Optimizing forecast-based actions to prepare for extreme rainfall in Peru

Natural disaster management has seen a major innovation recently with the advent of standardized forecast-based action and financing protocols. Given forecasts of sufficient skill and lead time, relief actions can be taken before, rather than after, a disaster, potentially saving lives and livelihoods while simultaneously transferring ex-post risk to ex-ante risk for relief and governmental agencies. Among other requirements, effective early action systems necessitate the availability of high-quality forecasts to inform decision making. While short-term predictions and early warning systems are not uncommon, there remains a dearth of applications at the monthly and seasonal scales which can foster actions requiring longer leads. Ideally, frameworks should include multi-stage actions, in which longer-lead forecasts allow for preparation while shorter-lead forecasts trigger direct actions. Explicit optimization of these plans, either through trigger mechanisms or forecast tailoring, can enhance performance by prompting appropriate actions at specific lead times. We apply this framework, in collaboration with the Peruvian Red Cross, to extreme rainfall events in coastal Peru. A sensitivity analysis of trigger thresholds, forecast methods, and levels of risk aversion is conducted to recommend optimal actions. Such optimization frameworks are attractive in that they can be utilized without post-disaster monitoring and evaluation, allowing for effective plan development in many disaster-prone regions.