| 10 Mar 2026
Diurnal cycles of cloud and rainfall over North–East Queensland during the coral bleaching season
Abstract. Mass coral bleaching events are becoming increasingly frequent over the Great Barrier Reef, posing a critical risk to Australia's marine ecosystems and the broader global ocean environment. These events are primarily driven by anomalously warm water temperatures, but their severity is strongly influenced by local cloud cover, which controls the amount of solar radiation reaching the ocean surface (including ultra–violet radiation which exacerbates bleaching). This study presents a characterization of the cloud and rainfall diurnal cycles over north–east Queensland during the coral bleaching season, providing a foundational step to untangling the complex relationships between clouds, rainfall, local–scale processes and the surface energy budget for this climate–sensitive region. Leveraging high–resolution Himawari–8 satellite brightness temperature data, C–band radar observations, and BARRA–R2 regional reanalysis, a multi–year analysis is conducted across three representative zones: coastal land, coastal ocean and open ocean. Results show that diurnal cycles vary distinctly by region and are strongly modulated by prevailing wind regimes. In general, westerly regimes are associated with clear skies and stronger daytime heating over the coastal ocean with enhanced convection over coastal land. In contrast, the frequently observed southeasterlies lead to relatively weaker development over the land and ocean. Cloud and rainfall maxima exhibit out–of–phase behavior between land and ocean, with rainfall often preceding cold cloud tops, indicative that cold brightness temperatures frequently correspond to decaying anvils rather than active convection. Latitudinal and topographic differences contribute to more intense convection near Cairns than Townsville. Variations in inland versus offshore propagation speeds further highlight regional complexity. Our findings emphasize the necessity for high–resolution simulations to advance understanding of convection initiation and propagation mechanisms.
This paper investigates the diurnal cycle in north-east Queensland under different regimes of coastal wind direction, with a view to understanding the variability of quantities that can affect coral bleaching, for example cloudiness (due to its impact on insolation).
I recommend that this manuscript be improved with major revisions before publication. My most significant comments are two-fold. i) The authors either need to provide stronger justification regarding the definition of the wind regimes or, more likely, do some further experimentation, as I suggest below. ii) The main thrust of the paper is presented as being the impact on coral bleaching, but the results are broadly descriptive and do not have any particular focus on coral reef locations (for example, there is a lot of discussion of inland propagation of convection).
Major comments
1). Wind regimes:
(a) The paper uses two sources of data for winds: the BARRA-R2 reanalysis and AWS observations. It is not clear why the paper mainly uses BARRA-R2 then switches to the AWS data for Fig 8. Why not use the observations to derive the wind regimes in the first place? I assume this is because there is no AWS at Willis Island (and there could also be missing data), but section 2 should clearly justify the choices of data sets.
(b) L99 says the reanalysis is hourly. For Jan-Apr 1979-2024 this would give 132,768 data points for each site, but Fig 4 has an order of magnitude more than this (1,168,776 if you add up the number of samples across regimes). Please clarify.
(c) I believe you define the wind regimes hourly, so successive hours could be in different regimes. Have you tried categorizing the regime for the entire day based on, say, the daily mean winds or the winds at a particular time of day, or averaged over a particular few hours? When you compute composite diurnal cycles for each regime (Figs 5-8) you must be piecing together different parts of different days, rather than full diurnal cycles. Does this affect the results? It may make the precip in Fig 7 less noisy, for example.
(d) You show in Fig 8 that the wind direction has a diurnal cycle. If you are selecting the wind regime hourly, certain regimes are more likely to exist at certain times of day. Therefore, there may be an aliasing effect when you produce composite diurnal cycles. For example, theta-e presumably has a diurnal cycle, especially near the ground (but I think Fig 4 is averaged over all times of day?). So are the differences between regimes really due to the wind patterns, or just a selective bias in the time of day that goes into each composite? This is another argument for assigning a single regime to each entire day.
(e) Similarly, you show that different regimes are more likely to occur at different times of the season (Fig 1b-d). You say (L184) that theta-e profiles are shown relative to the "seasonal composite mean". Is this a single mean profile over all of Jan-Apr? What if you show the anomalies relative to the corresponding month's mean? Or even use a smoothed mean seasonal cycle that varies at daily time resolution? This would remove any aliasing of the seasonal cycle onto your regime composites.
2). Much of this work is reminiscent of Dao et al. (2025; 10.1002/qj.70027) but surprisingly this is not referenced. The introduction should cite this paper and clearly explain how the present study differs from or builds upon it.
3).
(a) Throughout the paper, I found the discussion of the figures (particularly 4-7) to be very descriptive, going through each wind regime in turn and describing every feature of the plots. A better discussion would focus on the story that the authors want to tell. For example: In the discussion of Fig 7 (section 3.4) a lot is said about inland propagation, but the title of the paper and the motivation in the introduction relate to coral bleaching, so a key point is cloudiness over the reefs. Therefore, is the inland propagation really relevant to the aims of the paper?
