Abstract

Fine particulate matter (PM2.5) has adverse effects on human health, affects climate forcing through complex interaction with radiative transfer and reduces ambient visibility. The diurnal variation of PM2.5 refers to the variation of PM2.5 mass concentration within a day, 24 hours. Understanding this diurnal variation is critical for building exposure assessment of PM2.5 at hourly resolution, predicting severe air pollution events and relating satellite-derived PM2.5 concentrations from overpassing times to 24h-averages. The diurnal PM2.5 variation has been widely observed around the world, showing a similar pattern across different continents and seasons. The PM2.5 mass concentration generally peaks in morning and late evening, with minima near daybreak and middle afternoon. The driving mechanisms of this variation were not fully understood. This dissertation uses the chemical transport model GEOS-Chem to interpret the diurnal variation of PM2.5, which includes three studies. The first study develops and evaluates the processes affecting the simulation of diurnal fine particulate matter variation in the GEOS-Chem model. The base GEOS-Chem model exhibits excessive PM2.5 accumulation overnight, overestimated diurnal amplitude and early timing of the morning peak and afternoon minimum comparing to in-situ observations. To reduce the above simulation biases, an emission inventory with hourly resolution is used, vertical representation difference between model and observations is resolved by correcting for aerodynamic resistances, boundary layer heights in the GEOS-FP meteorological inputs are constrained by aircraft observations and the dry deposition scheme is revised. The final revised study notably improves the capability of GEOS-Chem to reproduce diurnal PM2.5 variation over the US. Nevertheless, the nighttime nitrate remains after all developments applied, contributed mainly by excessive N2O5 hydrolysis according to the results from sensitivity simulations. The second study follows the model developments in the first study and extends the simulation of diurnal PM2.5 to a global scale using the GEOS-Chem model with the high-performance configuration (GCHP). Mass fluxes for surface PM2.5 caused by different model processes (chemistry, boundary layer mixing, convection, emission, dry deposition and wet deposition) are developed from the model diagnostics to quantify the driving forces of global diurnal PM2.5. Results indicate that the morning peak of observed PM2.5 is likely contributed by fumigation, the PM2.5 dip in afternoon is driven by both the growth of boundary layer and the partitioning into gas phase of semi-volatile aerosols, the PM2.5 accumulation throughout evening is contributed by chemical production and boundary layer collapse, and the PM2.5 dip near daybreak is likely contributed by decreased anthropogenic emissions. The third study focuses on simulating the diurnal variation of PM2.5 composition over Southeastern US using GEOS-Chem driven by the GEOS-FP offline meteorological datasets (GCHP) and the online WRF simulation (WRF-GC). The two GEOS-Chem model configurations exhibit different performance on reproducing observed diurnal variation of PM2.5 composition, likely caused by different representation of relative humidity, cloud water, dry deposition and boundary layer mixing. Driving forces of the diurnal variation of black carbon, organic matter, sulfate, nitrate and ammonium is revealed based on budget analysis. In addition, this study uses a Long-short Term Memory (LSTM) deep learning model to reduce the diurnal biases of nitrate simulation in GCHP.

Committee Chair

Randall Martin

Committee Members

Jay Turner; Jenna Ditto; Jian Wang; Katherine Travis; Randall Martin

Degree

Doctor of Philosophy (PhD)

Author's Department

Energy, Environmental & Chemical Engineering

Author's School

McKelvey School of Engineering

Document Type

Dissertation

Date of Award

12-9-2025

Language

English (en)

Download

Available for download on Wednesday, December 08, 2027

Share

COinS