Advanced Approaches for the Collection, Quality Control, and Bias Correction of Smartphone Pressure Observations and Their Application in Numerical Weather Prediction

Abstract

Distributed worldwide, over 300 million smartphones are now capable of measuring atmospheric pressure, providing a potential surface observing network of unprecedented density and coverage. To examine the utility of such a network and test potential approaches for collection, quality control, and bias correction of smartphone pressures, a novel smartphone pressure app was developed. Within this app, observational uncertainty was quantified, quality control was performed, and sources of error were minimized. New machine learning techniques were developed to proactively correct observation bias. To test the performance of the app and evaluate the utility of smartphone pressure observations, WRF-EnKF experiments were performed for two case studies. In both case studies, smartphone pressures were able to constrain forecasts and analyses of observed and unobserved variables. In the second case-study full-cycling experiments assimilating smartphone pressures were able to successfully predict the forecast track and intensity of a major wind storm. Partial cycling experiments revealed that by improving initial conditions smartphone pressure assimilation can enhance forecasts at short lead times.