Abstract

Snowfall plays an important role in the Earth's water cycle, especially in orographically complex regions. However, snowfall in these regions is still poorly understood and subject to many uncertainties. CORSIPP aims to answer the following questions:

1. Which processes influence snowfall formation and snowfall rates in orographically complex terrain?
2. Which microphysical processes dominate precipitation formation and how do they influence precipitation rates?
3. What are the external forcing factors in complex terrain?

The project focuses on secondary ice production (SIP), especially in connection with riming processes, and analyses the influence of turbulence and frontal systems. For that purpose, a scanning W-band cloud radar and a novel video in situ snowfall sensor gathered extensive data for the entire winter season 2022-2023 in the Colorado Rocky Mountains as part of the SAIL campaign (Surface Atmosphere Integrated Field Laboratory: https://sail.lbl.gov). An overview of the field campaign is given in the Multimedia PageFlow created by the communication department of Leipzig University (https://unileipzig.pageflow.io/dem-schnee-auf-der-spur).

By simultaneously measuring snowfall with a Video In-Situ Snowfall Sensor (VISSS, Maahn et al., 2024a) and a 94 GHz dual-polarimetric W-band cloud radar (LIMRAD94, Küchler et al., 2017) direct information on the shape and size of individual snow particles is obtained and enables to derive particle size distributions and polarimetric quantities for the volume observed. The synergistic use of the two instruments and the forward operator PAMTRA allows us to answer the aforementioned key questions of the CORSIPP project.

In the second year the dataset was analyzed statistically to understand $K_{DP}$ signatures in snowfall, and the statistical analysis was complemented by in-depth case studies selected for detailed process analysis. Results of this research was published in Kötsche et al. (2025). Currently, we exploit the polarimetric variables using a novel technique to analyze the microphysical processes inside turbulent layers while avoiding the dampening effects of turbulence on radar polarimetry, a manuscript is in preparation. Additionally, the shape and degree of riming on snowfall particles are investigated using VISSS, and a comparison is made between orographically complex terrain and subpolar regions.

Investigating $K_{DP}$ signatures inside and below the dendritic growth layer with W-band Doppler Radar and in situ snowfall camera

Polarimetric radars provide variables like the specific differential phase ($K_{DP}$) to detect fingerprints of dendritic growth in the dendritic growth layer (DGL) and secondary ice production, both critical for precipitation formation. A key challenge in interpreting radar observations is the lack of in situ validation of particle properties within the radar measurement volume. While high $K_{DP}$ in snow is usually associated with high particle number concentrations, only few studies attributed $K_{DP}$ to certain hydrometeor types and sizes. We found that at W-band, $K_{DP} > 2\,^\circ\,\mathrm{km}^{-1}$ can result from a broad range of particle number concentrations, between $1$ and $100\,\mathrm{L}^{-1}$. Blowing snow and increased ice collisional fragmentation in a turbulent layer enhanced observed $K_{DP}$ values. T-matrix simulations indicated that high $K_{DP}$ values were primarily produced by particles smaller than $0.8\,\mathrm{mm}$ in the DGL and $1.5\,\mathrm{mm}$ near the surface.

Characterizing an orographic turbulent layer and the microphysical processes within it cloud radar and in situ snowfall camera observations

We have submitted another manuscript in which we focus on characterizing an orographic turbulent layer and the microphysical processes therein for the field campaign site in the Colorado Rocky Mountains. We would like to present a small excerpt of this manuscript: In Fig. 2, statistics of the turbulent layer height (TLH) between Sep 2021 and May 2023 (whole duration of the SAIL campaign) are shown. Liquid layer base (LLB) derived from the ARM high-resolution spectral lidar and cloud base height (CBH) were included in the statistics if a turbulent layer was detected. The most interesting feature is the collocation of TLH and CBH, with its peak just below the summit height of Gothic Mountain at just over $700\,\mathrm{m}$ AGL. This suggests that the turbulent layer plays a major role in cloud formation, possibly through enhanced moisture convergence in the lee of Gothic Mountain. The LLB was mostly detected a few hundred meters above TLH; still, the collocation of the turbulent layer and liquid layer base implies that the turbulent layer aids the formation of a supercooled liquid layer. The fact that LLB is slightly above TLH may arise from the fact that we look at a mean TLH and the turbulent layer has a few hundred meters vertical extent. Liquid water drops, due to their low weight, are most likely found at the top of the turbulent layer.

Analyzing riming and particle shape of snow particles at different locations

The properties of ice particles, including their number, size, shape, and growth processes such as aggregation and riming, are of central importance to the formation of precipitation, the lifetime of clouds, and the radiative properties of mixed-phase and ice clouds. The substantial variability in particle shapes poses challenges for cloud microphysics modelling and remote sensing retrievals. It is imperative to possess a comprehensive understanding of these shapes to enhance the reliability of cloud models, advance retrieval techniques, and refine climate projections.

