Use of blended evidence sources to build a history of flooding impact and an impact severity scale: A case study of Nairobi, Kenya

Many urban areas in the Global South are often data-poor and lacking in longer-term records of the occurrence and impact of natural hazards. Here we explore a methodology for using blended evidence sources to build a history of flooding impacts, along with a hazard severity scale, in Nairobi, Kenya, from 1978 to 2018. The evidence we use to build our Nairobi Flood Impact Database includes existing digital flood databases, newspapers, radio/TV broadcasts, government and NGO reports, peer-review journal articles, insurance company and emergency service records, online website reports, blogs, Google Analytic records and 330 photos/video from social media sources. For each record, we systematically extract from the source material, available information on the flood’s location, timing, and impact, with impact broken up into human (7 subcategories, e.g., fatalities), infrastructure (18 subcategories, e.g., building damage) and environment (6 subcategories, e.g., trees fallen). The resultant Flood Impact database has 1495 entries, which when entries that refer to the same flood event are grouped, result in a total of 354 flood events for 1978 to 2018 (41 years) in Nairobi, a much more extensive record than available previously. The flood database has the largest number of records for 2011 to 2018, given the increased use of social media and newspapers to report flood event impacts in recent years. We also see a peak in the years 1996 to 2000 (when there was a particularly heavy amount of rain due to El Niño) and then again 2016 to 2018. We then develop a five-point Likert scale for evaluating the adequacy of evidence types for recording location, timing and impact of floods. Finally, using a combination of existing impact-related natural hazard scales from the literature and our database, we build a five-part flood severity index combining the different types of impact, ranging from minor to catastrophic floods. Each of the impact types (31 subcategories) from our impact database is given a weighting from which inform this five-point severity index and map this severity index onto selected flood events from our Nairobi Flood Impact Database. Our database was then examined for temporal and spatial clustering in Nairobi and compared to different types of urban built up areas within Nairobi. This research provides a methodology and extensive blended database of the impact of floods over a 41-year period in a data poor area, providing a resource for better understanding the past history of spatial temporal hazard patterns, which can then be further expanded as to the causes, impact and how the flood events were dealt with in terms of recovery, lessons learnt and ways of mitigation and resilience building.