While we love deploying sensors, there are moments when useful data pops up in the most While we love deploying sensors, there are moments when useful data pops up in the most unexpected sources. In one collaboration, we used vehicles as hydrological sensors. Leveraging the extensive data collected by UMTRI's (University of Michigan Transportation Research Institute) connected vehicle program in Ann Arbor, we've correlated windshield wiper data from 80 vehicles with real-time radar information, giving rise to an entirely new rainfall data source. By introducing Bayesian update algorithms, we've not only enhanced the spatiotemporal measurement of precipitation but also laid the foundation for rainfall maps that seamlessly integrate radar data, traditional rain gauges, and windshield wiper states.

Simultaneously, accidental discovery of unconventional data sources has led us to another fun journey. In this study, we uncovered a hidden treasure trove of insights by fusing a staggering half a million previously untapped ship observations with a live hydrodynamic model. This unconventional fusion, made possible by our Probabilistic Process approach, has unveiled insights into hydro-meteorological processes spanning the vast expanse of the Great Lakes. It's a testament to the fascinating data that can be discovered when you venture beyond traditional methods.

Fundamental Advances: Data assimilation techniques, sensor fusion algorithms

Impacts: Entirely new meteorological data sets to inform management, mobility and navigation

Communities: Ann Arbor, Great Lakes