In order to examine the distribution of ice particle shapes, the Video In-Situ Snowfall Sensor (VISSS, Maahn et al., 2024a) was deployed in several field campaigns at different sites (SAIL: Gothic, Colorado; ACTRIS: Hyytiälä, Finland; PolarCAP/Cloudlab: Eriswil, Switzerland). The VISSS system integrates two perpendicular cameras with a backlit configuration, facilitating high-resolution video recordings of hydrometeors from two distinct perspectives. The collected data were analysed using a supervised classification algorithm (python: sklearn.ensemble.HistGradientBoostingClassifier) that was developed with 1,000 manually labeled particles per shape category. The algorithm was implemented using Dfit, Contour, and Amplitude Ratio as the input parameters. The calculation of the Contour and Amplitude Ratio was performed using the Fast Fourier Transform (FFT) method.

The investigation addresses the following questions: firstly, which particle shapes are dominant at each location, and what is the frequency of riming? Secondly, what are their typical measures, such as maximum dimension or aspect ratio?

A substantial amount of effort was invested in the labelling of the particles and the identification and evaluation of the most appropriate particle variables for the optimal algorithm. As illustrated in Figure 3, the confusion matrix is a statistical tool used to analyse the accuracy of a classification system. The lowest true positive value recorded was $0.79$ for the classification of aggregates and graupel. The fact that graupel is often mistaken for spherical precipitation is not inherently problematic, as they exhibit a similar shape. The phenomenon of aggregates being indistinguishable from other entities can be attributed to the gradual nature of transitions between shapes, which often renders clear delineations impracticable.

The proportion of particles that are too small is comparable at $70$–$80\,\%$. Of the particles that could be identified, the majority were found to be either aggregates or graupel. It is noteworthy that in orographically complex regions, the presence of pristine particles is almost entirely absent. The particles are heavily rimed, and a significant proportion of them are aggregated or broken dendrites. At other locations where the VISSS was deployed such as Hyytiälä or Eriswil, several days were observed during which only needles and needle aggregates were found, a phenomenon that is probably related to secondary ice formation or the presence of long-lasting supercooled liquid stratus clouds.

In the first project year, simultaneous snowfall measurements with VISSS and LIMRAD94 have been performed at the Rocky Mountain Biological Lab (RMBL) in winter 2022/2023, embedded in the SAIL measurement campaign (WP1). As a result, a unique synergistic dataset of snowfall in orographically complex terrain has been published (Kalesse-Los et al., 2023; Maahn et al., 2024b). Currently, the dataset is analyzed statistically to understand KDP signatures in snowfall. The statistical analysis is complemented by in-depth case studies selected for detailed process analysis.

Analysis of Radar Data

Wind and turbulence play an important role in orographically complex areas and lead to an increase in riming and secondary ice production (SIP, Ramelli et al., 2021). We analyzed wind and turbulence and found that many of our measurement days were strongly influenced by Gothic Mountain. Our measurement devices are located in the lee of the Gothic Mountain during westerly winds, which leads to an area of lee-induced flow disturbance (ALIFD), resulting in increased wind shear along the edges of the ALIFD (at about 500 m and 1000 m AGL). The two areas of increased wind shear can be clearly identified by the two maxima of eddy dissipation rate at 500 m and 1000 m AGL (Fig. 1). WSW-WNW is also the main wind direction for precipitation events with an amount >0.5 mm/h. These layers of turbulence, which were almost always present, complicate the further investigation for the determination of riming and SIP, as common retrievals used to detect these processes are not reliable in turbulent conditions. Another point we focused on was the investigation of different Specific Differential Phase (KDP) signals. Using the collocated in situ measurements from VISSS, we found that snow particle populations with different properties, sizes and number concentrations lead to similar KDP magnitudes. This is shown in Figure 2 where averaged KDP values close to the ground plotted against D32 obtained from the VISSS (proxy for the mean mass-weighted diameter of the particle population) and the total number concentration Ntot. Currently, we cannot rule out the contribution of bigger, low number concentration aggregates on W-Band KDP as particle populations with low Ntot and high D32 produce similar KDP values as populations with low D32 and high Ntot. Another interesting find was that blowing snow appears to be capable of producing high KDP values as well.

Analysis of VISSS In Situ Data

Ice particle properties like number, size and shape, and processes like aggregation and riming influence precipitation formation, lifetime and radiative properties of mixed-phase and ice cloudiness. The variety of ice particle shapes complicates cloud microphysical modelling and remote sensing, and understanding these shapes is essential for accurate cloud modelling, remote sensing and climate prediction. Therefore, we focused on analyzing the different single particle properties, like shape, size and degree of riming for the 883.874.476 particles observed by VISSS. We estimated particle shape using a supervised classification algorithm (1000 labels per shape class). Figure 3 shows the distribution of different particle shapes for snowfall in Gothic in winter 2022/2023 December, January and February. More than 70% of the particles are too small (DMax ⇐ 0.5mm) to be classified correctly. Among the particles for which the shape can be determined, the most common particles are aggregates, followed by stellars/dendrites. Riming is one of the main growth processes of ice particles and can also lead to SIP (via splintering during riming, Hallet Mossop Process). Therefore, the in situ method of Maherndl et al., (2024) was used to investigate the frequency of riming and the result is shown in Figure 4. During the winter at Gothic, the proportion of heavily rimed particles with DMax >= 1 mm was over 40%. This agrees well with the statement made at the beginning that in orographically terrain with complex wind and turbulence systems, an increased frequency of particle riming occurs.

