{"id":"87bdefa0-900f-44a0-a68d-8be062b9d048","arxiv_id":"2506.14918","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"In a whole-atmosphere model, gravity waves from non-tropospheric sources add at most modest changes to thermospheric winds, and including them does not improve agreement with ICON/MIGHTI satellite winds.","lead":"This study uses a global circulation model to test where atmospheric gravity waves come from and how much they affect winds and temperatures in the upper atmosphere. It finds that waves from the troposphere alone reproduce the observed wind structure, while extra sources at higher altitudes add only local changes.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'upper limit' claim in EXP3 is only an upper limit within the assumed source family: extra-tropospheric sources share the tropospheric phase-speed spectrum (λh=300 km) and are added in phase, so a different spectrum could produce larger effects elsewhere and change the dominance conclusion.","rationale":"The reader identified the same weakest assumption: the extra-tropospheric sources are prescribed with the same spectral shape and in-phase amplification as the tropospheric sources. My stress-test sharpens this into a load-bearing concern by noting that the 'upper limit' interpretation is only valid within that narrow source family. Real secondary waves can have different phase speeds and horizontal wavelengths, which could allow them to deposit momentum at altitudes where the primary spectrum is filtered, potentially producing effects larger than or different from EXP3. This directly affects the paper's central claim that non-tropospheric sources do not improve the global comparison. However, the paper is unusually explicit in Section 6 about these limitations, and the experiments as described support the weaker claim that the specific modeled sources have limited effect. Therefore, the reader's CONDITIONAL verdict remains appropriate, with no adjustment needed. The statistical uncertainty in Figure 11 is a secondary concern because the qualitative conclusion is already bounded by the upper-limit framing; the source prescription is the more fundamental unverified assumption.","tokens_in":25151,"tokens_out":7312,"duration_ms":70435,"concrete_test":"Repeat the EXP3 simulation with the same total added momentum flux but with a different phase-speed spectrum (e.g., a narrower Gaussian with c_w=15 m/s or a spectrum centered on ±20 m/s intrinsic phase speeds) and with a shorter horizontal wavelength (λh=100 km). If the zonally averaged zonal-wind differences in Figure 4k or the ICON/MIGHTI correlation/RMSE in Figure 11 change by more than the EXP3−EXP0 differences (about 0.05 correlation or 5 m/s RMSE), the conclusion that non-tropospheric sources do not improve the comparison is not robust to the assumed spectral shape.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central conclusion that non-tropospheric sources are relatively unimportant (Section 6, items 2 and 6) rests on EXP1–EXP3, in which the extra-tropospheric sources are prescribed as multiples of the tropospheric forcing G_trop with the same Gaussian phase-speed spectrum (Section 2.2.3, Eq. 1) and assumed to amplify the incident harmonics in phase. This is not a general upper bound: because the same spectrum is used, the added harmonics are subject to the same critical-level filtering and dissipation as the primary waves. A real secondary-wave spectrum, which is generated by body forces and typically contains shorter horizontal wavelengths and different intrinsic phase speeds (Vadas et al., 2018), could contain harmonics that survive at altitudes where the 300-km tropospheric spectrum is filtered, depositing momentum in regions that EXP3 leaves untouched. Figures 4 and 5 show the largest effects (up to ±30 m/s) precisely in wind-reversal regions, so an untested spectral shape could plausibly alter the ICON/MIGHTI comparison in the 110–140 km wave-breaking region, where the authors note qualitative improvements. Thus the claim that adding non-tropospheric sources 'does not improve global statistics' is conditional on the assumed spectral shape and phase coherence; the paper's own Section 6 limitation statements acknowledge this, but the abstract and conclusions present it as a stronger statement than the experiments can support.