Paul Nörenberg and Hugues Lantuit, from the University of Potsdam and the Alfred Wegener Institute, conducted what their paper describes as the first quantitative GIS-based flood-inundation analysis using terrain data reproduced from a commercial video game. They extracted elevation data from GTA V’s Los Santos, calibrated it against real-world measurements, and applied published IPCC sea-level scenarios to the result. That calibration step is what separates the work from novelty and positions it as a replicable methodology.
The study was published in Geoscientific Communication.
The Method Behind the Simulation
The game functions as a visualization platform, not as independent scientific evidence of flooding risk.
Los Santos is a fictionalized version of Los Angeles, and its terrain contains enough geographic detail to function as a climate-visualization aid. The researchers compared flooding patterns in the virtual city with published assessments of real-world Los Angeles risk.
The scientific value comes from calibrating its terrain, applying established climate projections, and observing whether the results correspond to known real-world vulnerability patterns.
Two Pathways, Three Centuries
The researchers modeled two emissions scenarios across three time horizons, with and without an added storm-surge layer.
The researchers modeled two IPCC AR5 emissions pathways: RCP 2.6, representing strong mitigation, and RCP 8.5, representing a high-emissions baseline. They ran projections at three time horizons: 2100, 2300, and 2500.
Each scenario included an optional storm-surge addition of 1.591 meters, derived from the highest still-water level recorded at the Los Angeles tide gauge.
Under RCP 8.5 with storm surge, projected water levels reach roughly 5.18 meters by 2300, covering about 5.1% of mapped land; by 2500, the modeled level rises to approximately 8.22 meters and inundation grows to 6.5%.
In those high-emissions scenarios, flooding progresses through the marina and beaches before reaching the port, wastewater systems, and coastal energy facilities. The paper specifically describes the harbor and gasworks as becoming inoperable, while identifying other coastal infrastructure as at risk.
The lower-emissions pathway tells a different story. Under RCP 2.6 with storm surge, approximately 2.2% of the modeled land area is inundated by 2300, with no major infrastructure affected in the simulation. By 2500, that share rises only slightly, to approximately 2.3%.
Why the Virtual City Holds Up
The sequence of modeled flooding in Los Santos closely resembles published vulnerability assessments for coastal Los Angeles, according to the study.
The flooding sequence in Los Santos begins around the marina and moves toward the port. The authors report that this pattern broadly mirrors published real-world assessments of coastal Los Angeles vulnerability.
Players who already know Los Santos’s waterfront, roads, and landmarks may be able to connect projected water levels to places they recognize. The paper argues that this spatial familiarity could make abstract climate projections easier to visualize than centimeter-scale statistical maps typically allow.
What These Results Are Not
The outputs are scenario-based model projections built on specified assumptions, not operational forecasts for Los Angeles.
These are scenario-based projections, not predictions. Uncertainty in climate modeling grows substantially at the 2300 and 2500 time horizons; the paper specifically notes low confidence in the ice-sheet contribution at the 2500 horizon.
The results illustrate possible outcomes under specified assumptions. They do not forecast exactly where Los Angeles will flood or when.
The Broader Implication
The study’s contribution is methodological: it demonstrates that existing entertainment software may carry enough geographic fidelity to serve public-facing climate communication.
Entertainment software built to render a convincing fictional city may already contain enough geographic detail to serve as a public-facing science communication tool. The researchers worked with an environment that already existed, applying scientific methods to terrain that a large existing audience already knows how to navigate.






























