This chapter introduces a Bayesian approach for the estimation of distributed phenomena based on discrete time-space measurements obtained by a sensor network. We introduce a new methodology for sensor network applications, which rigorously exploits mathematical models of the distributed phenomenon to be monitored. By this unobtrusive exploitation, the individual sensor nodes collect information not only about properties of the phenomenon but also about the sensor network itself. The novelty of the introduced estimation method is the systematic approach and the consideration of uncertainties not only occurring in the mathematical model but also arising naturally from noisy measurements.