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript uses the CMAT2 general circulation model, with the whole-atmosphere nonlinear gravity wave parameterization and the Medvedev et al. (2023) extension for vertically distributed sources, to isolate the effects of tropospheric versus non-tropospheric gravity wave sources on the middle and upper atmosphere. Four simulations are presented for northern summer solstice 2020: a benchmark with sources at 15 km only (EXP0), sources added at 90 km (EXP1) and 50 km (EXP2) at 10 times the tropospheric forcing, and sources equal to the tropospheric forcing at all levels above 15 km (EXP3). The paper analyzes zonal-mean temperature and wind changes, gravity wave drag and heating/cooling, longitudinal and local-time variability, and compares daytime ICON/MIGHTI winds with EXP0, EXP2, and EXP3. The central reported findings are that EXP3 produces the largest differences (up to ±30 m/s in zonal wind), that localized high-altitude sources have little effect, and that adding non-tropospheric sources changes wave-breaking regions but does not improve the global statistical comparison with MIGHTI wind observations.","tokens_in":25496,"tokens_out":6756,"duration_ms":63181,"significance":"If the experiment family is accepted as a proxy for unresolved middle-atmosphere sources, the paper provides a useful negative result: a whole-atmosphere GCM with tropospheric-only sources captures the basic vertical structure of thermospheric winds as seen by ICON/MIGHTI, and the prescribed extra-tropospheric sources do not improve the skill. The paper is commendably transparent about the speculative nature of the extra-tropospheric source prescription, labels EXP3 as an upper limit, and makes model output available through Zenodo. The ICON comparison is an independent, externally observed test. The principal limitation is that the 'upper limit' and 'does not improve' conclusions are conditional on a narrow family of assumed source spectra and on a daytime, single-season, low-latitude comparison; these qualifications need to be carried into the abstract and conclusions.","major_comments":[{"comment":"The paper repeatedly describes EXP3 as an upper limit or upper estimate of the importance of non-tropospheric sources, but this is an upper limit only within the assumed source family. In Eq. (1) and Section 2.2.3, every non-tropospheric source is prescribed as a multiple of the tropospheric momentum forcing G_trop with the same Gaussian phase-speed spectrum and, by construction, the added harmonics amplify the incident waves in phase. A secondary-wave spectrum generated by body forces typically contains shorter horizontal wavelengths and different intrinsic phase speeds (Vadas et al., 2018), and such harmonics could survive critical-level filtering in altitude regions where the 300-km tropospheric spectrum is filtered. Because Figures 4 and 5 show the largest modeled changes precisely in wind-reversal regions, an untested spectral shape could place momentum deposition elsewhere and change the conclusion. The Section 6 limitations acknowledge this in part, but the abstract and the final 'minor effects' inference present the result more strongly than the experiments support. I recommend rewording the abstract and Section 6 so that the dominance conclusion is explicitly conditional on the assumed source family.","section":"Section 2.2.3, Eq. (1); Sections 4.1 and 6"},{"comment":"The statistical comparison in Figure 11 reports correlation coefficients and RMSE for EXP0, EXP2, and EXP3 without confidence intervals, effective sample sizes, or significance tests. The differences among experiments are small relative to the reported ranges (e.g., zonal correlations 0.5–0.65), so the claim that adding non-tropospheric sources 'does not improve- and even slightly degrades' the comparison cannot be evaluated from the presented statistics. In addition, the comparison uses only daytime MIGHTI observations from June–July 2020 and only low latitudes (0–40°N in Figures 9–10), so the phrase 'global statistical comparison' in the abstract and Section 6 is too strong. The authors should either provide uncertainties and a statement of sample size/independence, or explicitly qualify the conclusions to 'daytime, single-season, low-latitude' comparisons.","section":"Section 5, Figure 11"}],"minor_comments":[{"comment":"The text and caption refer to the localized 90-km source as EXP2, but according to Table 1 EXP1 is the 90-km case and EXP2 is the 50-km case; this labeling inconsistency should be corrected.","section":"Section 4.1, Figure 5 caption"},{"comment":"The Holton and Alexander (1999) reference title contains 'tropospheric convention' and should read 'tropospheric convection.'","section":"Introduction, references"},{"comment":"The sentence 'ICON was launched on 10 October 10 2019' contains a duplicated '10' and should be corrected.","section":"Section 2.4"},{"comment":"There are duplicated words in the running text: 'with with a rate of' in Section 4.1 and 'the the global maximum' in Section 3; these should be fixed.","section":"Sections 3 and 4.1"},{"comment":"The manuscript uses 'extra-tropospheric' and 'non-tropospheric' interchangeably; choosing one term consistently would improve readability.