Quasi-biennial Oscillation Research Articles

Teleconnection of the Quasi-biennial oscillation with boreal winter surface climate in Eurasia and North America

An improved understanding of dynamical coupling from regional to global scales via tropospheric or stratospheric region can be helpful in improving seasonal forecasts for a given region of interest. Here we investigate dynamical coupling between the equatorial stratospheric Quasi-biennial oscillation and the boreal winter surface climate of the Northern Hemisphere mid and high latitudes using 42 years of data (1979–2020). For neutral El Niño Southern Oscillation periods, the Quasi-biennial oscillation westerly phase at 70 hPa favors high sea level pressure in the polar region, colder conditions and deeper snow over Eurasia and North America, and the opposite effects for the easterly phase. When Quasi-biennial oscillation anomalies arrive near the tropopause, it is observed that planetary wave activity is enhanced towards to extratropical region during westerly phase and reduced during easterly phase. This teleconnection pathway via the upper troposphere and lower stratosphere to the high latitude surface is independent of the “stratospheric pathway” (Holton-Tan mechanism). Diagnosis of this pathway can help to improve understanding of sub-seasonal to seasonal variations, and long-range forecasting over Eurasia and North America.

Quasi-biennial oscillation modulation of stratospheric water vapour in the Asian monsoon

Abstract. The Asian monsoon (AM) plays a key role in the transport of water vapour to the lower stratosphere and contributes significantly to the wet phase of the annual global stratospheric water vapour cycle. Although it is known that the quasi-biennial oscillation (QBO) is one of the main drivers of the interannual variability in the AM water vapour, the physical mechanisms responsible for this variability remain unclear. Here we have used daily microwave limb sounder data for the period 2005–2020 to characterize the QBO signature on the lower stratosphere AM water vapour during the boreal summer. We show that the QBO has the strongest impact during August, when QBO westerly minus QBO easterly differences may reach 1 ppmv at 100 hPa, although a significant signature is also observed during July. We find that the region whose temperature controls the QBO signal on water vapour over the AM differs between July and August. In July, when the key region is over the tropical Indian Ocean, the QBO modulation of the AM water vapour occurs in phase with the signal over the Equator, whereas in August, when the key region is at the subtropics, over the southern edge of the monsoon, the signal over the AM is opposite to that over the Equator. Our results reveal that the QBO signal on the temperature on the south side of the AM anticyclone, which ultimately has an impact on AM water vapour, is, in turn, modulated by the QBO impact on tropical clouds. Thus, we find that the QBO signature on clouds over the eastern Indian Ocean gives rise to Rossby wave trains that produce variations in the circulation over the southern side of the AM anticyclone such that weaker anticyclone over this region generates an increase in water vapour, and vice versa.

Tree-ring widths of Pinus tabulaeformis Carr reveal variability of winter half-year precipitation on the north-south transition zone in central China over the past 220 years

Long-term, high-resolution regional drought records contribute to understanding the impacts of drought on environmental and social systems in central China. Here, we develop a regional tree-ring width chronology of Pinus tabulaeformis Carr from the northern slope of Funiu Mountains on the north-south transition zone in central China. Monthly correlation analyses showed that temperature and humidity in current May and June are main limiting factors on tree growth. Despite that, the highest correlation with tree growth was found to be precipitation from previous December to current June (PreDJ, 0.718, p < 0.001), which was chosen for reconstruction. The reconstructed PreDJ revealed six drought periods and five wet periods over the past 220 years, and the recent dry spell would likely to continue. Spectral analyses indicated that the reconstructed PreDJ was closely related to the El Nino-Southern Oscillation (ENSO, 2-7a) and 35a climatic oscillation of Bruckner, and was also affected by the Quasi-Biennial Oscillation (QBO). Wavelet analyses showed that the quasi-cycle of 2-7a persisted over the past 220 years and strengthened after the 1980s, and the QBO signals appeared from the 1860s to 1970s and wear off thereafter, and 35a cycle only appeared during 1820–1920. Spatial analysis found that the reconstructed PreDJ had good spatial representation of precipitation in the central-eastern China. Therefore, the results of this study provide reliable information for understanding long-term drought impacts on environmental conditions and socioeconomic development in central China.

