the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 29 Jul 2024
Sensitivity of tropical orographic precipitation to wind speed with implications for future projections
Abstract. Some of the rainiest regions on Earth lie upstream of tropical mountains, where the interaction of prevailing winds with orography produces frequent precipitating convection. Yet, the response of tropical orographic precipitation to the large-scale wind and temperature variations induced by anthropogenic climate change remains largely unconstrained. Here, we quantify the sensitivity of tropical orographic precipitation to background cross-slope wind using theory, idealized simulations, and observations. We build on a recently developed theoretical framework that predicts enhanced seasonal-mean convective precipitation in response to cooling and moistening of the lower free-troposphere by stationary orographic gravity waves. Using this framework and convection-permitting simulations, we show that higher cross-slope wind speeds deepen the penetration of the cool and moist gravity wave perturbation upstream of orography, resulting in a mean rainfall increase of 20–30 % per m s−1 increase in cross-slope wind speed. Additionally, we show that orographic precipitation in five tropical regions exhibits a similar dependence on changes in cross-slope wind at both seasonal and daily timescales. Given next-century changes in large-scale winds around tropical orography projected by global climate models, this strong scaling rate implies wind-induced changes in some of Earth's rainiest regions that are comparable with any produced directly by increases in global mean temperature and humidity.
- Preprint
(3010 KB) - Metadata XML
-
Supplement
(4380 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-2180', Anonymous Referee #1, 05 Sep 2024
In this paper, the authors apply a theoretical framework they developed in previous work to evaluate the sensitivity of tropical orographic precipitation to changes in background wind speed. The theory predicts that precipitation increases with increasing wind speed by 20-30% per m/s, and find good agreement with this prediction in a numerical simulation using the Weather Research and Forecasting (WRF) model. They also find qualitative agreement with their prediction in observations.
The paper is well written. The result that tropical orographic precipitation doesn’t scale in proportion to the horizontal wind speed is interesting and has important implications for climate change impacts given possible future changes in the tropical circulation. The agreement between theory, models, and observations is imperfect but mostly convincing.
The main room for improvement, in my view, relates not to the analysis per se but to the way it’s explained. I like the authors’ approach of combining linear mountain-wave theory with a convective closure to study orographic enhancement of precipitation over tropical mountain ranges like the Western Ghats. However, aside from the two previous studies published by the same authors, this is a novel approach, and one that I suspect many readers will not be very familiar with. I think the authors should do more to explain the physical basis for their approach, including, for example, why they assume a dry static stability for the linear mountain wave solution, and why the convective closure of Ahmed (2022) makes sense in this context. I also think the authors could hold the reader’s hand a bit more on some of the more technical parts of the paper, especially when using their equations to draw physical conclusions. I note some specific areas of confusion for me below.
Sections 2 and 3: I may have missed it, but I couldn’t find any mention of the vertical humidity profile in either the theory or numerical modeling. Isn’t that important given that the lower tropospheric humidity perturbation is based on the vertical humidity gradient (Eq. 2)?
Line 93 and Fig. 1A: I understand why the depth of ascent would increase from a change in the vertical mountain wavelength, but I’m surprised there’s not an increase in the *magnitude* of eta (and thus T’) from an increase in U, given the lower boundary condition w=u*dh/dx (Fig. 1b). Later the authors note that the ratio of w/U increases in response to an increase in U, which I would expect to further amplify the T perturbation. What am I missing?
Section 3.3: I found this discussion of the role of theta_e’ to be confusing. The basic argument, as far as I can tell, is that there’s a theta_e perturbation that contributes to precipitation enhancement, and that this is tied to the greater vertical penetration of the mountain wave when U increases, since an increase in the vertical wavelength causes w/U to increase above the surface, and this in turn increases the vertical advection of theta_e. First, I didn’t find the evidence in Fig. 3c to be especially convincing given the large magnitude of the residual term. Also, shouldn’t the same increase in vertical advection also apply to T and q? In reality, the WRF simulations show no contribution from q' and a relatively modest contribution from T'. I think this disagreement between the theory and WRF simulations should receive more attention.
In summary, I think this is an important contribution, but I wish it provided a clearer physical explanation for why tropical orographic precipitation is so sensitive to perturbations in U. The simple theory that involves q and T seems relatively straightforward (aside from assumptions about the vertical profile of q). However, while this theory gives an accurate prediction for the precipitation increase simulated by WRF, it seems to do so for the wrong reasons, since WRF shows a negligible contribution from q’ and a large contribution from theta_e' (Fig. 2c). I think the authors should better explain how their theory can be reconciled with the WRF results.
