMaP NRT) are employed. Model predictions are confirmed by comparing with observed water discharges at control stations namely Rasul Barrage, Khanki Barrage, Qadirabad Barrage and Trimmu barrage. In addition, spatial performance of model is supported by remote sensing imaginary (MODIS) observed from first control station of River Jhelum and Chenab i.e. Mangla Dam and Marala barrage to confluence of both rivers at Trimmu Barrage. In this study, model is calibrated for two flood events (2013 & 2014) including major flood of 2014, and resulting optimized parameter are calibrated for consecutive two flooding years of 2015 and 2016 respectively. The simulated hydrograph showed good agreement with the observed discharge at a section upstream of the basin outlet i.e. Trimmu barrage with Nash–Sutcliffe efficiency of 0.9, 0.87, 08 and 0.85 for consecutive flood events of 2013, 2014, 2015 and 2016 vii vii respectively. Also, simulated flood inundation extent is well predicated for major flood event of 2014 and showed good agreement with the MODIS imagery having fit (%) of 0.87. For some affected areas that experienced short-duration flooding undetected by MODIS, local crop and housing damage data confirmed the simulated results. Although the simulated results showed good agreement with the observed dataset, the model showed some discrepancy in identifying the flood-affected areas located near the confluence point along the Jhelum River. Also, the simulated flood peak is under-predicted which could be due to error in flow measurement at high discharges. Nevertheless, the model has been successful in identifying the critically affected regions emphasizing its utility as a tool to help manage floods.