Investigating the sensitivity of 20th century seasonal hindcasts to tropospheric aerosol forcing
(2025)
Authors:
Matthew Wright, Antje Weisheimer, Tim Woollings, Retish Senan, Timothy Stockdale
Abstract:
Previous studies have identified multi-decadal variations in the skill of winter seasonal forecasts of large-scale climate indices, including ENSO, the PNA, and NAO. Forecast skill is significantly lower in the middle of the 20th century (1940—1960) than at the start or end of the century. We hypothesise that tropospheric aerosol forcing, which is spatially and temporally heterogeneous and poorly constrained in the hindcasts used in previous studies, contributes to this low skill mid-century period.This study assesses the sensitivity of ECMWF’s state-of-the-art seasonal forecasting model to tropospheric aerosol forcing, using a newly developed aerosol forcing dataset based on CEDS emissions data. We analyse DJF hindcasts initialised every November from 1925—2010, each with 21 ensemble members. For each year, we run hindcasts with ‘best guess’, doubled, and halved aerosol forcing (perturbing both anthropogenic and natural aerosols). All experiments exhibit similar multi-decadal variability in skill for large-scale climate indices. Aerosol forcing has no significant impact on forecast skill but some impacts on mean biases, suggesting other factors drive the mid-century skill minimum.Aerosol forcing has large regional impacts. Increasing aerosol forcing leads to cooler 2m temperature and SSTs globally, with amplified cooling in regions with large aerosol forcings, such as northern India and North Africa. Dynamical responses include an ‘anti-monsoon’ circulation over Africa, with a weakening of the trade winds and Atlantic Walker circulation, and local southwards shift of the ITCZ. The magnitude of the response increases when ocean initial conditions are perturbed to represent the cumulative impact of aerosol forcing, suggesting that coupling enhances the atmospheric response.These results highlight the model’s sensitivity to tropospheric aerosols, with large differences in bias and mean state after four months, despite limited impact on skill. The circulation changes over Africa warrant further investigation, with implications for future aerosol scenarios. Planned experiments will explore the impact in summer and quantify the timescale of the response to aerosols.