Several studies indicate a close correlation of environmental pressure (e.g. air pollution) and human health impacts. An effective development of political and academic measures and programs relies on the availability of sufficient and suitable data for the characterization of the considered environmental system. Currently, this is realised primarily by using data of administrative observation stations that are provided by public authorities. These administrative observation stations provide highly accurate measurements. However, spatial coverage and resolution are limited. We address this issue by applying an innovative approach based on Crowdsourcing and Citizen Science methods. By equipping citizens with a new kind of low-cost environmental sensor system an additional environmental data source is established. The observations of low-cost sensors in the urban area are used to densify data from administrative observation networks. These established approaches are based on the assumption that a real sensor is available at a particular location in the observed area. To overcome this mandatory requirement, we introduce the concept of Virtual Sensor Networks consisting of Virtual Sensor nodes. Different statistical models on particular locations characterise these Virtual Sensor nodes and the self-learning character of this approach enables a permanent monitoring and improvement. Finally, the use of crowdsourcing strategies increases the number of available Virtual Sensor nodes, arises the Virtual Sensor Network and leads to an improvement of spatial modelling of environmental parameters.