the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 19 Aug 2026
Regime-dependent evolution of water vapor, cloud liquid, and cloud base before rain onset in the southeastern Alpine forelands
Abstract. Liquid water path (LWP) and integrated water vapor (IWV) provide essential information on the moisture supply and condensation available for precipitation formation, while cloud base height (CBH) indicates where saturation and cloud formation start. The main goal of this study is to quantify how these parameters evolve before rain occurs.
We analyzed approximately 4 years of collocated observations from the WegenerNet 3D Open-Air Laboratory in southeast Austria, including an X-band dual-polarization radar, microwave and infrared radiometers, a GNSS station, and rain gauges. We separated convective and non-convective rain with a multi-criteria scheme based on radar polarimetric variables and the radiometer temperature profile, and warm, cold, and mixed rain types by comparing the echo top height with the -5 °C isotherm. During the six hours before onset, LWP stays low and then increases sharply within the last hour, with the steepest rise before convective events (about 300 g m-2), while cold-season events show a more gradual increase of about 150 g m-2. IWV changes earlier than LWP and is lowest in the cold rain type in both regimes. The joint evolution shows that convective events first move towards higher IWV and then towards a strong LWP increase near onset, while both parameters increase together in non-convective events. CBH decreases in all categories, most sharply before convective events (1000 m) and more modestly in the cold season (300 m). Overall, these signatures can serve as a reference to evaluate how models represent the transition from moisture to cloud and rain.
- Preprint
(14661 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 30 Sep 2026)
- RC1: 'Comment on egusphere-2026-4256', Anonymous Referee #1, 19 Aug 2026 reply
-
RC2: 'Comment on egusphere-2026-4256', Anonymous Referee #2, 23 Sep 2026
reply
The paper investigates the typical behavior of liquid water path (LWP), integrated water vapor (IWV; the first two derived from microwave radiometer, MWR), and cloud base height (CBH) at the Wegener Laboratory, Austria, before the onset of precipitation. This involves a statistical approach where a large number of precipitation events are identified from rain gauges and classified with the help of polarimetric X-band radar into warm/cold and convective/rest conditions. As explained in the conclusions this could lead to a new metric for evaluating forecasting models.
While the topic is generally interesting, the paper needs substantial revisions (in particular how it is written) before it can be published. My main criticism is
- There is no clear motivation given in abstract and introduction. The conclusions are much better written and help to put the paper into context. A discussion on physical processes, precipitation efficiency, the identified gap in understanding, and the addressed spatial/temporal scales is necessary for a paper in ACP, which should address a broader audience. Because a major result is a new methodology for model evaluation, the authors should highlight that, nowadays (particularly within the ACTRIS network and operational radar), similar observations are available from multiple sites, enabling broader application of the technique developed here
- No information on the accuracy of the measurements is given. I point out several details about that later and suggest integrating this information into Section 2, which is rather short. It does not provide the basis for understanding the limitations of the observations (parts of it are in section 3 or even later, but it would be better to have it together for all instruments)
- The methodology should be sharpened: The choice for the considered averaging and sampling times needs to be made clear. There is no consideration of cloud fraction (CF); it is not even discussed. In my opinion, CF is the main reason for the finding “The joint evolution shows that convective events first move towards higher IWV and then towards a strong LWP increase near onset, while both parameters increase together in non-convective events”. I suspect that conditional averaging (mean of LWP when present) would lead to a rather different results as downdrafts in convective conditions.
- It is assumed that the four-year statistics are large enough statistics, and all events are weighted equally. Similar to a previous study for this site, they also look at the maximum precipitation above the 90th percentile but don’t consider other metrics. I think the radar data includes much more information to characterize the precipitation events, for example, on the precipitation pattern (are the ground-based measurements in the center or at the edge) and propagation direction and speed.
- Finally, I find it rather strange that the case studies are presented after the statistical summary. I would first introduce the methodology with one or two case studies and then move to the statistics as this is the main result to be discussed in the final conclusion section. Also, the discussion of the cases is rather descriptive and lengthy, tiring the reader so maybe even more can be moved into the appendix.
DETAILED COMMENTS
Abstract: The first sentence is rather detailed
L23: “these parameters”. What does this refer to? In the sentence before you talk about processes and scales. As noted in my major concern 1, I would expect more discussion of the physical processes and the difficulty of determining precipitation efficiency from observations (see for example Kukulies et al., 2024, https://doi.org/10.1029/2024JD041924).
