Development of an Ensemble-based Data Assimilation System for a Chemical Transport Model and its Applications to Improve the Air Quality Predictions

Abstract

Improving air quality predictions remains a great challenge, mainly because of large uncertainties in the chemical transport models (CTMs). In recent decades, chemical data assimilation (DA) has become an effective tool to enhance the accuracy of air quality predictions via CTM simulations. This dissertation is mainly focused on the predictions of fine particulate matter (PM2.5) and its major components, which are the pivotal contributors to the air quality degradation in East Asia and South Korea. To improve the air quality predictions over East Asia, a data assimilation system has been developed by integrating the ensemble square root filter (EnSRF) DA method with Community Multiscale Air Quality (CMAQ) model. To build an efficient EnSRF–CMAQ DA system, a series of sensitivity experiments were conducted to determine the optimal values for key parameters, including prior inflation factor, horizontal localization width, and the number of ensemble members. This DA system was employed to optimize the initial conditions for total PM2.5 and the major aerosol species (i.e., NH_4^+, NO_3^-, SO_4^(2+), Organic aerosol, Black carbon, Sea salt, PM2.5,Others) related to PM2.5 by assimilating the ground-based PM2.5 over China and South Korea. In parallel with 6 h reanalysis runs, we also carried out 48 h prediction runs using the optimized initial conditions during the period of the KORUS-AQ campaign (i.e., 01 May to 11 June, 2016). Based on the reanalysis and prediction runs, the performance of the EnSRF DA method was also compared with those of other DA methods, such as ensemble Kalman filter (EnKF) and three-dimensional variational (3DVAR) method.