Citation: https://doi.org/10.5194/egusphere-2024-2180-RC1 -
RC2: 'Comment on egusphere-2024-2180', Anonymous Referee #2, 16 Sep 2024
Summary: in this manuscript, the authors apply their new theory of mechanically forced orographic convection (Nicolas and Boos 2022) to evaluate the sensitivity of orographic precipitation to background wind speed. The theory is augmented by convection-permitting WRF simulations and observations from selected mountainous tropical regions. In the end, the authors conclude that small increases in winds in current and future climates may give rise to disproportionate increases in orographic precipitation.
I found the manuscript to be mostly well written and scientifically rigorous. The figures are clear but also very densely packed with information, and the authors dutifully explain each detail in their very long figure captions. In terms of the methodology, the theoretical model is already vetted and on solid footing and the complementary simulations and observational analyses add substantial value. The results are scientifically interesting and mostly convincing. Overall, this is a strong manuscript, and my feedback consists of one major comment along with a number of minor and/or technical comments.
Major comment
- The physical explanation for the enhanced \theta_e over the windward slope in the U=12 m/s case in Figs. 3A-B is not convincing to me. The authors attribute this enhancement to increased vertical advection of \theta_e in Fig. 3C. They argue that this increase must be associated with a deeper mountain wave, which is evident at upper levels (well, above 800 hPa) in Fig. 1B. However, the clear enhancement in \theta_e in Fig. 3A is confined to the 900-1000 hPa layer, not above 800 hPa. Moreover, some of this enhancement reaches all the way down to the surface in Fig. 3A, where the authors concede that the mountain wave cannot explain the enhancement. I’m skeptical of this interpretation because, although vertical advection is stronger at larger U, so is horizontal advection, which tends to reduce \theta_e. Also, if the air crosses the mountain in both cases (which it does), and \theta_e is in fact conserved, the \theta_e over the mountain top should be the same in both cases even though the terms in the \theta_e budget may differ in magnitude. Therefore, my sense is that more analysis is needed to properly interpret this enhancement, which is important because it is largely responsible for the disproportionate increase in P’ with U. One possibility is that relative flow deceleration caused by the mountain wave over the windward slope (relative to the background wind speed U) may be smaller in the U=12 m/s case. If so, the near-surface W would be disproportionately enhanced. Also, because this is an orographic precipitation problem, one can never rule out cloud microphysics as a driver of changes. Is it possible that the U=12 m/s case sees a larger share of the precipitation evaporating over the lee slope, as opposed to over the windward slope? Weak evidence of this is seen in the slight downstream shift of the precipitation peak for the U=12 m/s simulation in Figs. 3A-B. Such a change could reduce evaporative cooling over the windward slope, thereby enhancing boundary-layer \theta_e there.
Minor comments:
- Abstract, L5-6: this sentence is confusing and misleading. First, it is said that precipitation is “enhanced”, but the term “enhancement” requires a reference state for comparison. What is that reference state? This context is necessary in the current study, where the term “enhancement” could refer to (i) orographic enhancement relative to surrounding regions, (ii) enhancement relative to cases with weaker or stronger winds, or (iii) enhancement relative to the current climate. Moreover, the description of the model seems off the mark. It’s a model for orographic precipitation, right? So why not describe it as such?
- L29 and L35-36: I agree that decades of research have facilitated progress in orographic “rainfall” (or “precipitation”). But here the authors only refer to studies of tropical orographic precipitation, and only a very small sampling of those. My concern is that the authors preemptively limit the scope of their literature review to a narrow topic. Doing so helps to shorten the discussion, but it also precludes the authors from benefiting from a huge tranche of relevant literature. There are numerous studies of midlatitude orographic precipitation, ranging from theoretical, to numerical, to observational, many with ideas that clearly transfer to the current study. For example, using idealized simulations, Colle (JAS, 2004) found 20 years ago that stronger winds enhance orographic precipitation by deepening the mountain wave. His result seems quite relevant to the present study. Of course, the goal is not to cite as many studies as possible, but to use the existing literature to help frame and inform the scientific investigation.
- L72: By “adiabatic”, do you mean dry or moist adiabatic?
- L85-86: “Thus, we expect Pm << Pa”: Where does this inference come from? It is not clear from eqn (3) nor the text of this sentence.
- L27: More information on these simulations would be helpful. Do they ever reach a quasi-steady state? How long does it take? If 250 days are required (as the text suggests), that would seem rather excessive.
- L132-133: Here the authors do cite a midlatitude orographic precipitation study but the physical reasoning is flawed. Yes, hydrometeor drift is significant, but Smith and Barstad (JAS, 2004) were exclusively interested in stratiform precipitation. They were assuming that part of the cause for the downstream shift was slow falling (~1 m/s) ice and snow being carried downwind. This effect is weaker for the convective precipitation studied here, where fall speeds are closer to 10 m/s. Therefore, I think the "10 km" estimate is overly generous.
