| 09 Oct 2025
Global identification of dominant ice-particle growth in cirrus clouds using EarthCARE satellite observations
Abstract. This study applies an ice-particle growth identification method to global observations obtained from the EarthCARE satellite. The method uses a joint probability density function of the equivalent radar reflectivity factor (Ze) and Doppler velocity (υd) on a common logarithmic scale (Ze–log10υd diagram), where the ratio of changes in Ze to changes in log10υd – referred to as the slope – serves as a quantitative indicator of dominant cloud microphysical processes. The analysis investigates the impact of random noise in Doppler velocity, which is a critical issue in EarthCARE products. In particular, three major error sources are addressed: observation window mode, along-track integration length, and bias correction related to antenna thermal distortion. These factors are found to significantly affect the derivation of slope values. To minimize the influence of noise, a representative slope is defined by calculating the median Doppler velocity in each Ze bin before applying the log₁₀ transformation. Using this revised method, EarthCARE's global observations reveal a systematic increase in the representative slope with atmospheric temperature across all latitude bands. While regional variations in slope are generally small, they nonetheless reflect distinctive microphysical characteristics specific to each region. These findings demonstrate that the EarthCARE satellite can be used to globally monitor cirrus cloud growth processes and offer a quantitative metric for evaluating the performance of climate models in representing cloud microphysics.
This manuscript presents a relevant contribution to the special issue “Early results from EarthCARE”, demonstrating the potential of combining the CPR radar reflectivity (Z) and Doppler velocity (UD) measurements to explore the universal ice Z–UD relationship across different temperature ranges. The paper is well structured and addresses an important topic, contributing to improving the representation of cloud microphysics in climate models using EarthCARE’s CPR observations.
My comments focus exclusively on Section 3.4, which describes the antenna pointing correction using surface Doppler velocity observations. The section provides a valuable characterization of the instrument error (intrinsic limitations and window mode-dependent uncertainty) but the mispointing correction approach raises methodological concerns. Several questions should be addressed to ensure the validity of the correction, particularly regarding its implementation, global applicability, and limitations over land surfaces. Providing additional context would also strengthen the section and is therefore advisable.
In the last few years, surface reference techniques have started to take increasingly more protagonism in spaceborne missions. The use of natural targets, like the Earth’s surface, has become fundamental to establish calibration records and to characterize and correct the geolocation and pointing accuracy of spaceborne remote-sensing instruments. The Earth’s surface provides globally distributed and temporally stable references that allow to detect small, systematic miscalibration, geolocation and pointing errors that cannot be captured by onboard telemetry alone.
These observation-based strategies have proven essential in missions like Aeolus, where surface and atmospheric returns were used to identify thermally induced pointing drifts modulated by orbital illumination (Rennie et al., 2021; Weiler et al., 2021). The use of natural targets has also been applied to geolocate CloudSat during the summer of 2019. The unexpected inclination of the satellite’s platform compromised the attitude data, and natural targets were effectively used in identifying and correcting the satellite’s geolocation (P Treserras and Kollias, 2024). The surface observations have also contributed to the calibration of climate records derived from spaceborne radar data (Kanemaru et al., 2024).
For the EarthCARE mission, using surface and atmospheric returns as natural reference points provides a direct and independent methodology for monitoring geolocation and correcting antenna mispointing (Tanelli et al. 2005; Tanelli et al., 2008; Battaglia and Kollias, 2015; Kollias et al., 2023; Scarsi et al., 2024; P Treserras and Kollias, 2024; P Treserras et al. 2025)
Based on the data collected following launch, it has been observed that the EarthCARE CPR experiences thermal deformation effects linked to solar illumination that influence its pointing accuracy. A detailed and comprehensive characterization of the mispointing behaviour as a function of both time of year and ANX time (time since ascending node crossing) is presented in P Treserras et al. 2025.
Although the surface may be treated as not producing vertical motion at nadir, EarthCARE observations have shown that this assumption does not always hold. In the absence of new insights or improved techniques, only the ocean and a limited number of land regions are suitable for antenna pointing assessment (P Treserras et al., 2025). Flat surfaces and uniform surfaces are expected to introduce no vertical motion at nadir, whereas heterogenic and rough topography can generate diffuse backscattering and significant terrain-induced Doppler effects due to slopes and variations in reflectivity causing non-uniform beam filling (NUBF) effects (Manconi et al., 2024).
Here, the authors acknowledge this limitation but still apply the pointing correction globally, later filtering out regions with apparent variability. The discussion seems to overlook and could better integrate some of these recent studies; for example, Kollias et al. (2023) is cited, even though that publication explicitly states that the surface pointing correction technique “is not recommended for application over the land surface”. As a result, several assumptions and interpretations in this section appear to diverge from previously established findings. A clearer connection with earlier results would help place the current analysis in context and ensure consistency with the existing understanding of the instrument’s pointing behaviour.