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within scope for JGR-Space Physics and the experiments form a useful sensitivity study. The main risks are the overstatement of the 'upper limit' result and the absence of uncertainty quantification in the ICON comparison. Both are fixable with revisions, and I would support publication if the abstract and conclusions are appropriately qualified and Figure 11 is strengthened with uncertainty information."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a legitimate, carefully hedged sensitivity study rather than a breakthrough, and the paper's own Section 6 does half my job for me. The genuinely new bit is the 3D GCM implementation of vertically distributed non-tropospheric sources and the first ICON/MIGHTI comparison with that setup; the source-prescription idea comes from Medvedev et al. (2023), and the paper says so plainly.\n\nWhat it does well: four clean experiments (EXP0–EXP3) with clearly described differences, honest labeling of EXP3 as an upper limit, and a straightforward satellite matching framework. The conclusion that tropospheric sources reproduce the basic vertical structure of MIGHTI winds, and that adding non-tropospheric sources does not improve global statistics, is supported by the figures. I checked the difference fields and the statistical summary; they are consistent with the text. The paper repeatedly acknowledges that the extra-tropospheric sources are proxies with assumed spectral shape and phase coherence. Data are on Zenodo, which is more than many papers in this area do.\n\nSoft spots, in proportion: the ICON comparison is daytime-only, one season, and correlations/RMSE come without confidence intervals, so the 'does not improve' claim is weaker than it looks. The stress-test concern about the source family is real but not fatal: because all extra-tropospheric runs reuse the tropospheric spectrum at 300 km horizontal wavelength and add in phase, the 'upper limit' label applies only within that family. A secondary-wave spectrum with shorter horizontal scales and different phase speeds could in principle deposit momentum in regions EXP3 leaves untouched. But the authors explicitly flag this in Section 6, so it is a limitation they own, not a hidden flaw. The bigger practical issue is that no code is released, so the validation framework is not fully reproducible; the Zenodo data help but don't fully solve that.\n\nWho this is for: anyone parameterizing gravity waves in whole-atmosphere models, and anyone building satellite-validation pipelines for thermospheric winds. It deserves serious refereeing; the right outcome is minor-to-moderate revision asking for error bars or significance tests on the statistics and a sharper statement of the domain of validity of the upper-limit claim.","headline":"A careful, honest sensitivity study that confirms tropospheric sources dominate in CMAT2, with a real but clearly acknowledged caveat about the assumed source spectrum.","tokens_in":26005,"tokens_out":2009,"would_cite":true,"duration_ms":20855,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper claims that in the CMAT2 general circulation model, gravity waves launched only in the troposphere reproduce the basic vertical structure of ICON/MIGHTI thermospheric winds, while added non-tropospheric sources change little and…","keywords":["atmospheric gravity waves","gravity wave drag","thermospheric winds","ICON/MIGHTI","CMAT2","secondary gravity waves","middle atmosphere vertical coupling","whole-atmosphere gravity wave parameterization"],"falsifier":"A decisive test would target the 110–140 km wave-breaking region, where the model has its largest bias and extra sources their largest effect: if denser ICON/MIGHTI sampling or case studies of well-characterized secondary wave events showed that adding realistic, phase-randomized middle-atmosphere sources systematically removes the troposphere-only bias in a global sense, the central conclusion would be overturned. Momentum flux measurements in the 50–90 km range at or above the assumed tropospheric-equivalent strength would likewise break the upper-limit interpretation.","tokens_in":24942,"feed_emoji":"🌊","tokens_out":11964,"duration_ms":95863,"temperature":0.7,"pith_summary":"This