Toward near real-time flood loss estimation: model structure and event definition

Abstract

Near Real-Time Loss Estimation Models (NRTLEMs) represent effective tools for developing improved parametric insurance products. This type of financial instruments enables rapid payments as they use one or more environmental variables measured immediately after the event and defined as trigger(s), to identify disaster events and predict the consequent impact. This study presents the preliminary development of such a NRTLEM, specific for floods. Given the importance of the event identification within the proposed methodology, different types of triggers are investigated and compared, with special focus on satellite precipitations estimates. NRTLE-based framework for identification of flood events in the Philippines using satellite precipitation estimates is investigated here. The methodology for event identification and the model calibration procedure are discussed. Finally, the model performance is investigated and the optimal configuration of model parameters minimizing basis risk, i.e., the mismatch between insurance claim settlement and the actual losses, is presented for the case-study application.