Contributions of Parameterized Gravity Waves and Resolved Equatorial Waves to the QBO Period in a Future Climate of CESM2

AbstractContributions of the resolved waves and parameterized gravity waves to changes in the quasi‐biennial oscillation (QBO) in a future simulation (2015–2100) under the SSP370 scenario are investigated using the Community Earth System Model 2 (CESM2) with enhanced vertical resolution and are compared with those from four CESM2 historical simulations (1979–2014). The maximum QBO amplitude of the future simulation is 26.0 m s−1, which is slightly less than that of the historical simulations (27.4–29.3 m s−1). However, the QBO period in the future simulation is much shorter: 21.6 months in the early‐future (2015–2050) and 12 months in the late‐future (2065–2100) period, than in the historical simulations (23.5–30.9 months). The shortened QBO period in the future is primarily due to increases in both resolved wave forcing and parameterized gravity wave drag (GWD) in the stratosphere, with a more significant contribution by the GWD. As convective activity becomes stronger in the future simulation, the momentum flux of parameterized convective gravity waves at the cloud top increases, resulting in stronger GWD in the stratosphere. The increases in the magnitude of westward GWD dominate those of eastward GWD in the stratosphere. This is due to a significant increase in westward momentum flux in the troposphere, especially during the descending easterly QBO, and enhanced westerlies in the lowermost stratosphere, which introduces a westward anomaly. For the resolved waves, Kelvin wave forcing is a key contributor to increased eastward forcing in the future simulation, with relatively minor contributions by other equatorial planetary waves.

Recovering Quasi-Biennial Oscillations from Chaos

Abstract The Quasi-Biennial Oscillation (QBO) is understood to result from wave-mean-flow interactions, but the reasons for its relative stability remain a subject of ongoing debate. In addition, consensus has yet to be reached regarding the respective roles of different equatorial wave types in shaping the QBO’s characteristics. Here we employ Holton-Lindzen-Plumb’s quasilinear model to shed light on the robustness of periodic behavior in the presence of multiple wave forcings. A comprehensive examination of the various dynamical regimes in this model reveals that increased vertical wave propagation at higher altitudes favors periodicity. In the case of single standing wave forcing, enhanced vertical propagation is controlled by the wave attenuation length scale. The occurrence of non-periodic states at high forcing amplitudes is explained by the excitation of high vertical unstable modes. Increasing the attenuation length scale prevents the emergence of such modes. When multiple wave forcing is considered, the mean flow generated by a dominant primary wave facilitates greater vertical propagation of a perturbation wave. Raising the altitude where most of the wave damping occurs favors periodicity by preventing the development of secondary jets responsible for the aperiodic behavior. This mechanism underscores the potential role of internal gravity waves in supporting the periodicity of a QBO primarily driven by planetary waves.

Long‐Term Variability and Tendencies in Mesosphere and Lower Thermosphere Winds From Meteor Radar Observations Over Esrange (67.9°N, 21.1°E)

AbstractLong‐term variabilities of monthly zonal (U) and meridional winds (V) in northern polar mesosphere and lower thermosphere (MLT, ∼80–100 km) are investigated using meteor radar observations during 1999–2022 over Esrange (67.9°N, 21.1°E). The summer (June‐August) mean zonal winds are characterized by westward flow up to ∼88–90 km and eastward flow above this height. The summer mean meridional winds are equatorward with strong jet at ∼85–90 km and it weakens above this height. The U and V exhibit strong interannual variability that varies with altitude and month or season. The responses of U and V anomalies (from 1999 to 2003) to solar cycle (SC), Quasi Biennial Oscillation at 10 and 30 hPa, El Niño‐Southern Oscillation, North Atlantic Oscillation, ozone (O3) and carbon dioxide (CO2) are analyzed using multiple linear regression. From analysis, significant regions of correlations between MLT winds and above potential drivers vary with altitude and month. The positive responses of U and V to SC (up to 15 m/s/100 sfu) indicates the strengthening of eastward winds in mid‐late winter, and poleward winds in late autumn and early winter. The O3 likely intensifies the eastward and poleward winds (∼100 m/s/ppmv) in winter and early spring. The CO2 significantly influence the eastward flow in late winter and summer (above ∼90–95 km) and strengthen the meridional circulation. The significant positive trend in U peaks in summer, late autumn and early winter (∼0.6 m/s/year), the negative trend in V is more prominent in summer above ∼90–95 km.

