OCHA Centre for Humanitarian Data webinar on flood modelling for anticipatory action
The OCHA Centre for Humanitarian Data supports humanitarian partners in developing mechanisms that trigger anticipatory action. In March 2022, the Centre hosted a webinar exploring the use of flood models for anticipatory action in humanitarian contexts.
This webinar brought together experts and practitioners from Google, the University of Reading and the Red Cross Red Crescent Climate Centre to shed light on how flood forecasts provide critical information to help people get ahead of rising water levels before a crisis unfolds. Using examples from Nepal and Bangladesh, the speakers examined different types of flood model, the scale at which models provide risk information, and how models are interpreted and their uncertainties. Some highlights from the speakers are below.
Sella Nevo gave an overview of flood forecasting and flood models, including hydrologic and inundation frameworks and outputs. He outlined some general principles.
There are often lots of trade-offs between different things that we want. For example, if we are doing forecasts for a longer lead time, there might be a cost in terms of accuracy. If you are using flood forecasting models, I highly recommend trying to ask: what trade-offs are you making? What costs are you paying? If you are increasing the lead time, what does that mean about the data you will need to provide, or the accuracy that you will end up getting?
Liz Stephens addressed how to integrate flood models into humanitarian response. She looked at how to evaluate a forecast, what to do in data-poor settings, and how to use unverified models in times of imminent crisis. She shared some advice for using specific models for a response.
It is important that the forecasts used by humanitarian organizations are robust. We don’t want to be spending money unnecessarily when it can be used more effectively elsewhere. I would be very cautious about using any off-the-shelf assessment of forecast skill, as it is unlikely to provide something usable for our decisions. We really need to consider the decision-making context; not just in terms of the lead-times and thresholds that we may want to trigger the humanitarian action, but also in the criteria for forecast skill.
Madhab Uprety spoke about the application of flood models for anticipatory action in Nepal. He discussed regional flood history, forecast sources and the October 2021 flood. He also noted the importance of early action plans.
It is really critical for us to understand the different types of impacts, so we can look at which impacts we need to prioritize. Particularly those which we are prioritizing through early action, such as early warning messaging, building temporary flood shelters, house strengthening, giving supplies like hygiene kits and WASH [water, sanitation and hygiene] materials, the evacuation of livestock and supplying ready-to-eat foods.
Ahmadul Hassan talked about using flood models for anticipatory action in Bangladesh. He explored the expected impacts, lessons from past floods, and evaluated the model used and its response plan.
What is possible, based on the limited lead time that we have from the model? We came up with cash grants. This can be the plan as a response to reduce the impact. The cash grants, as anticipatory action, can reduce the loss of assets… [In flood crises, people] are forced to sell their assets at a very low price. With cash, at least they can save valuable assets and also access food, which is most important for them to purchase. So, we provide unconditional cash grants once the trigger is reached, based on the model we used.
You can watch the webinar recording here.