The Radio Occultations and Heavy Precipitation (ROHP) experiment aboard the Spanish PAZ satellite was deployed in 2018 with the objective of demonstrating the ability of the polarimetric radio occultation measurement (PRO) concept in detecting rain (liquid-phase precipitation). Analysis of these data has also demonstrated the ability of PRO to detect horizontally oriented frozen-phase precipitation. To verify these observations, a global climatological comparison is performed using the CloudSat (94 GHz) radar as a reference for different heights and taking into account the radio occultation (limb-based) viewing geometry. A robust relationship (e.g., high correlation coefficient) is found between the polarimetric radio occultation observable differential phase shift (

The Radio Occultations and Heavy Precipitation (ROHP) experiment aboard the Spanish PAZ satellite

The shape and orientation of ice particles have a direct relationship with the lifecycle of precipitation

In

This strategy arises from the fact that there are no coincident observations between the CloudSat and PAZ due to different orbit parameters. However, the artificially generated products enable us to compare, in a statistical or climatological way, the PAZ

In this paper, we focus on the spatial correlations between the IWC as would have been sensed by RO geometries in CloudSat measured backgrounds and the

The CloudSat satellite was launched in 28 April 2006 and was regularly providing observations until August 2020. It orbits at an approximate height of 715 km in a polar orbit (inclination of 98.2

In brief, radio occultation observations consist of a low Earth orbiter (LEO) tracking a Global Navigation Satellite System (GNSS)-emitted signal while it is occulting behind the Earth's horizon

The approach followed in this study has been to re-map the CloudSat observed and retrieved parameters into the RO geometry. This is accomplished by using a set of actual RO rays and moving and rotating them in order to overlay them on the curtain-like CloudSat observations. RO rays often suffer from a drift as the occultation advances (e.g., the collection of rays does not form a perfectly vertical plane) due to the relative movement of the GNSS satellite and the LEO. However, for this study, the RO rays are collapsed into the CloudSat vertical plane. Each CloudSat orbit is split into 1200 km segments (resulting in approximately 29 segments per orbit), and the RO rays are placed in the middle of each of these segments. The first segment of each orbit is randomly displaced to ensure an even sampling of the Earth. After that, observed and derived CloudSat parameters are interpolated into the RO rays. See Fig.

Example of half of the CloudSat orbit observations (in this case, derived ice water content), with the RO rays overlapped. The CloudSat orbit granule is 01075. In panel

This approach permits the consideration of two relevant things. The first is that one can see the actual extent, and therefore the influence area, of the cloud structures as sensed by PRO. The second one is that the amount of IWC along RO rays can be quantified (e.g., in Fig.

Such an exercise is repeated for all CloudSat orbits between 2009 and 2013, resulting in 342 109 artificial RO observations. For each of these artificial ROs, the following parameters are stored:

The location and UTC time of each artificial RO observation are recorded. The location is determined by the latitude and longitude of the tangent point that has an altitude of 5 km. The UTC time is the one corresponding to the CloudSat observation.

The integrated IWC along each ray (as a function of tangent height) and the maximum IWC encountered along each ray are computed (e.g., Fig.

The distance each ray traveled within the influence of non-zero water content is computed (e.g., Fig.

Additionally, vertical profiles of some thermodynamic parameters at the location of each artificial RO, like the temperature, pressure, and specific humidity, are obtained from the ECMWF auxiliary files.

The resulting data are hereafter called the CloudSat-RO database. Since the focus will be only on the IWC, each CloudSat-RO event is truncated at the freezing level to avoid major contributions from liquid-phase precipitation. For this study, the effect of mixed-phase precipitation is not taken into account.

Given that CloudSat is orbiting on a polar orbit similar to that of the PAZ satellite and that RO rays are obtained from real RO events, the statistics gathered from the artificial collocation exercise should resemble the reality observed by PAZ. Furthermore, the integrated parameters along the RO rays mimic the behavior of

Figure

Statistics of the CloudSat-RO database corresponding to the rays whose tangent height is 9 km. Panels

The spatial/geographical patterns are also relevant. Higher concentrations of IWC at 9 km occur in the tropics (as expected), specially over land (South America and central Africa) and around the western Pacific Warm Pool. Regarding the distance, large values also appear over the mid- and high- latitude oceans. This results agree well with known patterns and the previous characterization of storm features, such as in

For this study, the DARDAR V3 retrieval has been used as a reference for the cross-comparison with

As discussed in the introduction, both

For the

The noise of the individual

Figure

Mean values of integrated IWC (left axes) from the CloudSat-RO dataset and mean values of observed

On a map, the gridded mean (or climatology) of

Global climatology maps for the mean PAZ

Panel

In addition to the mean, the 80th and 90th percentiles of both datasets are also computed at the same grid cells. That is, for each grid cell, the distribution for

Along with the correlation coefficient, the ratio between the PAZ

The correlation coefficients in Fig.

Correlation coefficients between PAZ

The results, in general (e.g., Fig.

When the data are split in different regions (e.g., tropics vs. extratropics), then more detailed features can be explored. The first relevant one is the fact that the correlation coefficient in the tropics (Fig.

The ratio between the mean climatologies of PAZ

Ratio between PAZ

The ratio is considered meaningful in the heights and regions where the correlation coefficient is high (e.g.,

The correlation coefficients obtained in Sect.

