Methods for Adjusting Biases in Ozone Model Outputs for Use in Attainment Demonstrations and Exposure Assessments, Deliverable 1B.

摘要

This report evaluates methods for reducing bias in CMAQ ozone predictions so that they can be more effectively used in attainment demonstrations and exposure studies. Attainment demonstrations need unbiased estimates of design values for future years. Exposure assessments require contemporaneous unbiased standard deviation estimates. Four methods were developed and tested along with raw CMAQ and Relative Reduction Factor (RRF) values, using a long-term simulation study covering the northeastern US. Three of the methods project base-year CMAQ bias metrics to prediction-years. The projected metrics are (1) mean and variance, (2) mean and variance of temporal components, and (3) regression parameters of a Quantile-Quantile (QQ) plot. The fourth method uses QQ regression parameters only for the base year. Adjustment of modeled values so that they more closely resemble observations usually improves performance in attainment demonstrations. Conclusions about performance depend on the metric used.