D5.3: Report on the benefits of SR-ESMs for dust generation and transport of a continental-scale dust storm

About this document This report describes the impact of model resolution on dust emission and transport and highlights examples for which dust modeling particularly benefits from storm-resolving model resolutions. Work package in charge WP5 Executive summary Aerosols are an integral part of the Earth's system and contribute significantly to the Earth’s changing energy budget. Mineral dust dominates the global aerosol mass with a contribution of more than two-thirds (Kinne et al., 2006). Key parameters to quantify dust emission are surface wind intensity and land-surface properties, such as surface roughness or soil texture. Hence, the accuracy at which both are described in models strongly influences that of the modeled dust patterns. Here, we demonstrate how the modeling of dust, one of the most important types of atmospheric aerosol, benefits from storm-resolving model resolutions. We further illustrate how storm-resolving dust modeling opens avenues for a more detailed characterization of dust phenomena. Conclusion & Results We coupled two state-of-the-art, physics-based dust emission schemes to the simplified aerosol module HAM-lite (Deliverable 22 [Del. Rel. No. D4.2]) in ICON. Using this model framework, termed ICON-HAM-lite-dust, we conducted a set of sensitivity experiments (EXP1 – EXP8) at a coarse resolution of 80 km to test different dust emission configurations with those two schemes against a reference simulation. Additionally, we conducted one experiment at a high resolution of 5 km (EXP4SR) with the same configuration as our coarse resolution experiment EXP4. With EXP4SR, we demonstrated the benefits of storm-resolving Earth system models for dust generation and transport exemplarily for (a) an area with strong topographical variation, (b) dust storms generated by the outflow from moist convection (haboobs), and (c) a continental-scale dust storm. In all three cases, we see great advances of the greater detail at storm-resolving resolutions for dust modeling, which we summarize in the following. Surface friction velocity, a key input parameter for dust modeling, shows much more realistic and spatially variable patterns in areas of pronounced topographical variations, for example, the Taklamakan desert, a well-known dust source north of the Tibetan plateau. Our experiments show that the locations of areas with high dust optical depth (DOD) differ considerably between resolutions. Climatological reference data based on satellite observations and model reanalysis (with assimilated satellite data) show stronger support for the results we obtain at our storm-resolving resolution of approximately 5 km (experiment EXP4SR) than for the coarser resolution experiment with the same configuration (EXP4). We developed an algorithm able to identify and track haboobs – severe dust storms generated by the cold pool outflow from moist convection – based on modeled temperature, vertical wind speed, and surface dust concentration. Using this algorithm, we showed that haboobs frequently occur in the storm-resolving simulation (EXP4SR), but, as expected, not at a coarse resolution (EXP4). The reason for this is the much more detailed representation of small-scale variability in meteorological parameters, such as winds and temperature. Storm-resolving modeling, together with our haboob tracking algorithm, allows for the first time to quantify and characterize haboobs globally in a consistent way. An equally consistent characterization is not possible from observations, because clouds interfere with dust detection in satellite data, particularly in the area of haboobs/thunderstorms, and the spatial coverage of in situ observations is too sparse to detect all haboobs. In EXP4SR, we identified a massive continental-scale dust storm comparable to the Godzilla dust storm occurring in June 2020. Similar to the latter, our analyses have shown that haboobs contributed significantly to the development of this storm. We also see from our results that dust outbreaks were often intense but small in extent. This spatial heterogeneity is hidden at coarser resolutions. As the next step, we plan to conduct a more detailed evaluation of the average (e.g., monthly or seasonal) spatiotemporal variability of dust patterns against observations (e.g,. DOD) to find the best working combination of dust emission scheme, drag partition,n and soil moisture corrections, and tuning parameters. We also plan to investigate possibilities for reductions of the computational costs of the newly implemented schemes. Once finalized, our developments in ICON-HAM-lite-dust will be included in and be available through ICON-HAM-lite.