paper asks whether gravity waves generated above the troposphere—secondary waves produced by breaking primary waves and other middle-atmosphere processes—add anything to the upper atmosphere beyond what primary tropospheric waves already provide. Using the CMAT2 general circulation model with a whole-atmosphere nonlinear gravity wave parameterization, the authors compare northern summer solstice runs with tropospheric sources only, with added tenfold sources at 50 and 90 km, and with sources of tropospheric strength at every altitude above 15 km. They find that tropospheric-only waves reproduce the basic vertical structure of ICON/MIGHTI thermospheric winds, that localized upper sources change almost nothing in the zonal mean, and that the all-altitude case—explicitly an upper limit—changes mean zonal winds by up to ±30 m/s without improving the global comparison. If the assumptions hold, current gravity wave drag schemes may already capture the main influence of these waves on the thermosphere.","feed_headline":"±30 m/s: the most extra wave sources can change winds","feed_subtitle":"Even that extreme case does not improve the model's fit to ICON/MIGHTI winds.","key_machinery":"The carrying mechanism is the whole-atmosphere nonlinear gravity wave parameterization of Yiğit et al. (2008), extended by Medvedev et al. (2023) to accept sources at arbitrary heights. At the 15 km source level the scheme prescribes a Gaussian spectrum of vertical momentum flux for 38 harmonics with intrinsic phase speeds of ±2 to ±80 m/s and a representative horizontal wavelength of 300 km; above that level each harmonic's flux evolves under density growth, critical-level filtering, and dissipation from nonlinear interactions, molecular viscosity and thermal conduction, radiative damping, and ion drag. Gravity wave drag is the vertical divergence of the resulting momentum flux, and net heating or cooling combines irreversible and differential terms. The extension expresses unknown non-tropospheric forcing as multiples of the tropospheric forcing $G_{\\mathrm{trop}}$ needed to generate the incident spectrum, keeping the same spectral shape and wavelength and assuming the extra sources amplify the incident harmonics in phase—a construction the authors present as an upper limit.","core_discovery":"The central claim is that in the CMAT2 GCM, primary gravity waves launched near the tropopause are sufficient to reproduce the basic vertical structure of thermospheric horizontal winds as measured by ICON/MIGHTI, and that adding idealized non-tropospheric wave sources does not improve—and even slightly degrades—the global statistical comparison. Localized sources placed at 90 km have negligible thermospheric impact even at ten times tropospheric strength, and 50 km sources produce changes about three times smaller than the largest case. The largest differences, up to ±30 m/s in zonal wind and ±40 K/day in heating and cooling rates, occur when sources of tropospheric strength are placed at every altitude above 15 km; since no observational evidence supports such persistent strong middle-atmosphere generation, the authors present this as an upper bound on the dynamical importance of missing secondary sources. The paper also establishes that gravity wave drag is longitudinally uniform in the lower thermosphere but localized in the upper thermosphere, that all-height sources raise longitudinal wind variability only up to about 150 km before ion-neutral coupling dominates, and that wave effects peak during the day in the upper thermosphere but at night in the lower thermosphere.","pith_inferences":["A run the paper does not perform—extra sources with random phases rather than in-phase amplification—would sharpen the upper-limit claim; out-of-phase sources could partly cancel the incident harmonics, making the true impact of secondary waves smaller still.","Because the upper-thermosphere drag is localized in longitude while ion drag erases wind variability above about 150 km, replacing the empirical ionosphere model with a fully interactive one could shift the height at which ion-neutral coupling overtakes wave-driven variability.","The correlation-and-RMSE comparison against MIGHTI winds is directly transferable to other whole-atmosphere models, which would test whether tropospheric dominance of thermospheric wind structure is a property of the real atmosphere or of this particular drag scheme and its 300 km reference wavelength.","The paper's own construction implies that if persistent middle-atmosphere wave generation is ever found, its clearest signature in the thermosphere would be longitudinal