- L134: Does the theory even consider multiple levels? This isn't apparent in equation (1) or the Appendix. Anyway, shouldn't the precipitation amount scale with low-level specific humidity? If not, the theory would have little hope of capturing the sensitivity of orographic precipitation with respect to temperature variations in a changing climate.
- L158: What is meant by “equally”? The two contributions don’t seem precisely equal to me, so this is misleading.
- L160-161: I don't follow this reasoning. You just said that qL' has minimal impact but θ_e does, so how can one compensate for the other? And why is there an “absence of free-tropospheric moistening”? Doesn't the mountain wave ascent produce this moistening?
- Equation (4): I don’t think it’s fair to claim that the \theta_e budget is equal to the highly simplified (4) without stating which simplifying assumptions were made.
- L174: “Little accuracy is lost…”: is this shown somewhere? Given my major comment, I would be interested to see the evidence behind this statement.
- L175 and L141: Why do you average 4000-2000 km upstream of the mountain here but over a different region (4000-2500 km upstream) on L141? Is there a justification for this inconsistency?
- L183: I’m not so sure that a \theta_e change on the order of 0.1 K in can be called a “sharp” increase.
- L206-207: Do all reanalyses used in this paper really use a “model prior to 1979”? I thought ERA5 used a modern numerical model. Or maybe I am not reading this sentence correctly.
- L225: Measured how? Using gauges, as you claimed earlier? Surely the gauges don't align perfectly with the line shown in Fig. 4B. What data set are you using, and how are you analyzing these data? In this section you use a variety of data sets, and the reader needs some help to keep track as you weave through different analyses.
- L229-230: OK, but what about P'? Are you saying here that P’ should be a function of P0? That's not what was assumed in the linear model.
- L259-260: “assuming a 3.5 K warming”: measured where? At the surface???
- L263-269: Related to this text, another midlatitude study is Kirshbaum and Smith (QJ, 2008), who found < CC scaling w.r.t. temperature changes in simulations designed to test this exact question. Their result and physical explanation is very similar to the one posed by O'Gorman and Schneider (2009) for the climate at large, but with a slightly different mathematical formulation.
- L288-291: Another limitation that should be acknowledged is the fact that orographic convection is poorly resolved at Δ= 3 km (Kirshbaum, JAS 2020). The model effective resolution is no better than 20 km, and orographic cells typically have horizontal scales of 1-10 km. Also, for larger temperature deviations one must also consider cloud-microphysical changes (e.g., less ice-phase microphysics) that could offset the projected enhancements in regions experiencing significant warming.
Citation: https://doi.org/10.5194/egusphere-2024-2180-RC2 -
AC1: 'Preliminary response to reviewers - egusphere-2024-2180', Quentin Nicolas, 08 Oct 2024
We thank both reviewers for their thorough analysis of the manuscript and their thoughtful comments. Our manuscript argues for a large sensitivity of mechanically forced tropical orographic precipitation to changes in background wind speed, based on three lines of evidence: a theoretical model, two convection-permitting simulations, and a set of observations. Of these three lines of evidence, both reviewers expressed concerns about our explanation of the behavior in the convection-permitting simulations. Specifically, their most important comments relate to our analysis of the changes in boundary-layer equivalent potential temperature induced by changes in the background wind speed . Here we summarize the revisions we plan to make in response to these comments. These revisions will provide a clearer, more detailed explanation of the deepening of the mountain wave and its consequences, together with a more complete discussion of alternative hypotheses for the large sensitivities that we document.
Our argument for the increase in θeB’ with increased U can be summarized as follows: horizontal gradients in θeB’ adjust so that horizontal advection (by the total wind u, close to the background wind U) balances vertical advection (produced by vertical velocity w in the mountain wave) and diabatic sources. When the background wind U is increased, we argue that vertical advection increases faster than U over the mountain, due to the vertical stretching of the mountain wave; while diabatic sources change at approximately the same rate as U (Fig. 3C). This leads to a positive anomaly in the horizontal gradient of θeB’ over the upwind slope in response to increased U, hence a positive anomaly in θeB’ (obtained through the lateral boundary condition θeB’≃0 over the upstream ocean, since SST does not change).
