Top-down estimates of SO2 degassing emissions from the Turrialba Volcano using in situ measurements from Unmanned Aerial Systems and the WRF-STILT model

In recent years, there has been a growing interest in the continuous volcanic degassing emissions as an important natural source of sulfur-rich gases and aerosols. To investigate the impact of volcanic degassing on atmospheric chemistry and climate forcing, chemical transport models rely on emission inventories compiled from various sources. For example, the emission database from the Aerosol Comparisons between Observations and Models (AEROCOM) project derives eruptive SO2 emissions from past literature, Volcanic Explosivity Index (VEI), and limited observations from satellite and in situ instruments. Additionally, for all volcanoes with historic eruptions, AEROCOM simply assigns a constant SO2 degassing rate of 6.2×10-4 kt/day. This rudimentary estimate can lead to large uncertainties in model simulations of the volcanic SO2 lifecycle and its impact on the atmospheric composition. In this study, we propose to apply inverse modeling techniques to estimate top-down SO2 emission rates from the Turrialba Volcano (10.025°N, 83.767°W) using in situ SO2 measurements from unmanned aerial systems (UASs) during March 2013. We predict SO2 mixing ratios along the UASs' flight paths based on the AEROCOM a priori SO2 emission dataset and the atmospheric trajectories and surface footprints simulated by the WRF-STILT model. We incorporate a high-resolution (~ 30 m) terrain data into the model in order to account for the effects of the complex orography on the wind conditions near the volcano. The predicted SO2 mixing ratios are compared with measurements in a statistical procedure to minimize the model-data difference thereby yielding improved posterior estimates of volcanic SO2 degassing emission rates. A detailed uncertainty analysis will be conducted during this study taking into account all sources of error in the inverse modeling approach, such as the SO2 measurements, meteorological inputs, model configurations (e.g., spatial resolution, model physics parameterizations), and back trajectory calculations.