Bayesian History Matching Applied to the Calibration of a Gravity Wave Parameterization

AbstractBreaking atmospheric gravity waves (GWs) in the tropical stratosphere are essential in driving the roughly 2‐year oscillation of zonal winds in this region known as the Quasi‐Biennial Oscillation (QBO). As Global Climate Models (GCM)s are not typically able to directly resolve the spectrum of waves required to drive the QBO, parameterizations are necessary. Such parameterizations often require knowledge of poorly constrained physical parameters. In the case of the spectral gravity parameterization used in this work, these parameters are the total equatorial GW stress and the half width of phase speed distribution. Radiosonde observations are used to obtain the period and amplitude of the QBO, which are compared against values obtained from a GCM. We utilize two established calibration techniques to obtain estimates of the range of plausible parameter values: History matching &amp; Ensemble Kalman Inversion (EKI). History matching is found to reduce the size of the initial range of plausible parameters by a factor of 98%, requiring only 60 model integrations. EKI cannot natively provide any uncertainty quantification but is able to produce a single best estimate of the calibrated values in 25 integrations. When directly comparing the approaches using the Calibrate, Emulate, Sample method to produce a posterior estimate from EKI, history matching produces more compact posteriors with fewer model integrations at lower ensemble sizes compared to EKI; however, these differences become less apparent at higher ensemble sizes.

Influence of the Quasi‐Biennial Oscillation on tropical convection and its teleconnection to the midlatitudes in boreal winter

AbstractRecently, a connection between the quasi‐biennial oscillation (QBO) and the Madden–Julian oscillation (MJO) was found in observations. The boreal winter seasonal mean amplitude of the daily MJO is anticorrelated with the QBO winds measured at 50 hPa. We investigate whether this relationship is captured in reforecasts of past winters in a seasonal prediction system, and find the ensemble mean response does not show this association, with essentially zero correlation. By repeated subsampling of the ensemble reforecast data, we also show that the observed correlation is very unlikely to arise merely as the result of sampling in the relatively short observational record. Instead, we conclude that the reforecasts genuinely cannot capture the observed QBO–MJO seasonal relationship. We also consider the boreal winter seasonal mean response of tropical convection to the QBO and its influence on the subtropics. We find that the reforecasts are able to capture the stationary pattern found in the observed outgoing long‐wave radiation (OLR) over the tropical west Pacific in response to the QBO. We also find that the associated upper tropospheric divergence in observations interacts with the west Pacific subtropical jet to produce atmospheric Rossby wave patterns that propagate across the midlatitude Pacific Ocean. These extratropical waves have the potential to interfere with the climatological waves in midlatitudes, modifying the strength of the midlatitude annular mode through interactions with the extratropical stratosphere. However, the observed influence on wavenumber 1 is not captured by the reforecasts, suggesting a weak extratropical influence and a partial explanation for the weaker‐than‐observed QBO teleconnection to the Arctic Oscillation and North Atlantic Oscillation.

