Blending Unconventional Data with Geopotential Height Anomalies to Understand Urban Floods in Hyderabad, India

Abstract:

One of the challenges for urban flood modeling is data scarcity, and the scarcity is the result of various factors including lack of details of storm water management system, absence of flow measurements and rainfall measurements at finer spatial and temporal scales. Also, importantly, absence of information on the floods is commonly observed, i.e., no archive of historical floods. This seriously limits understanding of floods in meteorological- and land use land cover- context. The focus of this study is to develop a detailed archive of historical floods for the city Hyderabad, India. Unconventional data sources such as newspaper archives, along with scientific literature and other sources for the period between 2000 and 2018 are used. A total of 50 flood events were identified and most of the flood events, ~ 76%, occurred during the monsoon season. The database comprises the following fields: date of the event, rainfall amount, meteorological conditions, water logged areas, impacts in terms of human losses and property damages. The key aspect observed from this database is that urban flood events have risen by 63% in the period of 2011-2018 in comparison to the decade of 2000 to 2009. This rise may be attributed to rise in impervious cover over the city in conjunction with inadequate drainage systems. However, to identify the meteorological influence, i.e., to understand synoptic conditions leading to these floods, the use of machine learning techniques are being pursued.