The spatio-temporal variability of ice-nucleating particle concentrations quantified with a new benchmark dataset

Abstract. Ice-nucleating particles (INPs) catalyze primary ice formation, thereby raising the glaciation temperature of mixed-phase clouds. Glaciation substantially alters cloud microphysical and radiative properties. Therefore, the development and validation of models that predict INP concentrations receive considerable attention. Current model validations rely on sparse observational datasets. These validations compare modeled and measured INP concentrations at individual measurement points, but do not consider their temporal variability. This variability reflects the physical processes controlling INP abundance, and it therefore offers a largely unexploited constraint on the model physics. This study presents a multi-campaign dataset consisting of 110 000 immersion-mode INP observations. Statistical adjustments for differing sampling intervals and campaign durations are derived that allow the heterogeneous dataset to be synthesized into a global characterization of INP concentrations. The campaign means of INP concentrations at -23 °C span more than four orders of magnitude, from Antarctica to agricultural continental sites. The within-campaign spread is around half an order of magnitude and increases with temperature. A multi-campaign time-series analysis shows that almost half the variance of daily values resides within a week, and roughly 20 % beyond 3 months. Across the resolved temperatures (-28 to -15 °C), organic material accounts for up to 90 % of the INP population in the sampled environments. Earlier modeling studies that attribute most INPs to mineral dust therefore likely overestimate the dust contribution. The extensive collection of INP observations and the derived statistics of INP variation provide a strong foundation for improved model evaluation.