Starting L33: Rather detailed results from previous studies are given but the main challenge of the study is ignored: How can local measurements contribute to the understanding of the end result of a complex chain (surface precipitation) that is influenced by atmospheric dynamics on various scales. The authors need to think a bit more broadly in the introduction as for example advection is never mentioned.
L45: The distinction between convective and non-convective scenes is a good approach to move toward more local - less advective - dominated condition and can be strengthened here. The rather detailed description of thresholding should be replaced by the idea behind and not the details.
L 51: This question is not explicitly answered in the conclusions
L62 onward: Refer to Fig.1 How large is the area? What is the typical precipitation? There is one sentence referring to the VB – but how much does this contribute? How does this relate to the investigations? Fig.1a shows the impressive ground-based station network. As far as I understand, these sites all measure precipitation. Why don’t the authors show the average precipitation rate over all events studied and separated into their different categories (warm, cold, convective..) such that the reader gets an overview how different their distributions are. In fact, it might also be promising to distinguish the events based on the circulation weather type, as these lead to different interactions with orography.
L79: How far are the precipitation measurements away from the central = radiometer (define) site?
Section 2: Table 1, which gives an overview of the data, is not referred to. There is no information on the key products, their uncertainty, and their temporal sampling in the text. For all data, a common resolution of 2.5 min is used. What is the reason for that? Typically, MWRs measure with second resolution, while GNSS have longer integration times as several slant path measurements to satellites are needed. The difference in spatial (and temporal) samples causes a different sensitivity to water vapor variability, especially in convective situations (Steinke et al., 2015, doi:10.5194/acp-15-2675-2015). Thus, this partly explains IWV differences observed between both instruments, as for example stated on line 236. In general, it was not clear when you use MWR or GNSS and why? Another missing information is how CBH is derived. The IR temperature is affected by water vapor emission between cloud base and surface so there is a sensitivity to IWV here. Also, I don’t know how meaningful the measurement is during virga precipitation. Another point, it is the temperature profile: Section 2 should explain that the MWR retrieval is used (always?) and what the uncertainty/limitation is. In summary, there is important information to be provided here.
My major concern with the 2.5 min resolution for LWP is that the information on broken cloud situations (typical for convective conditions) is lost. Are the original data still available? At least it would be good to have the information on cloud fraction in this time period. LWP is likely lower in convective than in stratiform cases if it is averaged over cloud free periods. Maybe the authors can argue that 2.5 min could mimic an atmospheric model grid cell of 1.5 km, assuming cloud advection of 10 m/s.
L 99: “which are not derived from interpolation” what does that mean?
L105: The multiplication is trivial
L 114: It was unclear there the temperatures come from. BTW, did you perform a sensitivity study especially because the retrieval is most uncertain during thick clouds?
L135: You don’t mention what rho_hv is? – why do you have 0.85 here and 0.9 later?
L136: The polarimetric variables are not defined and explained. I think it is a good idea to go not too much into detail here but maybe you can put the threshold values into the table with the sensitivity study in the appendix and only explain the physical principle in section 3.
Section 4.1: The monthly mean IWV/LWP analysis is a bit separated from the rest of the study. Nevertheless, I find it valuable but it needs to be put into context, i.e.. saying how unique this is, as observed LWP climatologies only exist over ocean (MAC-LWP; MODIS) and that large discrepancies exist between them (Lohmann and Neubauer, 2018 https://doi.org/10.5194/acp-18-8807-2018)
L 166: How many samples are necessary to get a robust monthly estimate? Do you apply any thresholding? I ask because in Fig.4, March 2022 has a monthly mean of 500 gm-2. As this is a value often connected already with precip (and retrievals of MWR are not made for that), I doubt that this is a representative monthly mean. As MWR estimates exclude precip events, the data are biased. This also affects the intercomparison with ERA5 and might explain the higher values there. For IWV this bias can be estimated from the GNSS data – it would be interesting to see if these data give the same results on the “precip bias” as found by Henken et al. 2020, https://doi.org/10.3390/rs12071170
L167: Repetition of L99 – still don’t understand “interpolated”
L170: How are warm and cold season defined?
L175: Hersbach, H.L179: Figure captions don’t belong into the main text. The text should be understandable without looking at figures. Figures are there to support your statements. This comment also applies to other parts of the paper,
L186: What does “changes” mean? Differences between both data set? LWP variability?
L188: As the scale issue is of high importance for the study the choice of the considered interval should be motivated earlier.
L193: I wouldn’t call it a jump, but more an exponential decay. The authors might consider to reduce bin size in Fig.5 and similar ones (or increase bin size with increasing time) as the grey bars dominate the plot. This may even allow adding a smoothed line / fit to become more quantitative.
