For Environmental Protection Administration (EPA) provides high resolution AQI information, this project focuses on the methodology of statistically post-processing the direct outputs from weather prediction model and air quality modeling model. First, the Model Output Statistics (MOS) is utilized to establish multi-variable linear regression models to predict the 0~72 hour PM2.5, PM10 and O3 concentration for each station. Procedures for data accessing, model building, real-time forecasting and performance verification are all developed in the project. On the other hand, the current EPA CMAQ modeling is migrated to the Central Weather Bureau (CWB) HPC under the EPA-CWB bilateral agreement and establishes the CMAQ OP1.0 process. The CMAQ OP1.0 can provide the 0~72 hour gridding pollutant concentrations forecast once per day. On the daily forecast operations, all of the PM2.5, PM10 and O3 predictive concentrations come from the MOS calculations and CMAQ OP1.0 simulations are aggregated to the objective AQI guidance for monitoring zones or counties.

Objective forecast, Model Output Statistics, Community Multi-Scale Air Quality modeling system