A new method is introduced to evaluate nonlinear, nonstationary modeled PM_2.5 time series by decomposing decadal PM_2.5 concentrations and its species onto various timescales. It does not require preselection of temporal scales and assumptions of linearity and stationarity. It provides a unique opportunity to assess the influence of each species on total PM_2.5. The results reveal a phase shift in modeled EC/OC concentrations, indicating the need for improved model treatment of organic aerosols.