Implementation of fast and extensible landscape evolution models
In geomorphology as well as in many other areas of scientific research, the growing use of computer programs, notably for running simulations, is affected by issues of reproducibility and reusability. In these areas, a lot of numerical experimentation often leads to full-featured model implementations with complex codes and interfaces that become hard to maintain. Following good software engineering practices, we try to overcome these issues by providing a common,generic framework for building computational models and running simulations. This framework encourages model creation or extension using a fine-grained modular approach, which is suited for development of scalable implementations and which leaves much room for experimentation. Highly connected to the Python scientific ecosystem,this software is also designed to increase interactivity. We use the framework to implement a set of efficient algorithms (FastScape) into versatile models of landscape evolution that will potentially include many different erosion processes (e.g., bedrock river incision, hillslope erosion, marine transport and sedimentation, glacial erosion, etc.) and their control by climate or tectonic factors.
Project investigators: Benoît Bovy and Jean Braun
Collaborators: open to external contributions (open-source software)