L202: Can you better quantify the impact of the depth of convection using radar rather than distinguishing between warm=shallow and deep=ice. In fact that you have this definition should be made clear in the second question posed in the introduction.
L222 Did I miss a figure on the diurnal cycle? If there is none it might be good to better quantify the effect, e.g. by giving the median times of day when the events happen in the different classes.
L224: What is the “moist” part of the day? It strongly depends which moisture quantity you are looking at. In IWV, the diurnal cycle mostly can’t be detected.
L227: In general, the discussion of the rather different ranges of IWV and LWP needs to be clarified at the beginning. There are many descriptions along the line that “IWV increases less than LWP”. The statement might change if relative units are used as for example a 10 % increase in IWV (e.g. 20 to 22mm) and an 10% LWP increase (eg,. 0.10 to 0.11mm) are compared.
L228 “more ice particle column (cold rain) holds less water vapor. That sounds rather strange. The highest IWV worldwide is found in deep (ice) convection in the tropics. In mid latitudes, I think it mostly depends on how high the tropopause level is…and tropopause is mostly higher in warm airmasses.
L235: onward – see discussion on the different footprint of MWR and GNSS.
L242: what is a “milder” CBH?
L247: more persistent clouds? Do you have information on cloud fraction?
L249: Very sloppy text – describe what you investigate
L258: Are the correlations significant – for r=0.4 the explained variance is only 16%
L270: Was this a cold front passing? For the case studies you should be able to check the synoptic situation.
L274: You mention which data you use in Section 2 so you don’t need to repeat Level2… radar derived precipitation is sufficient. The sentences is again just a figure caption which also makes the section tiring to read.
L280: Before rain starts – it might also be mismatching and shear not seen by the radar. How far down does the radr get without being affected by clutter?
L300: was this a single cell storm?
Fig. 11: Why do you have so much ZDR up into high altitudes before the event?
L308: 0.6 mm/h is rather weak – is this typical for the non-convective class? In fact it would be nice to see how the precipitation rate frequency of occurrence differs between the different categories.
It would be nice to see how the different cases fit into the overall statistics – how are they chosen. The transition from the end of 4.3 to the conclusions is rather abrupt. There is no discussion about what was learned from them. In my major comment 5 I suggest to move them before the statistics.
L333: How do you know that where the liquid is in the cloud? It might well be that there is a lot of supercooled liquid transported aloft by strong updrafts but can you proof it?
L380: Be careful: This is different in relative units. However, here one has to be careful as well because LWP is bounded by zero while IWV is continuous.
Citation: https://doi.org/10.5194/egusphere-2026-4256-RC2
Data sets
WegenerNet 3D Observing System L1b v1.0 Andreas Kvas https://wegenernet.uni-graz.at/
WegenerNet 3D Observing System L2 v1.0 X-band radar Andreas Kvas https://wegenernet.uni-graz.at/
WegenerNet L2 v8.0 climate station data Jürgen Fuchsberger https://wegenernet.uni-graz.at/
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 85 | 54 | 68 | 207 | 69 | 64 |
- HTML: 85
- PDF: 54
- XML: 68
- Total: 207
- BibTeX: 69
- EndNote: 64
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The article of Ghaemi et al. is about the evolution of IWV, LWP and CBH before rain onset
observed in the southeastern Alpine forelands. The article is well written and contains almost all explanations and details.
There are a few related studies, however the article contains new methods of a regime-dependent analysis and new results. In particular the data from the X-band radar are important for recognition of the rain areas. For my taste, the article is a bit
long but the plus point is that the details of the analysis are described.
Generally I recommend a publication in ACP because of the detailed analysis which showed that the evolution of water vapor and cloud droplets before rain is different for convective and non-convective rain events. It would be interesting if the weather models also show such a sharp transition from cloud droplets to rain droplets as the observations indicate for the increases of LWP before rain onsets.
A minor revision is needed.
Minor comments:
1) I think one should mention the manufacturer of the measurement instruments
2) Please discuss LWP during transition from cloud droplets to rain droplets.
3) line 100 ... time steps which ....
maybe it is better to write ... time steps when .... ?
4) Figures 10 and 12 Because of the related colors it is difficult to distinguish the rain gauge from the radar measurements
5) Figure 10 and later : please explain meaning of the horizontal black line
6) Conclusion section is too long.
I would put some parts of the conclusion section into a new section discussion.
7) Just a possible extension idea: it would be great to see the sharp LWP increase before rain onset in a model such as WRF.