Big Questions

How does flood modeling enable communities to prepare for risk?

How models, maps and AI turn uncertain flood scenarios into clearer plans for people and infrastructure
Environmental Water Cities & Places
Elise Ibendahl
Global Principal, Flood Modeling and Planning
Flood platform Flood mapping

During recent flooding on a major river, I received an urgent call from a client. They could see the water level was nearing a record high, but needed to know what that meant for a nearby highway: Would a nearby levee overtop? Which low points would be affected first? Where should they sandbag? I had a high-quality flood forecast, but it didn’t align directly with the highway’s elevation information, so I had to quickly translate it into the locations and decisions they needed. We worked through the uncertainty and identified where to act. 

That experience captured how challenging flood risk management can be. A community might know the river is high and a gauge reading looks concerning, but understanding what that means for people, properties and infrastructure requires more detailed, focused information. 

Flood modeling contributes to that information, building a fuller picture of risk beyond traditional floodplain designations. A “100-year flood” means a 1% chance of that level of flooding in any given year; it doesn’t mean a community is safe for the next 99 years. Flood risk is a spectrum, and a single line on a map can’t show the impacts from every possible storm and localized condition. In many cities, floodplains for creeks and rivers are mapped, while localized flood risks from rainfall in sewered areas may be less visible. 

Comprehensive modeling can support communities with planning, preparing for emergencies and deciding where to invest. For example, we’ve used modeling to identify flood risk at proposed community facilities and infrastructure assets even when risk is not visible on regulatory flood maps. This presents an opportunity to consider protective measures before facilities are built. 

Models also allow us to test solutions before committing to construction. Looking across an entire watershed, we can assess if a project that reduces risk in one place increases it elsewhere. If there’s an increase in flood depth elsewhere, modeling contributes to an optimized solution that avoids the impact. Working with cost estimating and economics teams, we can compare benefits against investment, identify the most beneficial solutions and find the right sequence of projects in a capital improvement plan. For me, that’s where modeling delivers its greatest value: turning a clearer understanding of flood risk into practical decisions that strengthen community resilience. 

Using AI to test more scenarios and improve flood models 

One of the most useful applications of AI is enabling the understanding of what happened during past floods. A river gauge provides information at one location, but news reports, social media and service records may contain evidence of impacts elsewhere that weren’t previously comprehensively understood. AI can support with interpreting those sources and understanding where and why flooding occurred at a given location. Characterizing the impacts from past storm events gives us additional evidence to validate our models. 

Combining AI with traditional, physics-based modeling also allows us to explore more scenarios and potential solutions. Previously, an engineer might manually change individual pipe sizes or other features, run the model, review the results and repeat the process. Across a large watershed, that can take considerable time. 

Machine learning and optimization tools can accelerate those processes, identifying where changes could be most effective. We can incorporate project costs and test combinations of measures, including green infrastructure, to compare outcomes across the system. We took this approach with Melbourne Water, developing a “virtual flood laboratory” to test the performance and cost-effectiveness of mitigation measures at unprecedented speed.  

Tools such as Jacobs’ Flood Platform provide the computing capacity to automate, run and analyze thousands of different mitigation scenarios within a watershed. These AI-powered approaches provide communities and other decision-makers with confidence that the solutions they’ve selected are the right ones from a performance, community benefit and cost perspective. 

Detailed models can also inform faster, simplified models for flood forecasting, making their insights more useful as rainfall approaches. The value comes from combining these capabilities with engineering judgment and local knowledge. Timely, visual flood forecasts can clearly present the likelihood and scale of impact to local critical infrastructure, improving decision-making while understanding the assumptions and uncertainty present in forecasting.