École Nationale Supérieure de
Techniques Avancées

Abstract

This report encapsulates an 11-week research internship focused on the simulation and prediction of braided river behavior using techniques from geomorphology and computer science. This study is a comprehensive exploration of a novel approach that employs image processing, data analysis and geostatistics to recreate and predict the dynamism of braided rivers. The goal is to simplify complex natural processes into understandable segments, linking theoretical principles with practical outcomes. Our investigation builds upon the groundbreaking work of scientists like Julio Hoffimann and Céline Scheidt. J. Hoffimann’s methodology involves intricate image processing, advanced statistical, and mathematical modeling, which have practical implications for predicting the behavior of braided rivers and potentially preventing floods in certain regions of the world. This internship has led to new insights and robust methodologies that draw us closer to a comprehensive understanding of braided rivers. The report offers a detailed step-by-step guide to Hoffimann’s methods, a critical evaluation of these, and suggestions for improvement.