Idealized models or emulators of volcanic aerosol forcing have been widely used to reconstruct the spatio‐temporal evolution of past volcanic forcing. However, existing models, including the most recently developed Easy Volcanic Aerosol (EVA, Toohey et al. (2016): i) do not account for the height of injection of volcanic SO2; ii) prescribe a vertical structure for the forcing; and iii) are \NEW{often} calibrated against a single eruption.

We present a new idealized model, EVA_H, that addresses these limitations. Compared to EVA, EVA_H makes predictions of the global mean stratospheric aerosol optical depth that are: i) similar for the 1979‐1998 period characterized by the large and high‐altitude tropical SO2 injections of El Chichón (1982) and Mt. Pinatubo (1991); ii) significantly improved for the 1998‐2015 period characterized by smaller eruptions with a large variety of injection latitudes and heights. Compared to EVA, the sensitivity of volcanic forcing to injection latitude and height in EVA_H is much more consistent with results from climate models that include interactive aerosol chemistry and microphysics, even though EVA_H remain less sensitive to eruption latitude than the latter models.

We apply EVA_H to investigate potential biases and uncertainties in EVA‐based volcanic forcing datasets from phase 6 of the Coupled Model Intercomparison Project (CMIP6). EVA and EVA_H forcing reconstructions do not significantly differ for tropical high‐altitude volcanic injections. However, for high‐latitude or low altitude injections, our reconstructed forcing is significantly lower. This suggests that volcanic forcing in CMIP6 last millenium experiments may be overestimated for such eruptions.