An intriguing natural phenomenon occurs every polar spring, namely the bromine explosion, in which plumes of tropospheric bromine monoxide (BrO) are formed. These plumes are observed in the BrO vertical column densities (VCDs), retrieved from satellite sensors. Tropospheric BrO depletes tropospheric ozone and facilitates the deposition of mercury. Bromine molecules are mainly released from young sea ice, and meteorological parameters determine the formation and evolution of enhanced BrO VCD plumes. Due to the complexity of the physicochemical processes involved in the bromine explosion, the modeling of tropospheric BrO VCDs in chemical transport models is challenging and not yet adequate. The first of its type, this study demonstrates the potential of using an artificial neural network (ANN), which uses meteorological parameters and sea ice age as inputs to simulate and predict tropospheric BrO VCDs in the Arctic. The ANN is trained and validated using a 22-year satellite remote sensing dataset of Arctic tropospheric BrO VCDs. A generally satisfactory spatial agreement between observed and simulated tropospheric BrO VCDs is observed. However, the magnitude of the observed BrO VCD plumes is underestimated. Air temperature and mean sea level pressure are the most important parameters influencing the magnitude of tropospheric BrO VCD simulations. Although the changing spatial distribution of tropospheric BrO VCDs over time is well captured, the trend reported in the observations of tropospheric BrO VCDs is not reproduced by the ANN, suggesting that additional parameters not included in the ANN also influence the formation of tropospheric BrO VCD plumes. •The prediction of BrO plumes is required to assess the loss of tropospheric ozone•We present an artificial neural network, which simulates Arctic tropospheric BrO•The neural network reproduces spatial patterns of many tropospheric BrO plumes•The trend reported in the observations is not evident in the simulations