The graft‐versus‐host problem for data‐driven gravity‐wave parameterizations in a one‐dimensional quasibiennial oscillation model

AbstractTwo key challenges in the development of data‐driven gravity‐wave parameterizations are generalization, how to ensure that a data‐driven scheme trained on the present‐day climate will continue to work in a new climate regime, and calibration, how to account for biases in the “host” climate model. Both problems depend fundamentally on the response to out‐of‐sample inputs compared with the training dataset, and are often conflicting. The ability to generalize to new climate regimes often goes hand in hand with sensitivity to model biases. To probe these challenges, we employ a one‐dimensional (1D) quasibiennial oscillation (QBO) model with a stochastic source term that represents convectively generated gravity waves in the Tropics with randomly varying strengths and spectra. We employ an array of machine‐learning models consisting of a fully connected feed‐forward neural network, a dilated convolutional neural network, an encoder–decoder, a boosted forest, and a support‐vector regression model. Our results demonstrate that data‐driven schemes trained on “observations” can be critically sensitive to model biases in the wave sources. While able to emulate accurately the stochastic source term on which they were trained, all of our schemes fail to simulate fully the expected QBO period or amplitude, even with the slightest perturbation to the wave sources. The main takeaway is that some measures will always be required to ensure the proper response to climate change and to account for model biases. We examine one approach based on the ideas of optimal transport, where the wave sources in the model are first remapped to the observed one before applying the data‐driven scheme. This approach is agnostic to the data‐driven method and guarantees that the model adheres to the observational constraints, making sure the model yields the right results for the right reasons.

Representation of the Stratospheric Circulation in CRA-40 Reanalysis: The Arctic Polar Vortex and the Quasi-Biennial Oscillation

Representation of the Stratospheric Circulation in CRA-40 Reanalysis: The Arctic Polar Vortex and the Quasi-Biennial Oscillation

Exploring Spatial and Temporal Patterns in the Debrecen Solar Faculae Database: Part I

Photospheric faculae are markers of the solar magnetic field, appearing as bright regions along the edges of granules on the Sun’s surface. Using data from the Debrecen Solar Faculae Database, we investigated the spatiotemporal distribution of photospheric faculae between 2010 May 1 and 2014 December 31 and found the following. (i) At lower latitudes, there is an enhanced abundance of faculae appearing as stripes at given Carrington longitudes, which are interpreted as indicative of the presence of active longitudes. (ii) At higher latitudes, we identified so-called crisscross patterns of facular appearance. These patterns are likely the result of faculae in regions situated along the boundaries of supergranules. Last but not least, (iii) various periods of oscillatory phenomena were identified in this facular data set, including a longer periodic range consistent with the quasi-biennial oscillations and shorter ones with periods of 4–12 days. Our findings are supported by the visualization of a simple heuristic thought experiment and more complex dynamo simulations, strengthening the proposed interpretation of the three observed solar phenomena reported.

On The Interannual Variability of Indonesian Monsoon Rainfall (IMR): A Literature Review of The Role of its External Forcing

The IMR variability is notorious for its hydrometeorological disasters. This paper examines recent studies on IMR and the main factors controlling its variability. The focus of this study is to investigate the impact of the atmosphere-ocean interaction that acts as the external forcing of IMR in the tropical Indian and Pacific Oceans. Specifically, the study will examine the influence of two climate phenomena, namely the El Nino Southern Oscillation (ENSO) and the Indian Ocean Dipole (IOD) and their interdecadal changes associated Pacific Decadal Oscillation (PDO), on the IMR. The review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) approach. Furthermore, data sets (such as rainfall, wind field, and SST) spanning 1990-2020 were used to verify the key findings. In general, this study concludes that the majority of the authors coincided with the following conclusion: ENSO and IOD events impact IMR by changing its amplitude, duration, intensity, and frequency of mean and extreme rainfall. Additionally, it has been shown that their impacts on IMR are most substantial during the dry seasons, specifically in June, July, and August (JJA), and not as strong as during the wet seasons, specifically in December, January, and February (DJF). Spatially, the effects of ENSO and IOD on IMR variability are clearly found more eastward and westward of the region, respectively. The expansions towards the east and west directions were facilitated by the displacement of the ascending and descending of Walker circulation patterns in the Indonesian region, respectively. Given the interannual fluctuations in IMR, caused mainly by ocean-atmosphere interactions, the knowledge gap of atmospheric factors like the Quasi-Biennial Oscillation (QBO) must be investigated in the future, as suggested by previous research and our preliminary study.