For this, the idea is to compute the

The forward-scattering simulations used for this study have been done using Rayleigh approximation. This formulation assumes that the particles can be approximated as oblate spheroids, with a certain axis ratio and effective density. The list of hydrometeors and ice particle habits that have been used are pristine ice crystals, aggregates of pristine ice, and wet snow. The adequacy of using Rayleigh approximations is justified by the long wavelength of GNSS signals, i.e.,

The simplest type of particles used in this study are pristine ice dendrites and thick plates. For the Rayleigh computation, the density of solid ice is used, and the values in

For more complex shapes, we use aggregates of pristine ice particles. For the Rayleigh approximation, a spheroid of air filled with portions of ice is used, following

Finally, wet snow aims to represent frozen particles at the initial stages of melting. For this, the same spheroids as in the previous subsection are used, but for this case, a mass fraction of liquid water of a 10 % of the total mass is used, in addition to pure ice and air, to compute

The results for pristine ice, aggregates, and wet snow are computed using single-particle scattering and forcing the different particles to be horizontally oriented. This is unlikely to be the case in real clouds and storms, where the orientation of these particles is more complex. To account for different orientation angles, we assume a Gaussian distribution of tilt (or canting) angles centered at 0

Using 1 week of CloudSat-retrieved

Results of simulated

Results of the distributions of simulated

The ratios between the

Ratio between

In these results, it can be seen how, for more pristine and thin particles, a certain canting angle is required (e.g., fully horizontal orientation is overestimating the observed

To further validate observations and results obtained in Sect.

For simplicity, these results are obtained assuming the same hydrometeor type at all heights for all situations, which is unrealistic. Similarly, the same effective density has been assumed for each hydrometeor at different sizes, while a more realistic approach would be to assume that the density decreases with particle size. However, these simulations can be used to explain the bulk of the observations and further assess the effect of considering more or fewer tilted particles in the simulations in reproducing the observations. Comparing the simulated results with the observations (represented in the top row of Fig.

The relationship between the PRO-observable

The results in Fig.

The PRO-observable

When the correlation coefficient is evaluated for the higher ends of the distribution (i.e., the climatologies for the 80th and 90th percentiles), then similar behavior is observed in the tropics, whereas differences between the mean climatology and the higher percentiles are observed in the mid-latitudes, especially over the southern oceans. This means that the features exhibited in the tail of the distribution of the retrieved IWC in this region are not well captured by PAZ

The ratio between the mean climatology of

This has in fact been observed in

Another interesting feature of the ratio between

The robust relationships between PRO

Seeking to assess the feasibility of the results in Sect.

Using a subset of

The results of the ratio between

Statistically comparing the simulated and observed

An important note to add is that the results presented here are valid for the chosen IWC retrieval algorithm (in this case, the DARDAR V3) and are therefore dependent on that choice. In particular, the DARDAR V3 algorithm assumes non-spherical ice particle shapes

This study assessed the global relationship between polarimetric radio-occultation-observable

The global climatology of CloudSat-retrieved integrated IWC has been compared to the

The ratios between the mean

Furthermore, forward-scattering simulations of simple snow aggregates accounting for a range of effective densities, axis ratios, and tilt angle distributions can explain the bulk of the ratios (i.e., the mean climatological values). The simulation framework described here, although being very simple, can easily scale in complexity for arbitrary complex inputs of ice particle habits from, e.g., numerical weather prediction models or to account for more realistic assumptions about the effective density.

Despite the high correlation between the mean values of

A comparison of the integrated IWC from the CloudSat retrieval using the 2B-CWC-RO and DARDAR V3 products has been performed and is shown in Fig.

Comparison between the integrated IWC from the CloudSat retrieval using the 2B-CWC-RO and DARDAR V3. Panel

The differences arise from different assumptions regarding the treatment of the CloudSat observations near the freezing level and the assumptions about the particle size distributions, and DARDAR includes Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) observations to complement CloudSat observations

The differences in the integrated IWC would change the results from Sect.

The datasets used for this work are the PAZ polarimetric radio occultation data, which can be obtained from

All co-authors have reviewed, discussed, and agreed to the final version of the paper. RP, EC, and FJT conceptualized the paper. RP analyzed the data and prepared the original draft. The investigation was led by RP, EC, and FJT. RP reviewed and edited the paper with EC and FJT, who also acquired the funding.

The contact author has declared that none of the authors has any competing interests.

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

The authors would like to thank Patrick Eriksson and one anonymous reviewer, for their comments and suggestions, which definitely helped to improve the paper.

Ramon Padullés has received funding from the postdoctoral fellowship program, Beatriu de Pinós, funded by the Secretary of Universities and Research (Government of Catalonia) and by the Horizon 2020 Research and Innovation program of the European Union under the Marie Skłodowska-Curie Actions. The ROHP-PAZ project is funded by the Spanish Ministry of Science and Innovation and by the European Regional Development Fund (ERDF), as part of “A way of making Europe” for the European Union. Part of the investigations has been done under the EUMETSAT ROM SAF CDOP4 and is based upon work from CSIC-PTI TELEDETECT members. F. Joseph Turk acknowledges support from NASA under the U.S. Participating Investigator (USPI) program. The work performed by F. Joseph Turk has been conducted at the Jet Propulsion Laboratory, California Institute of Technology, under contract with NASA.

This research has been supported by the Agència de Gestió d'Ajuts Universitaris i de Recerca (grant no. 2019 BP 00110), the H2020 Marie Skłodowska-Curie Actions (grant no. 801370), and the Agencia Estatal de Investigación (grant nos. RTI2018-099008-B-C22, PID2021-126436OB, and MCIN/AEI/10.13039/501100011033). This work has also been partially supported by the program Unidad de Excelencia María de Maeztu (grant no. CEX2020-001058-M). We acknowledge support of the publication fee by the CSIC Open Access Publication Support Initiative through its Unit of Information Resources for Research (URICI).

This paper was edited by Thijs Heus and reviewed by Patrick Eriksson and one anonymous referee.