Advances in Water Resources 146 (2020) 103785 Contents lists available at ScienceDirect Advances in Water Resources journal homepage: www.elsevier.com/locate/advwatres Flood risk forecasting at weather to medium range incorporating weather model, topography, socio-economic information and land use exposure Shrabani S. Tripathy a , Hari Vittal b,c , Subhankar Karmakar a,b,d , Subimal Ghosh a,d,e,∗ a Inter-Disciplinary Program in Climate Studies, Indian Institute of Technology Bombay, Mumbai – 400076, India b Environemntal Science and Engineering Department, Indian Institute of Technology Bombay, Mumbai – 400076, India c UFZ–Helmholtz Centre for Environmental Research, Leipzig D-04318, Germany d Centre for Urban Science and Engineering, Indian Institute of Technology Bombay, Mumbai – 400076, India e Department of Civil Engineering, Indian Institute of Technology Bombay, Mumbai – 400076, India a r t i c l e i n f o Keywords: Extreme rainfall Flood risk Weather forecast Hazard Vulnerability Exposure a b s t r a c t Non-structural mitigation measures to the globally increasing flood events include forecast based alert generation. However, the extreme rainfall forecasts are associated with low hit rate, high false alarm, and spatiotemporal bias; which makes it difficult to rely on them. Further, the losses due to flood in a region not only depend on rainfall severity but also on topography, socioeconomic conditions and exposure of the region to floods. Here, we introduce a new concept of spatial flood risk mapping and forecasting at weather to medium range based on forecasted hazard, embedded with vulnerability (topographic and socioeconomic) and exposure information. We define hazard as the probability of extreme rainfall event during upcoming days given an available weather forecast for the same days. As hindcast is used for computation of probabilities, hazard contains prior information about the false alarm, hit rate and spatiotemporal bias of the forecast. Vulnerability is calculated by averaging the topographic and socioeconomic indicators, and exposure is calculated using a land use land cover map. Topographic vulnerability is computed with digital elevation model using Height Above the Nearest Drainage method, and Data Envelopment Analysis is performed to derive the socioeconomic vulnerability based on the demographic census data. For a specific region and a specific event, the relative flood risk maps are generated at an administrative level (e.g., district, subdistrict or village level for India) and the high-risk areas can be identified from those maps for mitigation. The methodology is demonstrated for a very recent extremely severe flood event that happened in Kerala, India in August 2018. It is evident from the results that the high-risk areas forecasted well in advance (as high lead time as 15 days) match fairly well with the areas, which suffered maximum losses because of direct flood. 1. Introduction Climate change has caused a considerable impact to the global water cycle which lead to changes in seasonal patterns as well as an increase in the frequency of extreme rainfall events (Oki and Kanae, 2006). Inter- governmental Panel on Climate Change (IPCC, 2012) reports that this increase in extreme rainfall is statistically significant in many parts of the globe. It also states with medium confidence, that these changes in extremes are attributed to anthropogenic influences. In the Indian sub-continent as well, both empirical methods and model projections have shown an increase in the frequency and magnitude of extreme rainfall events (Goswami et al., 2006; Preethi et al., 2017; Roxy et al., 2017). A decrease in the total rainfall and intensification of extremes in the tropics for 21st century has been projected in the IPCC reports (Seneviratne et al., 2012). Irrespective of climate change, the impor- ∗ Corresponding author at: Department of Civil Engineering, Indian Institute of Technology Bombay, Mumbai – 400076, India. E-mail address: subimal@civil.iitb.ac.in (S. Ghosh). tance of forecast and understanding its uncertainty have a very high societal relevance. Increasing extremes in a changing climate further in- creases its importance. Extreme rainfall events not only affect our day to day lives but also result in floods and landslides which cause huge loss of lives and property (Dottori et al., 2018; Fowler et al., 2010). Every year a billion people are affected and thousands die because of these extreme events (IFRC/RCS, 2011). According to the National Disaster Management Authority (NDMA, 2008), floods have become a cause of concern because of an increasing trend in flood related losses in India. Some examples of such extreme events include the heavy rainfall event in Mumbai, India on 26th July 2005, which recorded 944 mm rainfall in 24 h (Jenamani et al., 2006). The rainfall and resulting flood caused the death of almost 1000 people and an economic loss of about US$100 https://doi.org/10.1016/j.advwatres.2020.103785 Received 29 April 2020; Received in revised form 24 September 2020; Accepted 5 October 2020 Available online 7 October 2020 0309-1708/© 2020 Elsevier Ltd. All rights reserved.