The next step is to compare the times with a high occurrence of one particle shape (e.g. only needles, only graupel, only dendrites, only aggregates) with the measured radar variables in order to detect differences in the radar variables depending on the particle shape.

Using SAIL measurements to support the ESA Earth Explorer 11 candidate mission WIVERN

WIVERN (WInd VElocity Radar Nephoscope, Illingworth et al., 2018), one of the two remaining ESA Earth Explorer 11 candidate missions, is planned to be equipped with a conical scanning 94 GHz radar and a passive 94 GHz radiometer. While the main objective of the mission is to measure global in-cloud winds, WIVERN reflectivity data can also be used to derive IWC and SR. Compared to CloudSat and EarthCARE, WIVERN's 800 km swath provides better coverage (70 times better than CloudSat) leading to significantly reduced the uncertainty of polar snowfall estimates (Scarsi et al., 2024). Further, it will be the first space-born cloud radar with polarimetric capabilities. Therefore, the slanted LIMRAD94 observations during SAIL were used to obtained statistics of KDP in snowfall for developing a WIVERN instrument simulator (Rizik et al., 2023). Further, SAIL data is currently used in a separate ESA-funded activity to explore using the potential of using relatively noisy space-born KDP observations for enhancing snowfall rate retrievals.

References:

• Illingworth, A. J., A. Battaglia, J. Bradford, M. Forsythe, P. Joe, P. Kollias, K. Lean, M. Lori, J.-F. Mahfouf, S. Melo, R. Midthassel, Y. Munro, J. Nicol, R. Potthast, M. Rennie, T. H. M. Stein, S. Tanelli, F. Tridon, C. J. Walden, and M. Wolde, 2018: WIVERN: a new satellite concept to provide global In-cloud winds, precipitation, and cloud properties, Bull. Am. Meteorol. Soc., 99, 1669–1687, https://doi.org/10.1175/BAMS-D-16-0047.1.
• Kalesse-Los, H., M. Maahn, A. Kötsche, V. Ettrichrätz, and I. Steinke, 2023: Leipzig university W-band cloud radar, gothic (colorado), SAIL campaign second winter (15.11.2022 - 05.06.2023), https://doi.org/10.5439/2229846.
• Küchler, N., S. Kneifel, U. Löhnert, P. Kollias, H. Czekala, and T. Rose, 2017: A W-band radar–radiometer system for accurate and continuous monitoring of clouds and precipitation, J. Atmos. Oceanic Technol., 34, 2375–2392, https://doi.org/10.1175/JTECH-D-17-0019.1.
• Maahn, M., D. Moisseev, I. Steinke, N. Maherndl, and M. D. Shupe, 2024a: Introducing the Video In Situ Snowfall Sensor (VISSS), Atmos. Meas. Tech., 17, 899–919, https://doi.org/10.5194/amt-17-899-2024.
• Maahn, M., V. Ettrichraetz, and I. Steinke, 2024b: VISSS raw data from SAIL at gothic from November 2022 to june 2023, https://doi.org/10.5439/2278627.
• Maherndl, N., M. Moser, J. Lucke, M. Mech, N. Risse, I. Schirmacher, and M. Maahn, 2024: Quantifying riming from airborne data during the HALO-(AC)3 campaign, Atmos. Meas. Tech., 17, 1475–1495, https://doi.org/10.5194/amt-17-1475-2024.
• Ramelli, F., J. Henneberger, R. O. David, A. Lauber, J. T. Pasquier, J. Wieder, J. Bühl, P. Seifert, R. Engelmann, M. Hervo, and U. Lohmann, 2021: Influence of low-level blocking and turbulence on the microphysics of a mixed-phase cloud in an inner-alpine valley, Atmos. Chem. Phys., 21, 5151–5172, https://doi.org/10.5194/acp-21-5151-2021.
• Rizik, A., A. Battaglia, F. Tridon, F. E.Scarsi, A. Kötsche, H. Kalesse-Los, M. Maahn, and A. Illingworth, 2023: Impact of crosstalk on reflectivity and doppler measurements for the WIVERN polarization diversity doppler radar, IEEE Trans. Geosci. Remote Sens., 61, 1–14, https://doi.org/10.1109/TGRS.2023.3320287.
• Scarsi, F. E., A. Battaglia, M. Maahn, and S. Lhermitte, 2024: How to reduce sampling errors in spaceborne cloud radar-based snowfall estimates, EGUsph. (rev. TC), 1–23, https://doi.org/10.5194/egusphere-2024-1917.
• Vogl, T., M. Maahn, S. Kneifel, W. Schimmel, D. Moisseev, H. Kalesse-Los, 2011: Using artificial neural networks to predict riming from doppler cloud radar observations, Atmos. Meas. Tech., 15, 365–381, https://doi.org/10.5194/amt-15-365-2022.