Unraveling ice multiplication in winter orographic clouds via in-situ observations, remote sensing and modeling

Mixed-phase clouds (MPCs) play a crucial role in Earth's climate system, influencing radiative forcing and the hydrological cycle. The distribution of ice and liquid water within MPCs is highly heterogeneous, impacting the efficiency of the Wegener-Bergeron-Findeisen (WBF) process and the rate of cloud glaciation. Accurately representing these processes in numerical weather prediction (NWP) and climate models remains a significant challenge, leading to model biases. Ice crystal number concentration (ICNC) is a key microphysical parameter, often exceeding expectations based on the limited availability of ice nucleating particles (INPs) at temperatures above -20°C. This discrepancy highlights the importance of secondary ice production (SIP), a process where new ice crystals are formed from existing ice, which is often poorly represented in models. This study aims to improve our understanding of SIP in winter orographic clouds by combining in-situ observations, remote sensing data, and high-resolution modeling, focusing on the identification of characteristic radar signatures associated with specific SIP mechanisms.

The importance of SIP has been widely acknowledged in various studies, including laboratory experiments, field observations, remote sensing, and modeling efforts. Common SIP processes include the Hallett-Mossop (HM) process, rime-splintering, ice-ice collisional break-up (BR), and droplet-shattering (DS) during freezing. While the HM process is often included in atmospheric models, its efficiency is limited to a narrow temperature range and specific cloud microphysical configurations, with some studies suggesting potential overestimation. Other processes, like BR and DS, may play more significant roles at colder temperatures, potentially skewing Doppler spectra towards slower-falling hydrometeors. However, ambiguity remains in interpreting radar signals, with downdrafts, winds, and turbulence potentially affecting the interpretation. Existing remote sensing methods using lidar and radar retrievals have shown widespread SIP occurrence, but further analysis using Doppler spectrograms is needed to investigate specific SIP mechanisms in under-explored temperature regimes.

This study utilized the Weather Research and Forecasting (WRF) model coupled with an updated Morrison microphysics scheme (M09) incorporating detailed descriptions of SIP processes. The model was used to investigate an intense snowfall event observed in mainland Greece during the CALISHTO campaign. The Cloud Resolving Model Radar Simulator (CR-SIM) was coupled with WRF outputs to simulate radar observations from a W-band spectral zenith profiler (WProf) deployed at Mount Helmos. Three sensitivity experiments were performed: CONTROL (default WRF PIP scheme), DEMOTT (updated aerosol-aware PIP scheme), and ALLSIP (aerosol-aware PIP scheme with four SIP processes: HM, BR, DS, and SUBBR). The simulations were compared against WProf observations, including Doppler spectrograms, timeseries of radar moments (Ze, MDV, skewness), and radiometer-derived liquid water path (LWP). The study analyzed radar data such as reflectivity (Ze), median Doppler velocity (MDV), and spectral skewness, to identify potential SIP fingerprints. In-situ measurements from the CALISHTO campaign, including aerosol size distributions and liquid water content, provided additional constraints for model evaluation.

The ALLSIP simulation, including all four SIP mechanisms, showed significantly better agreement with the observed radar reflectivity (Ze) compared to simulations considering only primary ice production (PIP). The activation of SIP mechanisms, particularly BR, significantly shifted the simulated Ze towards higher values, especially within the dendritic growth layer (DGL) between -20°C and -10°C. The analysis revealed that BR dominated SIP during the snowfall event, with limited contributions from DS and SUBBR. Snowflake aggregation was found to enhance BR within the DGL. Doppler spectrograms exhibited bimodal distributions, with a primary mode attributed to aggregated ice particles and a secondary mode, emerging at temperatures above -17°C, associated with BR-generated particles. The secondary mode showed a positive skewness in the Doppler spectra, indicative of slower-falling particles. The model also showed that BR efficiency increases at temperatures exceeding -15°C within the DGL, with a tenfold increase in production rates when nimbostratus cloud tops rise below the -25°C isotherm. In the seeder-feeder cloud configuration, BR and aggregation were consistently aligned. Overall, the study demonstrated a strong connection between positive skewness in Doppler spectra within the DGL, spectral bimodality, aggregation, and BR, suggesting that positive skewness could serve as a fingerprint for SIP via ice-ice collisions.

The results strongly support the importance of BR-driven SIP in orographic clouds during ice seeding events. The model simulations successfully replicated the observed increases in snowfall and radar reflectivity, demonstrating the crucial role of SIP in accurately representing these processes. The identification of positive skewness in Doppler spectra within the DGL as a potential fingerprint for BR offers a new avenue for detecting SIP using existing cloud radar data. This is particularly significant because it does not require polarimetric measurements, making it applicable to a broader range of datasets. Future studies should expand the analysis to larger spatial and temporal scales to assess the statistical significance of these findings and explore the applicability of this method to other regions and cloud types.

This study demonstrates the significant impact of SIP, particularly BR, on the microphysics and precipitation in winter orographic clouds. The findings highlight the importance of including detailed SIP parameterizations in NWP and climate models to improve the representation of MPCs and their effects on the climate system. The identification of positive spectral skewness within the DGL as a potential indicator of BR-driven SIP offers a promising tool for analyzing existing cloud radar data to improve our understanding of SIP at a global scale. Future work should focus on validating these findings using a larger dataset and exploring the limitations and applicability of this method to other cloud types and regions.

While the study successfully demonstrated the connection between positive spectral skewness and BR-driven SIP, the analysis was based on a single snowfall event. Further investigation using a larger dataset is necessary to confirm the generalizability of these findings. The model's representation of other SIP processes, such as DS and SUBBR, may also need further refinement. Uncertainties in the representation of PIP and the effects of wind shear and turbulence could also affect the interpretation of the results. Additionally, the study relied on a simplified representation of ice habits in the WRF model, limiting the detail provided about individual ice particle shapes.