“Predictive data modelling approaches for disaster forecasts and mobilizing pre-emptive responses”
Red Cross and Red Crescent (and other actors) have been working on developing more sophisticated approaches to predicting the likelihood of disasters and then using this early warning to mobilise early action. We want to ensure that analytical rigour shapes plans and actions for potentially life-or-death situations. If for instance, we can predict a very high chance of disaster impacts due to floods or other hydro-meteorological hazards we can begin mobilizing volunteers or ensuring pre-positioning of stocks or even distributing cash or enacting evacuation protocols (or other supports) with vulnerable communities ahead of the potential disaster to ensure that its impacts are reduced. However, given the probabilistic nature of the forecast, it is possible to act in vain if, for example the preventive evacuation of people is implemented but then the flood doesn’t materialize.
Another factor that limits the decision making process to act ahead of the disaster, is the lack of financial mechanisms to implement actions, that can be automatically triggered once a given forecast reaches a danger level. Understanding forecast skills, potential disaster impacts, early actions and preparedness for response actions that will reduce risks are critical to facilitate the decision making process to release funding in a timely manner.
What we need?
In order to address this challenge effectively we need to have;
- extremely good monitoring systems that send early signals when problems are looming;
We need enhanced data and models that can help us more accurately predict the likelihood of various disasters occurring. For this, hydro-meteorological services need methodologies and tools to enhance their capacity to analyse the skill of their forecasts, including reliability, false alarms ratios and lead times. At the same time to enable an efficient mechanism that allow governments and humanitarian organisations to act on time ahead of disasters, it is essential to understand disaster risks, (for example the potential number of population to be affected, where this population lives, potential effect on infrastucture, at which level of water or wind speed the houses are damaged, or at which level of water crops are damaged, etc).
(b) very detailed protocols that establish a relationship between a wide range of plausible warning signals (which describe the risk) and the corresponding decisions and actions (which should lead to the desired outcome).
We need protocols that clearly define;
- the magnitude of climate and weather parameters and potential disaster risks that pose a threat
- the early actions and preparedness for response actions that can be implemented to minimize risks, protect at-risk populations and ensure safety
- the forecast probability thresholds at which the forecast-based action should be activated – in other words, the ‘forecast trigger’
- the standard operating procedures (protocol) for staff addressing all the associated tasks: communicating change of plans, enabling financial flows to cover new expenses, describing step by step process to implement prioritized forecast-based actions etc.
- The critical piece in all of this is essentially to know whether or not any science-based signal of predicted conditions should be transformed into motion.
How can the Red Cross translate the data (e.g., a forecast for an impending flood exceeding the ‘danger level’) into a simple “Yes/No” output that can trigger early action and preparedness for response action on the part of humanitarian staff, volunteers, donors and partners, based on pre-defined standard operating procedures?
We do not need to wait until a disaster is looming to establish danger levels for forecast-based action. The humanitarian sector has to come up with criteria, protocols or other decision-support systems that can receive science-based predictions and recommend an adequate course of action – ensuring that financial systems are in place to support action in the precious time window before a disaster and after science says a hazard-causing extreme event is likely. We need smart, forecast-based humanitarian decisions as well as simple, decision-based scientific forecasts and observations.