Integrated Geo-Sensing: a Case Study on the Relationships between Weather Conditions and Mobile Phone Usage in Northern Italy

Meteorological phenomena, in particular rapid events with local dynamics such as the onset of heavy rainfall, may have several instantaneous impacts on human behaviour. These impacts include, for example, the utilization of private or public transportation or the scheduling of personal and business activities. Given the increasing accuracy of weather forecasting at any scale, understanding the relationship between weather patterns and collective mobility behaviour or activity can potentially provide valuable insights into understanding urban dynamics and/or demand for public resources such as transportation. Scientific studies show that user-generated traffic in wireless communication networks can serve as a proxy for spatio-temporal patterns of human behaviour. In this paper we explore the relationship between weather and mobile phone usage and indirectly on human behaviour. We link the timespace pattern of meteorological measurements with that of mobile phone usage in the same large scale area (a region in Northern Italy). Taking the spatial context into account, we compare frequency-domain statistics correlations between weather and telecom activity and how they change between mountainous, urban, and coastal areas. The results indicate significant relationships between weather conditions, telecom activity, and area type. Keywords— sensor networks, user-generated network traffic, environmental monitoring, human behaviour, factor analysis, spectral analysis, squared spectral coherence estimation