Crucial role of obliquely propagating gravity waves in the quasi-biennial oscillation dynamics

Abstract. ​​​​​​​In climate modelling, the reality of simulated flows in the middle atmosphere is largely affected by the model's representation of gravity wave processes that are unresolved, while these processes are usually simplified to facilitate computations. The simplification commonly applied in existing climate models is to neglect wave propagation in horizontal direction and time. Here we use a model that fully represents the propagation of unresolved waves in all directions, thereby elucidating its dynamical effect upon the most important climate mode in the tropical stratosphere, i.e. the quasi-biennial oscillation. Our simulation shows that the waves in the equatorial stratosphere, which are known to drive this climate mode, can originate far away from the Equator in the troposphere. The waves propagating obliquely toward the Equator are found to play a huge role in the phase progression of the quasi-biennial oscillation as well as in its penetration into the lower stratosphere. Such waves will require further attention, given that current climate models are struggling to simulate the quasi-biennial oscillation down to the lower stratosphere, which may be needed to reproduce its observed impacts on the surface climate.

Skillful prediction of length of day one year ahead in multiple decadal prediction systems

Despite a small amplitude, Length of Day (LOD) change, which varies from one year to another due to changes in Atmospheric Angular Momentum (AAM), determines the accuracy of Global Positioning System (GPS) time calculation. In this study, we examine the prediction skill of LOD and AAM in nine decadal prediction systems archived for the Decadal Climate Prediction Project. A persistence and rebound in LOD prediction skill at one year or longer lead time is found in most models. A poleward propagation of AAM anomaly via wave-mean flow interaction is also qualitatively well reproduced. This long-lead prediction of LOD and AAM is attributed to reliable predictions of the El Niño–Southern Oscillation (ENSO) and the Quasi-Biennial Oscillation (QBO), the former being more systematically related than the latter. This result indicates that the improved ENSO prediction and atmospheric wave-mean flow interaction may lead to better prediction of LOD, AAM and related extratropical climate in the decadal prediction systems.

Forecast Skill and Signal‐To‐Noise Ratio for Stratospheric Polar Vortex Variations in Northern Hemisphere Winter Seasonal Hindcasts

AbstractThis study explores seasonal forecast skill and signal‐to‐noise (SN) ratio for stratospheric polar vortex (SPV) variations during Northern winter using hindcast (HC) data of six systems initialized around early November in the Copernicus Climate Change Service database. Results show high skill for December‐January‐February (DJF) mean El Niño/Southern Oscillation and Quasi‐Biennial Oscillation (QBO) variations, although it is suggested that some systems underestimate the QBO amplitude. The ENSO and QBO variations in the HC data are also suggested to be underdispersive having too high SN ratio due to too strong signal and/or too weak noise (spread). The skill for DJF mean SPV variations is limited, with four systems having marginal skill near the 95% level, whereas the SN ratio is suggested to be realistic. It is also shown that no system has skill (evaluated using the Brier Score) at the 95% level in forecasting if each DJF winter experiences one or more major sudden stratospheric warmings, when a bias of the polar vortex strength in each system is simply considered. A comparison of the ensemble members for each system, when divided into two groups by the strength of the QBO‐SPV teleconnection, suggests an increase in the skill for SPV variations for the group of stronger teleconnection. This confirms the idea that the QBO‐SPV teleconnection contributes to the skill for SPV variations, and also suggests that the skill for SPV variations can be improved as the HC QBO and QBO‐SPV teleconnection are better simulated.

