Reliable floodplain delineation from hydrological predic-tions and digital terrain models in Iguaçu watershed, Brazil
ISBN 978-85-88783-11-9
Authors
1Madruga de Brito, M.; 2Mioc, D.
1UNIVERSITY OF BONN Email: marii.britoo@hotmail.com
2TECHNICAL UNIVERSITY OF DENMARK
Abstract
Floods are the most common and devastating disaster among all natural hazards in Brazil, with nearly 3,310 fatalities between 1991 and 2012. The Iguaçu watershed, located in Paraná State, is a critical area for flood management due to its importance in terms of hydropower production, with nine major hydropower plants. Flooding along this basin not only affects the production of electricity, but also cities and their inhabitants, causing economic losses, social impacts and, in extreme cases, losses of human lives. As an example, in 2014 severe floods occurred in União da Vitória municipality, resulting in one fatality and affecting at least 52,600 people. In order to reduce such losses, it becomes essential to develop flood forecasting systems. These studies yield useful information for emergency management, allowing appropriate measures to be taken on time. In spite of the advancements provided by flood forecasting systems, most of them provide simply the predicted water levels. Only based on this, it is difficult to determine which buildings will be affected by the expected flood event. Another issue is that the elaboration of hydrological and digital terrain models (DTM) contains a great deal of uncertainty and limitations, which are not always considered in flood forecasting. This paper presents part of an ongoing research which attempts to develop a methodological proposal for near real-time floodplain mapping in Iguaçu watershed, southern Brazil. The floodplain will be delineated through the integration of a DTM derived from airborne LiDAR (light detection and ranging) data with hydrological forecasts for the following 24 and 48 hours provided by COPEL (Companhia Paranaense de Energia). The modelling uncertainties will be computed by means of fuzzy logic, in order to provide the levels of uncertainty for a given forecast. Based on that, different scenarios with the spatial extend of the predicted floods are going to be displayed in a Web-based GIS. This platform will also show real-time telemetry data of river level and rain gauges, potential evacuation routes, shelters locations, satellite imagery and historical floods events, helping to perform a rapid dissemination of alerts and information about flooding. The results will be validated by comparing the extend of the forecasted flood with real data obtained by high resolution satellite images. The results of this research are expected to provide decision makers a cost-efficient floodplain delineation tool, which will enable the identification of areas affected by predicted flood events in a more reliable way. Thus, it will be possible to design more effective flood management strategies, allowing communities and individuals threatened by hazards to act in time to reduce damages.