The reviewers’ comments concern several points in this argument:
- Our explanation of the change in the three-way balance between horizontal advection, vertical advection, and diabatic sources (reviewer 2). The reviewer argues that although vertical advection increases, so does horizontal advection. Our point is precisely that horizontal advection increases to balance the increase in vertical advection (itself explained by mountain wave dynamics), and that this in turn leads to an increase in dθeB’/dx. We plan on modifying the associated paragraph lines 176-183 to make this point clearer; specifically, we will stress that vertical advection increases faster than the background wind speed U, forcing an increase in dθeB’/dx to maintain the balance. Our argument concerning the sensitivity of the ratio w/U to wind speed may have been too subtle, and we will modify the text to better convey the relevant dynamics.
- The fact that equation 5 is only approximate, leading Fig 3C to show a large residual in some places (reviewer 1). The important point here is that the residual is small upstream of the ridge top, where our analysis is focused. We plan on showing averages of each term in Fig 3B-D over the upwind slope as small bar plots on the right of the figure to highlight the magnitude of each term in the region where it matters. This will make more readily apparent the fact that the residual in Fig 3C is small there. We will further clarify the processes responsible for the residual (e.g., changes in horizontal wind perturbations and nonuniformities in the vertical θe gradient).
- The fact that the increase in vertical advection is due to a deeper mountain wave (reviewer 2). The reviewer argues that the deepening of the mountain wave illustrated through the T anomaly in Fig. 1B is not observed below 900 hPa, while the increase in θeB’ in Fig. 3A is confined to the boundary layer. In a related point, reviewer 1 argues that the same reasoning should apply to lower-tropospheric temperature and moisture anomalies, whereas our manuscript showed that q’ changes little there. In order to properly address these points, we will better illustrate the changes in the mountain wave velocity field and its consequences for the various thermodynamic quantities. Specifically, we will show that the deepening of the mountain wave simultaneously increases anomalies in lower-free-tropospheric T and θeB, and we will add a figure illustrating this deepening of the mountain wave in the vertical velocity field (the temperature anomalies shown in Fig. 1B do not reveal this deepening in the boundary layer, as the stratification is dry neutral there). We will further discuss why the lower-free-tropospheric moisture anomaly does not change with increased U in simulations (whereas the theory predicts that it should).
- Reviewer 2 states that alternative explanations based on diabatic sources (specifically rain evaporation in the boundary layer) should be explored. We argued based on Fig. 3C that changes in diabatic sources of θe cannot explain the observed increase in θeB’. We will modify Figure 3 to include an average of changes in this term in the region of interest (the upwind slope) and show that it is much smaller than the change in vertical advection. We will stress that changes in diabatic sources are unlikely to explain the increase in θeB’ with increased U.
In summary, our changes will make a clearer connection between the processes at play in sections 2 and 3. We will explain why the simulations do not feature the increased lower-free-tropospheric moisture anomaly predicted by the theoretical framework. We will make our explanation of changes in the θe budget clearer by modifying Figure 3 and adding a figure showing the deepening of the mountain wave into the boundary layer in the vertical velocity field.
We also note that reviewer 2 raised two points in their major comment with which we disagree. The first one is that if θe is conserved, θeB should not change when the background wind is changed. In fact, θe is only expected to be conserved along streamlines, and the 900 hPa level that defines the top of our boundary layer is not a streamline; hence a boundary-layer average of θe can change even if θe is conserved (this is in fact what we are arguing for; there are diabatic sources in our case, but their change is less important than changes in vertical velocity, i.e. in the shape of streamlines). The second point is that the ratio of near-surface w to the background wind U may be larger at higher U due to a weaker relative deceleration of the near-surface flow. The simulations show that this is not the case (the near-surface w/U exhibits little change when U is increased from 10 m/s to 12 m/s).
The reviewers also raised several minor points, all of which we will address through relatively minor changes to the manuscript. Among these, we will:
- Give more details and explanations in the presentation of our theoretical model (reviewer 1).
- Explain why the vertical moisture profile is important in setting the orographic precipitation perturbation P’, but not its relative sensitivity to changes in wind, d ln(P’)/dU (reviewer 1).
- Add a paragraph in the introduction to more comprehensively summarize past work that has examined related questions for midlatitude orographic precipitation (reviewer 2).
Citation: https://doi.org/10.5194/egusphere-2024-2180-AC1
Data sets
Data for Nicolas & Boos, "Sensitivity of tropical orographic precipitation to wind speed with implications for future projections" Quentin Nicolas https://doi.org/10.5281/zenodo.11479598
Model code and software
qnicolas/windSensitivity: Submission stage for Nicolas & Boos, "Sensitivity of tropical orographic precipitation to wind speed with implications for future projections" Quentin Nicolas https://doi.org/10.5281/zenodo.12735240
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 208 | 90 | 92 | 390 | 51 | 8 | 10 |
- HTML: 208
- PDF: 90
- XML: 92
- Total: 390
- Supplement: 51
- BibTeX: 8
- EndNote: 10
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1