Boundary tracking and estimation of pollutant plumes with a mobile sensor in a low-density static sensor network

• We propose a method for tracking and estimating the boundary of a dynamic plume with a mobile sensor.
• It is an efficient method with low computational effort.
• No prior information about the plume or meteorological data is needed.
• We evaluate several variations of the method using simulated plume data.
• We give a number of practical recommendations on how to use the method in different scenarios.

Static sensor networks are used to detect concentrations of pollutants exceeding critical thresholds or to delineate the boundaries of contaminated regions in the air or in water. As the density of sensors deployed in the field may be locally insufficient to accomplish this task, it is necessary to add adaptive mobile sensors to the permanent sensing infrastructure. We propose a method for tracking and estimating the boundary of a dynamic plume with a mobile sensor in addition to binary below/above threshold signals from a low-density static sensor network. It is a heuristic method with low computational effort, which is efficient in terms of sensor movement. Neither prior information about the plume nor meteorological data is required. Our approach is applicable for dynamic plumes and operating with a small number of static sensors. We evaluate several variations of the method and parameter combinations using a plume simulated with the atmospheric dispersion model RIMPUFF. We give a number of practical recommendations.