rather than zonal-mean structure below about 150 km—a target that coordinated airglow and radar campaigns could seek."],"forward_implications":["Global models that launch gravity waves only near the tropopause may already capture most of the gravity wave influence on thermospheric circulation, since the troposphere-only run reproduces the observed basic wind structure.","Secondary-wave sources near the mesopause are unlikely to matter for the zonally averaged thermosphere: even a source ten times stronger than the troposphere at 90 km changed the mean fields negligibly.","The all-height source run bounds the possible dynamical effect of missing middle-atmosphere waves at about ±30 m/s in zonal wind and ±40 K/day in heating and cooling; any real secondary source weaker or less coherent than the assumed one would produce smaller changes.","Model skill depends on wind component and altitude: correlations with ICON/MIGHTI are 0.5–0.65 for zonal winds and 0.25–0.45 for meridional winds, with better agreement in the upper thermosphere than in the lower thermosphere, where an eastward bias places the wind reversal too high.","If future observations quantify middle-atmosphere wave generation, its effects could be folded into existing drag parameterizations by retuning, since the extra sources mostly amplify what the tropospheric waves already do."],"supporting_citations":[{"why":"Supplies the nonlinear whole-atmosphere gravity wave parameterization that propagates the 38-harmonic spectrum, computes dissipation, drag, and heating/cooling.","marker":"Yiğit et al. (2008)"},{"why":"Provides the framework for expressing non-tropospheric sources as multiples of the tropospheric forcing G_trop; the EXP1–EXP3 design directly follows this approach.","marker":"Medvedev et al. (2023)"},{"why":"Describes the CMAT2 GCM—its dynamical core, vertical extent, lower-boundary forcing, and the upper-atmosphere physics used in all experiments.","marker":"Yiğit et al. (2009)"},{"why":"Defines the ICON mission and data set whose MIGHTI winds are the observational benchmark for the comparison.","marker":"Immel et al. (2018)"},{"why":"Documents the MIGHTI instrument and its airglow wind measurements used in the validation.","marker":"Englert et al. (2017)"},{"why":"Describes the MIGHTI wind retrieval algorithm that produces the Level 5 wind profiles compared with the model.","marker":"Harding et al. (2017)"},{"why":"Gives the theory of gravity wave heating and cooling that the parameterization uses for thermal effects.","marker":"Medvedev & Klaassen (2003)"}],"fun_headline_variants":["Non-tropospheric wave sources don't improve ICON wind fits","Even ±30 m/s wave effects fail to beat tropospheric-only model","Tropospheric waves capture thermospheric winds; extra sources don't help","Adding wave sources above troposphere yields no ICON/MIGHTI gain"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that missing upper-atmosphere wave sources can be represented as in-phase, same-spectrum multiples of the tropospheric forcing; if real secondary waves are weaker, spectrally different, or out of phase with the primary waves, the modeled effects and the conclusion that tropospheric sources dominate could be wrong.","fun_headline_variants_meta":{"raw":{"variants":["Non-tropospheric wave sources don't improve ICON wind fits","Even ±30 m/s wave effects fail to beat tropospheric-only model","Tropospheric waves capture thermospheric winds; extra sources don't help","Adding wave sources above troposphere yields no ICON/MIGHTI gain"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000841,"raw_usage":{"total_tokens":3731,"prompt_tokens":1077,"completion_tokens":2654,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":693,"completion_tokens_details":{"reasoning_tokens":2573}},"tokens_in":693,"tokens_out":2654,"duration_ms":17389,"temperature":1.0,"reasoning_tokens":2573,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T00:08:59.954399+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive test would target the 110–140 km wave-breaking region, where the model has its largest bias and extra sources their largest effect: if denser ICON/MIGHTI sampling or case studies of well-characterized secondary wave events showed that adding realistic, phase-randomized middle-atmosphere sources systematically removes the troposphere-only bias in a global sense, the central conclusion would be overturned. Momentum flux measurements in the 50–90 km range at or above the assumed tropospheric-equivalent strength would likewise break the upper-limit interpretation.","supporting_citations":[],"review_version":1}