Regions where the pointing correction is unreliable are acknowledged and excluded from the study, but the criteria used for this filtering should be revisited to ensure a more rigorous identification of the mispointing correction behaviour, as it directly affects the quality and reliability of the presented results.
The authors state that the antenna pointing correction is derived from surface Doppler velocity measurements integrated over 100 km segments, yet they evaluate this correction using spatial and temporal aggregations that are much larger (10×10 deg lat/lon boxes and multi-orbit averages). This inconsistency introduces a significant scale mismatch between the correction and its verification, likely masking the actual variability and bias patterns that exist at the correction scale.
Such coarse averaging inevitably smooths out local biases and artificially reduces the observed variability. It also risks producing apparent Gaussian-like distributions even when the underlying signal might not be necessarily Gaussian. In practice, this means that the reported biases and standard deviations may not reflect the actual performance of the antenna pointing correction at its native 100 km scale.
The study should also clarify how these 100 km along-track averages are computed. For instance, directly averaging the Doppler velocity is only valid when the phase distribution is narrow and well within the Nyquist limit. The mean may be underestimated if the measurements are affected by velocity folding or if the phase distribution is broad, which is often the case over land.
In our own independent analysis using the same C-NOM L1b data (corrected for platform motion and antenna mispointing), using the technique described in P Treserras et al. 2025, the seasonal and orbital-phase dependencies are very accentuated. Over the ocean and snow-covered land, the surface Doppler velocity exhibits a near-zero bias, confirming that the antenna pointing correction performs as expected where the surface backscatter is homogeneous and stable. In contrast, substantial biases appear over land, sometimes apparently larger in seasons not reported in the study.
Because the ocean shows no corresponding bias, the deviations observed over land likely reflect surface-related effects rather than errors in the global correction. This suggests that trusting the surface Doppler over land introduces systematic errors rather than represent genuine antenna mispointing. Seasonal effects are also relevant and may obscure the characterization if they are not considered. The authors acknowledge that the offset values over land may be contaminated and that residuals depend on surface characteristics; however, these residuals can be significant and may even exceed the magnitude of the antenna mispointing correction itself.
Furthermore, the discussion about orbit phase appears incomplete. The text claims that “σfit does not vary significantly between the ascending and descending nodes (Figure 10b), except in a few specific areas (e.g., interior regions of China). Therefore, the ascending and descending nodes are not treated separately in the subsequent analysis.”. However, CPR data shows clear and systematic differences between ascending and descending orbits, including over regions not highlighted by the authors. These discrepancies suggest that important orbit-phase dependencies remain unresolved and should not be neglected.
It is assumed that the expected Doppler velocity at the surface is nominally zero. However, this should not be interpreted as a strict condition for all observations, since non-uniform beam filling (NUBF) effects (which are not considered in the study), surface characteristics and topography can introduce deviations from zero. Ascending and descending orbits reveal distinct surface Doppler velocity patterns that must be treated carefully. Correcting for antenna pointing by trusting the surface Doppler signal over land and using only the standard deviation to assess orbit-phase dependency and discard regions, may lead to an incomplete representation of the underlying problem and a misleading mispointing correction.
For example, the authors consider Alaska, Brazil, and northern and central parts of Russia, to be reliable regions because σfit falls below a certain threshold. However, seasonal climatological trends suggest that an underlying bias remains, which should not be neglected.
This effect can be seen in Figures 8 of the manuscript. The surface Doppler values over Brazil display similar colors for both ascending and descending orbits, whereas the signal over the ocean at the same latitudes shows an opposite sign. Although the variability difference is negligible and the region is considered reliable (Figure 10d), the Doppler velocities have opposite signs (Figure 8b, orange over Brazil versus blue over the ocean at the same latitudes), indicating that artifacts are affecting the pointing correction.
The antenna pointing correction is expected to exhibit a homogeneous pattern as a function of latitude for a given period of time. If the Doppler correction differs between land and ocean at the same latitude (as suggested), then the correction is not accurately capturing the true antenna mispointing but is biased by land-surface effects. A figure showing the global distribution of the antenna pointing would help evaluate the effectiveness of the correction technique.
The text also mentions that the 90ºS-60ºS is highly advantageous for the analysis, as it not only has smaller intrinsic Doppler velocity errors but also exhibits lower uncertainty in the antenna pointing correction. However, the authors also acknowledge that Antarctic sea ice produces surface Doppler returns that tend to approach zero regardless of the actual pointing error. Therefore, it is unclear how the apparent low uncertainty in this region is interpreted.