Aeolus wind lidar observations of the 2019/2020 quasi-biennial oscillation disruption with comparison to radiosondes and reanalysis

Abstract. The quasi-biennial oscillation (QBO) was unexpectedly disrupted for only the second time in the historical record during the 2019/2020 boreal winter. As the dominant mode of atmospheric variability in the tropical stratosphere and a significant source of seasonal predictability globally, understanding the drivers behind this unusual behaviour is very important. Here, novel data from Aeolus, the first Doppler wind lidar (DWL) in space, are used to observe the 2019/2020 QBO disruption. Aeolus is the first satellite able to observe winds at high resolution on a global scale, and it is therefore a uniquely capable platform for studying the evolution of the disruption and the broader circulation changes triggered by it. This study therefore contains the first direct wind observations of the QBO from space, and it exploits measurements from a special Aeolus scanning mode, implemented to observe this disruption as it happened. Aeolus observes easterly winds of up to 20 m s−1 in the core of the disruption jet during July 2020. By co-locating with radiosonde measurements from Singapore and the ERA5 reanalysis, comparisons of the observed wind structures in the tropical stratosphere are produced, showing differences in equatorial wave activity during the disruption period. Local zonal wind biases are found in both Aeolus and ERA5 around the tropopause, and the average Aeolus-ERA5 Rayleigh horizontal line-of-sight random error is found to be 7.58 m s−1. The onset of the QBO disruption easterly jet occurs 5 d earlier in Aeolus observations compared with the reanalysis. This discrepancy is linked to Kelvin wave variances that are 3 to 6 m2 s−2 higher in Aeolus compared with ERA5, centred on regions of maximum vertical wind shear in the tropical tropopause layer that are up to twice as sharp. The enhanced lower-stratospheric westerly winds which are known to help disrupt the QBO, perhaps with increasing frequency as the climate changes, are also stronger in Aeolus observations, with important implications for the future predictability of such disruptions. An investigation into differences in the equivalent depth of the most dominant Kelvin waves suggests that slower, shorter-vertical-wavelength waves break more readily in Aeolus observations compared with the reanalysis. This analysis therefore highlights how Aeolus and future DWL satellites can deepen our understanding of the QBO, its disruptions and the tropical upper-troposphere lower-stratosphere region more generally.

Interannual Variability of Zonal Mean Temperature, Water Vapor, and Clouds in the Tropical Tropopause Layer

AbstractWater vapor and cirrus clouds in the tropical tropopause layer (TTL) are important for the climate and are largely controlled by temperature in the TTL. On interannual timescales, both stratospheric and tropospheric modes of the large‐scale variability could affect temperatures in the TTL. Here multiple linear regression (MLR) is used to investigate explained variance in the cold point tropopause temperature (CPT), cold point tropopause height (CPZ), 83 hPa water vapor (WV83), 83 hPa ozone (O383), and total cirrus cloud fraction with cloud base (TTLCCF) and top (ALLCF) above 14.5 km, all averaged over 15°S‐15°N. Predictors of the MLR are a set of stratospheric and tropospheric large‐scale modes of variability. The MLR explains significant variance in CPT (76%), CPZ (78%), WV83 (65%), O383 (62%), TTLCCF (52%), and ALLCF (36%). The interannual variability of CPT and WV83 is dominated by stratospheric processes associated with the Quasi‐Biennial Oscillation (QBO) and Brewer‐Dobson Circulation (BDC), whereas the variability of CPZ, O383, TTLCCF and ALLCF is also controlled by 500 hPa temperature (T500). Residual variability in CPT and CPZ not captured by the MLR are further significantly correlated to stratospheric temperature. It is shown that the portion of the BDC's shallow branch missed by the eddy heat flux based BDC index contributes significant amounts of the explained variances.

