In a world where 86% of coastal areas face the looming threat of extreme sea-level rise, the urgency of accurate predictions cannot be overstated. Sea-level rise (SLR) can disrupt coastal agriculture, imperil food security, and endanger coastal ecosystems critical for climate resilience and biodiversity. Sinking cities would face catastrophic infrastructure damage and loss of homes. Understanding the multifaceted impacts of SLR on these interdependent systems is crucial for developing effective mitigation and adaptation strategies – but are our current approaches adequate?
The Challenge
Most current approaches to modeling inundation from SLR use what is known as the ‘bathtub model’. This assumes that all areas at an elevation lower than a projected flood level will be flooded, often done by projecting a flat flood surface over a global digital elevation model (DEM). It is easy to use, compute, and verify across extensive areas while still providing spatially-explicit information. Importantly, it does not require large, detailed datasets that are often lacking for understudied rural coastal zones.
However, this makes the accuracy of predictions heavily reliant on the accuracy of the DEM used. Most free satellite-based DEMs, which are often used for modeling large areas, have high vertical errors generally in the range of 1 to 10m - so areas could be anywhere between bone-dry and three storeys underwater! Ignoring such a wide error of margin potentially renders any decisions made to adapt to short-term SLR estimates as good as a coin toss.
Presenting global error estimates helps make this error explicit, but it does not solve the uncertainty issue at hand, nor does it inform us of the relative level of confidence across different topographies. Understanding the spatial distribution of error is especially important for efficient spending of climate adaptation finance, by funneling limited funds to priority areas that are most susceptible to permanent inundation.
Adatos' solution: Probabilistic SLR inundation risk analysis
That is why at Adatos, we developed a low-cost, scalable approach that provides spatially-explicit confidence levels for strategic climate adaptation and risk management to address SLR-associated challenges. We conduct inundation analysis based on site-specific elevation and tidal height, while accounting for the spatial distribution of DEM vertical error to understand relative risk levels for planning and management.
Adatos uses a probabilistic approach by combining our flooding algorithm with hundreds of simulations to confidently project future scenarios. We employ a Monte Carlo approach to model the effect of DEM errors on predictions by evaluating the distribution of simulated flood events that also accounts for spatial autocorrelation. Our algorithm also factors in land-to-sea connectivity and leverages water indices from open-source satellite data, adapting to ever-changing coastlines. This low cost approach allows for the solution to be adapted to different satellites, locations, and scales, while allowing the projections to be dynamically updated with changing sea levels.
Our proprietary technology allows us to go beyond a binary prediction of inundated areas by providing users with a confidence level for every pixel, facilitating more precise evaluations of local sea-level rise-related flooding risks. Rather than ignoring DEM uncertainty, we translate global error estimates into a spatially-explicit risk map that is meaningful output for decision-making at a local scale. This information is valuable for assessing vulnerability, prioritising intervention areas, and developing management strategies for both short- and long-term wetland analysis. Accurate SLR risk mapping for threat analysis allows us to establish the feasibility and permanence of potential project areas, predict mangrove migration, and assess climate risk to coastal agriculture and blue carbon ecosystems.
Comments