(b) You return to the point about coral bleaching in the final paragraph of the paper, but very little has been said about this previously. Could you shift the focus of the results and discussion to the locations of the reefs? Do you have SST data that you could composite by regime? That would demonstrate whether these regimes really do have an impact.
4) There are occasional speculative comments which should be better evidenced or else removed:
(a) L222-223: "likely reflecting the reduced influence of orographic uplift and elevated terrain heating compared to Cairns"
(b) L327: It is not very encouraging that you admit some of the features of your concluding schematic diagram are "speculative". You say this is a limitation of observations, but can you not use BARRA to investigate processes that are not observed? There is nothing about gravity waves in the paper so this should not be included in Fig 9 (although you could make references to other papers on gravity waves in the text as a possible explanation for the offshore convection).
(c) The only mentions of orographic uplift in your results section are also speculative, but you use this as an explanation in the discussion (L336). This should also be removed unless you can make some relevant arguments. For example, if you mark orography on Fig 7 somehow, could you deduce an impact of uplift?
5) You argue that there is a lag between precipitation and BT. This is a well-known phenomenon but I struggle to see it in Fig 6. Please be quantitative about the times of day of minimum BT and maximum precip, to make the lags clear (L225, L249).
Minor comments
L19: "influences" -> "influence"
L28: "interact with" -> "interacts with" to be consistent with "accounts for" earlier in the sentence.
L33: Cite something more recent here regarding state-of-the-art models' ability to represent the diurnal cycle.
Fig 1: The Cairns radar and AWS are not easy to see. Suggest drawing them on top of the Hovmoller transect.
L150: Replace "composite" with "histograms of"?
L154: I'm not convinced that the increase in SE through the months is delayed at Cairns and Willis Island compared with Townsville. The three sites look very similar in this respect.
L154: Replace "stable" with "similar"?
L158-159: Remove "on average" and "on a seasonal mean basis"
L160: "variation" -> "variation in"
L162: "[NE]'s decline coincides with the strengthening south-easterlies, indicating a shift in dominant flow". But SE is always the dominant regime, in all months shown.
Fig 2: Gray "hatching" (actually shading) is too faint.
L164-170: The SW and NW regimes are so uncommon, do you really need this paragraph?
Fig 3: Caption says (a) shows the wind regimes, but you actually divide up the direction into 16 bins on the plot, whereas there are only 4 regimes. I take it the polar plots are histograms. What are the tick labels on the radial axes? Are they percentages? I would use the same tick labels on each plot and use integer values (rounding errors make it look like the ticks are unevenly spaced). There is no explanation of the colors. Have you ordered the values within each histogram bin by speed and colored them accordingly? In panels b-d, consider changing the color for either NE or SW, as red and green together should be avoided.
L177-181: Move the explanation of theta-e to section 2.
L179: Remove "typically".
L195: What is meant by "energetically favorable air"?
L214: More useful to quote the wavelength than the Himawari channel number.
L214: How do you define rainfall frequency? Rainfall rate above some threshold?
L215, L308, L313 and Fig 9 caption: Fix figure references.
Fig 5: Suggest subtracting the daily mean (as a function of p) from each panel to make the diurnal variations clearer.
L240: "BTs peak": better to say something like "BTs are largest" since you invert the y-axis (which I agree is sensible), so the "peak" actually appears as a trough.
Fig 6: Can remove "hourly means". Again, quote wavelength of satellite channel and define rainfall frequency.
Fig 7: Not easy to see the precip contours on top of the BT. Consider removing the filled BT contours and just show the contour lines for 273 and 263K, since these are what you refer to in the text anyway. Then consider showing the precip as filled contours. It's not always clear what feature you think is propagating at the speed shown by the propagation lines, e.g. the 30 m s-1 line in (b). In the caption, replace "diagram" with "diagrams". Remove "mean daily"?
L318: "prior tropical studies" - which ones?
Fig 8: How is average direction defined? (e.g., What is the average of 1deg and 359deg? Arithmetically it is 180deg but common sense dictates it is 0deg.) Or do you average the zonal and meridional components, then show the speed and direction of the result? This is what I would recommend. At 15LT at Cairns, (c) indicates the average is roughly easterly but (b) shows the wind is mostly from the south-east at this time.
Fig 9:
(a) Are the five images supposed to represent equal time periods, or particular times? This is not clear from the vertical time axis.
(b) The theta-e arrow is too faint.
(c) Consider indicating SB speed by the length of the arrow. Using thickness looks like the SB depth is changing.
L462: This is the discussion version of the paper, not the published version in WCD (2026).