A simple logistic model to understand the occurrence of flood events into the Biobio River Basin in central Chile

We predict flood events by fitting two simple logistic regression models between mesoscale rainfall and El Nino- Southern Oscillation (ENSO) index with local scale flooding into the Biobio River Basin in Chile from 1948 to 2002. Two different and complementary logistic models were studied: Logit and Probit. The models perform alike suggesting that the methodology is robust. The best model is most accurate during the autumn over 80% hit rate and in winter 66% hit rate, which are the seasons with high risk of floods. This work represents a first step in the development of a complete hydrological framework for the Biobio River Basin. The Logit distribution shows better results; thus, we suggest to use this distribution to relate flood events with mesoscale precipitation and ENSO index in the Biobio River Basin. Finally, these results explain the flood events and its relation with risk management. In forthcoming studies we will extend the methodology over this region and Chile taking advantage of the availability of higher resolution (temporal and spatial) weather products along with investigating climatic patterns and Intergovernmental Panel on Climate Change (IPCC) climate change scenarios.