The use of new artificial intelligence (AI) techniques to learn how landscapes evolve is demonstrated. A few “snapshots” of an eroding landscape at different stages of its history provide enough information for AI to ascertain rules governing its evolution. Once the rules are known, predicting landscape evolution is extremely rapid and efficient, providing new tools to understand landscape change.

pyTopoComplexity is an open-source Python tool for multiscale land surface complexity analysis. Applied to a landslide-affected area in Washington, USA, it accurately identified landform features at various scales, enhancing our understanding of landform recovery after disturbances. By integrating with Landlab’s landscape evolution simulations, the software allows researchers to explore how different processes drive the evolution of surface complexity in response to natural forces.

The time it takes for a landscape to adjust to new environmental conditions is critical for understanding the impacts of past and future environmental changes. We used different computational models and methods and found that predicted times for a landscape to reach a stable condition vary greatly. Our results illustrate that reporting how timescales are measured is important. Modelers should ensure that the measurement technique addresses the question.