Flood monitoring and damage assessment using water indices: A case study of Pakistan flood-2012

This paper uses Normalized Difference Water Index (NDWI) of McFeeters (1996), Water Index (WI) introduced by Rogers and Kearney (2004), referred to as Red and Short Wave Infra-Red (RSWIR) and WI suggested as the best by Ji et al. (2009), referred to as Green and Short Wave Infra-Red (GSWIR) for delineating and mapping of surface water using MODIS (Terra) near real time images during 2012 floods in Pakistan. The results from above indices have been compared with Landsat ETM+ classified images aiming to assess the accuracy of the indices. Accuracy assessment has been performed using spatial statistical techniques and found NDWI, RSWIR and GSWIR with kappa coefficient (kappa) of 46.66%, 70.80% and 60.61% respectively. It has been observed using statistical analysis and visual interpretation (expert knowledge gained by past experience) that the NDWI and GSWIR have tendencies to underestimate and overestimate respectively the inundated area. Keeping in view the above facts, RSWIR has proved to be the best of the three indices. In addition, assessment of the damages has been carried out considering accumulated flood extent obtained from RSWIR. The information derived proved to be essential and valuable for disaster management plan and rehabilitation. (C) 2015 National Authority for Remote Sensing and Space Sciences. Production and hosting by Elsevier B.V.