Wednesday, 31 January 2024: 11:45 AM
310 (The Baltimore Convention Center)
The hydroxyl radical (OH) lies at the nexus of climate and air quality as the primary oxidant for both reactive greenhouse gases and many hazardous air pollutants. The lifetime of methane, a major anthropogenic greenhouse gas causing about one-half of the warming of carbon dioxide since pre-industrial times, is largely set by the global integral of OH concentrations, which vary strongly in space and time. To better understand the spatiotemporal dynamics of OH, we utilize an existing 13-member ensemble of the CESM2-WACCM6 chemistry-climate coupled model spanning the years 1950 to 2014, in which ensemble members vary only in their initial conditions of the climate state in 1950. We show a substantial spatial variation of historical trends of tropospheric column OH. The OH trends in one decade between 2005 and 2014 ranges from -0.25%/year over Europe to +0.4%/year over South America. We then use a machine learning (ML) technique, a fully connected neural network, to emulate annual mean tropospheric column OH and apply this emulator to identify the dominant drivers contributing to the regional variation in the OH trends. Chemical drivers, including the tropospheric NO2 column, HCHO column and CO column, are the dominant drivers of OH trends during 2005-2014. Specifically, we find that a 1 %/year change in NO2 column results in a 0.11 %/year increase in column OH based on sensitivity tests from ML. In contrast, HCHO column and CO column reflect changes in the OH sink pathway, with a 1%/year change in HCHO column and CO column associated with a -0.1%/year and -0.04%/year change in OH, respectively. As these chemical drivers can be observed from space, we also use existing satellite observations to evaluate the chemical drivers in the CESM2 model, and then calculate the OH trends using the ML and input features constrained by the satellite observations.

