A Statistical-Physical Approach for Air Quality Forecasting

• Activity Number: 175
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Section on Physical and Engineering Sciences
• Author(s): YoungDeok Hwang*+ and Emre Barut and Kyongmin Yeo
• Companies: IBM Research and IBM Research and IBM Research
• Keywords: physics ; Forecasting ; multiscale ; air quality

Abstract

Physical processes, such as advection and diffusion, can transport atmospheric pollutants from the location at which they are generated, resulting in adverse impacts which can be distant from the pollution source. A physical model to incorporate the change of fluid field is a crucial tool in building an accurate forecasting model; but this requires instantaneous calibration of the physical model. In this paper, we propose a method to build a statistical forecasting model that couples the physical knowledge and observed data. The method obtains a regression fit for each of the spatial locations by solving a convex program, while simultaneously finding the input configurations for the physical model. The proposed approach is demonstrated through a real application.