Tracking active nests in solar-type pulsators: Ensemble starspot modelling of Kepler asteroseismic targets

The satellite Planetary Transits and Oscillations of stars (PLATO), due to be launched late 2026, will provide us with an unprecedented sample of light curves of solar-type stars that will exhibit both solar-type oscillations and signatures of activity-induced brightness modulations. Solar-type pulsators only have moderate levels of activity because high levels of activity inhibit oscillations. This means that these targets represent a specific challenge for starspot modelling. In order to assess the possibilities that PLATO will soon open, we wish to characterise the morphology of active regions at the surface of stars for which we also have a detection of solar-like acoustic oscillations. In this context, we report the results of an ensemble starspot modelling analysis of the Sun and ten solar-type pulsators observed by the Kepler satellite. We implement a Bayesian starspot modelling approach based on a continuous-grid model, accounting for the combined starspot and facular contribution to activity-induced brightness modulations. From our analysis, we find that several stars of our sample exhibit clear signatures of stable longitudinal active nests while sharing activity levels and convection versus rotation regimes similar to the solar regime. By searching for modulations in the reconstructed starspot coverage, we found significant periodicities that we identify as possible signatures of cyclic modulations similar to the quasi-biennal oscillation or the Rieger cycle. We can infer the corresponding intensity of the magnetic field at the bottom of the convective envelope based on the hypothesis that internal magneto-Rossby waves acting on the tachocline cause these modulations.

An empirical model for the tropical Indian region using in-situ and space-borne observations taking into account of tropical oscillations

An empirical model of temperature is developed using long-term observations of radiosonde (52 years), M-100 Rocket (17 years), and Sounding of the Atmosphere by Broadband Emission Radiometry (SABER) (22 years) data. The dominant oscillations such as Annual Oscillation (AO), Semi-Annual Oscillation (SAO), Terr-Annual oscillation (TAO), and Quasi-Biennial Oscillation (QBO) variability has been incorporated into the temperature to obtain the empirical atmosphere. The monthly mean of temperature obtained from radiosonde from 0 to 25 km, M-100 rocket from 26 to 80 km and SABER from 26 to 109 km has been subjected to the Fast Fourier Transform technique to retrieve the amplitudes and phases of AO, SAO, TAO, and QBO. The amplitudes and phases obtained are put in an empirical formula along with the temperature obtained from the observations to retrieve the empirical temperature. The empirical model is developed for two epochs (1971–1990, and 2002–2023). The amplitude is similar in both epochs for the TAO, and it is high in the first compared to the second epoch for SAO, AO, and QBO. The peak in the amplitude (3–4 K) is obtained between 80 and 90 km for TAO, and between 70 and 80 km (2–3 K) for SAO. The amplitude is less than 1 K for AO in both the epochs. The peaks in the QBO are observed between 20 and 30 km (2–3 K), another between 60 and 70 km (3–4 K) in the first epoch and above 80 km (2–3 K). Downward phase propagation is dominant for all the oscillations observed below 50 km, and upward propagation above. The empirical temperature obtained compares well with the observations below 20 km (RMSE <1 K) and shows differences above it (RMSE ∼ 4–6 K). This is due to the fact, that the amplitude of the oscillation is more in the stratosphere and mesosphere which results in perturbation of the background temperature.

Explainable Offline‐Online Training of Neural Networks for Parameterizations: A 1D Gravity Wave‐QBO Testbed in the Small‐Data Regime

AbstractThere are different strategies for training neural networks (NNs) as subgrid‐scale parameterizations. Here, we use a 1D model of the quasi‐biennial oscillation (QBO) and gravity wave (GW) parameterizations as testbeds. A 12‐layer convolutional NN that predicts GW forcings for given wind profiles, when trained offline in a big‐data regime (100‐year), produces realistic QBOs once coupled to the 1D model. In contrast, offline training of this NN in a small‐data regime (18‐month) yields unrealistic QBOs. However, online re‐training of just two layers of this NN using ensemble Kalman inversion and only time‐averaged QBO statistics leads to parameterizations that yield realistic QBOs. Fourier analysis of these three NNs' kernels suggests why/how re‐training works and reveals that these NNs primarily learn low‐pass, high‐pass, and a combination of band‐pass filters, potentially related to the local and non‐local dynamics in GW propagation and dissipation. These findings/strategies generally apply to data‐driven parameterizations of other climate processes.