Environmental Engineering Reference
In-Depth Information
4.17 Application of Wavelet Filters in an Evaluation
of Photochemical Model Performance
P.S. Porter
1
, C. Hogrefe
2
, E. Gégo
3
, K. Foley
4
, J.M. Godowitch
4
,
and S.T. Rao
4
1
University of Idaho, Idaho Falls, ID, USA
2
ASRC, University at Albany, Albany, NY, USA
3
GEGO and Associates,
4
Atmospheric Modeling Division U S Environmental Protection Agency
1. Introduction
Air quality model evaluation can be enhanced with time-scale specific comparisons
of outputs and observations. For example, high-frequency (hours to 1 day) time
scale information in observed ozone is not well captured by deterministic models
and its incorporation into model performance metrics lead one to devote resources
to stochastic variations in model outputs. In this analysis, observations are compared
with model outputs at seasonal, weekly, diurnal and intra-day time scales. Filters
provide frequency specific information that can be used to compare the strength
(amplitude) and timing (phase) of observations and model estimates.
2. Methods and Time Series
2.1. Modeling system
Model outputs were produced by MM5-v3.7.2., CMAQ-v4.5.1., CB4 and aero3
set to simulate the time period 1988-2005 (Hogrefe et al., 2009). The domain was
the northeastern U.S. at a grid of 12 × 12 km. Emissions included NEI 1990,
1996-2001, OTC2002, and OTC2009, processed by SMOKE.
2.2. Observations
Observations (ozone concentrations and meteorological variables) used for time
series examples were recorded by the
Clean Air Status and Trends Network
(CASTNET, www.epa.gov/castnet/data/metdata/) operated by the Environmental
Protection Agency's Clean Air Markets Division. CASTNET sites are located in
mostly rural and remote areas such as national parks and monuments. Illustration
