Refining Emission Source Databases Using Aircraft Data and Inverse Modeling: A Case Study of Arkhangelsk

The accuracy of emission source databases is a critical factor in air quality modeling. Remote sensing data, often used to compile such databases, can contain significant systematic and random errors related to retrieval algorithm features, atmospheric conditions, and spatiotemporal resolution. To improve database accuracy, we propose using aircraft measurement data through an inverse modeling approach to derive emission source power correction factors.
Using Arkhangelsk as a case study, we present an approach to refine the EDGAR emission inventory. The method is based on combining two models: the detailed WRF-Chem 4.7, which includes a full chemical transformation scheme and meteorological driver, is used for forward simulation of pollutant transport. A simpler atmospheric transport model is applied for source information correction, enabling repeated solutions of forward and adjoint problems at acceptable computational cost. Aircraft sounding data are used for result verification. This strategy is justified by the need for computational efficiency, as inverse modeling requires repeated solutions of both forward and adjoint problems and becomes prohibitively expensive when using only a detailed model.
Calculations using in situ aircraft measurements show that the proposed emission source power correction factor allows for effective and justified source correction, reducing the discrepancy between modeled and measured concentrations. However, this approach requires further development to account for boundary layer dynamics, the influence of diurnal and weekly emission cycles, and the accuracy of meteorological parameter specification, including wind fields, temperature, and turbulence characteristics. These tasks are planned for future research. The work was supported by by State Assignment topics of ICMMG SB RAS (data assimilation algorithms) and IAO SB RAS (data storage and processing), Russian Science Foundation Grant No. 23-77-30008 (regional model configuration).