The ARCO project focuses on the search and rescue of drowning victims in urban rivers. In a context of global warming and urban growth, the risk of drowning is expected to increase. Unfortunately, despite professional and well-trained rescuers, the proportion of rescued victims remains unsatisfactory, as urban rivers are harsh environments with strong currents, often turbid and cold water, many obstacles on the bed and very large search areas. ARCO proposes to significantly improve these rescue operations; firstly, by defining the locations, times and factors conducive to drowning in cities in order to build effective prevention and surveillance strategies; secondly, by estimating the location probability distribution of the victim at any given time in order to narrow down the search area; and thirdly, by improving underwater search techniques. Lyon, France, with the Rhône and Saône Rivers, is its main case study but ARCO will produce general results and methods. To meet these operational objectives, ARCO will address three major scientific questions. The first is the epidemiology of drowning: to whom, why, where and when river drowning occurs. The second question relates to the physical, biological and social parameters influencing the drift of a body within a river flow. The third is the method of hydraulic and probabilistic modelling of the victim's trajectory as an Eulerian-Lagrangian problem. ARCO will call upon various scientific disciplines that will combine their concepts, methods and protocols: observation in the field with rescuers, consultation of their archives, geo- and video-questionnaires to investigate practices and representations, experiments on body drift in the laboratory and in the field with dummies, and finally theoretical modelling and numerical calculations. ARCO will be organised in four work packages (WP). WP1 will produce a precise, complete and harmonised database on the occurrences of drownings and their context (e.g. the location, time, victim) in Lyon. It will be used to identify the main physical/biophysical (e.g. river discharge and temperature) and socio-cultural factors (uses, practices, representations) governing these drownings (for WP2 and 3). The extension of the database to other cities will help building a generalized model of hot-spot/hot-moment of drowning incident to help urban planners, policy-makers, river brigades and rescuers in their missions of prevention and security. WP2 will focus on the key parameters that influence the drift of a body. These parameters are many and relate to both the victim (clothes, weight and height, age, sex) and the river where it drowned (water velocity, temperature). The necessary parametric study will be based on tests with small-scale models in the laboratory, and in the field with a specific training dummy whose trajectory will be followed. WP3 will develop a model predicting the distribution of a victim’s locations in the river at a given time. To do so, it will integrate the flow conditions (WP1) and parameters (WP2) into a model of the victim's trajectory, itself applied to the river flow computed using an Eulerian modelling. The integration of this model into a smartphone or tablet application for use by the emergency services will not be done during the project, but it is the future target for WP3. ARCO will adopt an interdisciplinary scientific approach at the crossroads of human and physical geography, fluvial hydraulics and geomorphology. Our consortium includes five academic laboratories, a river operating company and end-users from a local fire and rescue service. In addition to scientific publications, in WP4 (results transfer task) the project will provide a drowning database, a guide to rescue and patrol practices, and the functional specifications of the future smartphone/tablet application to help divers from rescue services in urban areas with large rivers.