Evaluating trends and seasonality in modeled PM_2.5 concentrations using empirical mode decomposition

Area of Science:

• Atmospheric Science
• Environmental Modeling
• Air Quality Research

Background:

• Regional air quality models simulate fine particulate matter (PM2.5) and its components.
• Decadal simulations enable advanced model evaluation beyond traditional methods.

Purpose of the Study:

• To propose and demonstrate a novel process-based evaluation for speciated PM2.5 simulations.
• To assess the WRF-CMAQ model's ability to simulate long-term trends and cyclical variations in PM2.5 and its species.

Main Methods:

• Utilized improved complete ensemble empirical mode decomposition with adaptive noise (improved CEEMDAN).
• Evaluated version 5.0.2 of the WRF-CMAQ model using PM2.5 data from three monitoring sites.
• Analyzed time-dependent trends, sub-seasonal, annual, and interannual variations of PM2.5 and its components (SO4, NO3, NH4, Cl, OC, EC).

Main Results:

• The WRF-CMAQ model better simulates the rate of change in long-term trends than their absolute magnitudes.
• Amplitudes of sub-seasonal and annual cycles for total PM2.5, SO4, and OC were well reproduced.
• Significant phase shifts (up to half a year) were observed in the annual cycles of PM2.5, OC, and EC.

Conclusions:

• The findings highlight the need for improved temporal emission allocation and organic aerosol treatment in air quality models.
• WRF-CMAQ shows stronger capability in replicating sub-seasonal cycles compared to interannual variations.
• The proposed improved CEEMDAN method offers a valuable tool for sophisticated air quality model evaluation.