The authors also attribute part of the residual uncertainty in σfit to “lack of reliable topographic data.” It is unclear why inaccuracies in the digital elevation model (DEM) are cited as a major source of uncertainty. Why the antenna mispointing correction relies on an a priori DEM, and not from the detected surface return in the radar profile. If the correction depends on DEM-based surface height rather than the measured surface echo, the method stops being a pure observational-method and risks introducing additional geometric and geoid-related errors instead of capturing the true surface height revealing the antenna mispointing.
Addressing these issues becomes important in the context of the study, particularly if the goal is to characterize ice-particle fall speeds. The magnitude of these terminal velocities often lies below or within the same range as the expected residual bias from antenna mispointing. Without a robust and scale-consistent correction, any quantitative interpretation of the Z–UD relationship risks being dominated by pointing uncertainties rather than true microphysical variability.
Since the CPR antenna mispointing plays a critical role in the quality and reliability of Doppler velocity measurements, it would be advisable to address the correction methodology in greater detail, perhaps in a dedicated manuscript. Clarify the implementation, validate the correction at spatial scales consistent with its derivation (∼100 km) and re-examine the NUBF effects, latitude, seasonal and orbit-phase dependences, particularly over continental regions compared to the ocean. Otherwise, this and future studies study should directly exclude measurements collected over land, to ensure consistency with the existing literature and understanding of the correction technique.
References:
Battaglia, A. and Kollias, P.: Using Ice Clouds for Mitigating the EarthCARE Doppler Radar Mispointing. IEEE Transactions on Geoscience and Remote Sensing, vol. 53, no. 4, pp. 2079-2085, doi: 10.1109/TGRS.2014.2353219, April 2015.
Kanemaru, K., T. Iguchi, T. Masaki, N. Yoshida, and T. Kubota, 2024: Development of a Precipitation Climate Record from Spaceborne Precipitation Radar Data. Part II: Temporal Adjustment of the Calibration Change Using the Ocean Normalized Radar Cross Section. J. Atmos. Oceanic Technol., 41, 921–933, https://doi.org/10.1175/JTECH-D-23-0151.1.
Kollias, P. and Puigdomènech Treserras, B. and Battaglia, A. and Borque, P. C. and Tatarevic, A.: Processing reflectivity and Doppler velocity from EarthCARE's cloud-profiling radar: the C-FMR, C-CD and C-APC products. Atmospheric Measurement Techniques, Atmospheric Measurement Techniques, doi: 10.5194/amt-16-1901-2023, 2023.
Manconi, F., Battaglia, A., and Kollias, P.: Characterization of surface clutter signal in presence of orography for a spaceborne conically scanning W-band Doppler radar, EGUsphere [preprint], doi: 10.5194/egusphere-2024-2779, 2024.
Puigdomènech Treserras, B. and Kollias, P.: An improved geolocation methodology for spaceborne radar and lidar systems. Atmospheric Measurement Techniques, doi: 10.5194/amt-17-6301-2024, 2024.
Puigdomènech Treserras, B., Kollias, P., Battaglia, A., Tanelli, S., and Nakatsuka, H.: EarthCARE's Cloud Profiling Radar Antenna Pointing Correction using Surface Doppler Measurements, EGUsphere, 1–16, https://doi.org/10.5194/egusphere-2025-1680, 2025
Rennie, M., L. Isaksen, F. Weiler, J. de Kloe, Th. Kanitz, O. Reitebuch: The impact of Aeolus wind retrievals in ECMWF global weather forecasts. Q. J. Roy. Meteorol. Soc., Vol 147, Issue 740, 3555-3586, doi: 10.1002/qj.4142, 2021.
Scarsi, F. E., Battaglia, A., Tridon, F., Martire, P., Dhillon, R., and Illingworth, A.: Mispointing characterization and Doppler velocity correction for the conically scanning WIVERN Doppler radar, Atmos. Meas. Tech., 17, 499–514, doi: 10.5194/amt-17-499-2024, 2024.
Tanelli, S., E. Im, S. Kobayashi, R. Mascelloni, and L. Facheris, 2005: Spaceborne Doppler Radar Measurements of Rainfall: Correction of Errors Induced by Pointing Uncertainties. J. Atmos. Oceanic Technol., 22, 1676–1690, doi: 10.1175/JTECH1797.1, 2005.
Tanelli, S., Durden, S. L., Im, E., Pak, K. S., Reinke, D. G., Partain, P., Haynes, J. M., and Marchand, R. T.: CloudSat's Cloud Profiling Radar After Two Years in Orbit: Performance, Calibration, and Processing, IEEE T. Geosci. Remote, 46, 3560–3573, doi: 10.1109/TGRS.2008.2002030, 2008.
Weiler, F., M. Rennie, Th. Kanitz, L. Isaksen, E. Checa, J. de Kloe, Ngozi Okunde, and O. Reitebuch: Correction of wind bias for the lidar on-board Aeolus using telescope temperatures. Atm. Meas. Tech., 14, 7167–7185, doi: 10.5194/amt-14-7167-2021, 2021.