Automated identification of dominant physical processes

Kaiser, Bryan; Sáenz, Juan A.; Sonnewald, Maike; Livescu, Daniel

doi:10.1016/j.engappai.2022.105496

Cited by 10 publications

(11 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Similar applications of machine learning as a tool for empirical leading order analysis is ideally placed to accelerate discovery within oceanography and beyond [24,25]. Specifically within the context of understanding the dynamical balances within gyre circulation, [42] highlights the impact that the choice of geographic area within which to assess the dynamical balances matters.…”

Section: Discussionmentioning

confidence: 99%

“…The unifying theory was arrived at using machine learning based objective regime discovery that can highlight where coherent physical drivers are acting [24,25]. While the regimes in Sonnewald et al (2019) are global, the focus here is on the dynamical regimes present in the Southern Ocean (Fig.…”

Section: Understanding Wind Gyre Circulation Using a Barotropic Vorti...mentioning

confidence: 99%

“…Understanding the interplay of the governing forces of ocean circulation, discovering in what areas certain driving forces are more prominent than others can focus work and accelerate insight, for example, as illustrated through the problematic nature of arbitrarily choosing gyre boundaries. Within physical systems ranging from the ocean and climate, through to general relativity and medicine, identifying subregimes in driving forces have driven progress by guiding intuition [25]. Unsupervised machine learning is well suited to finding subregimes, and can act objectively without human bias.…”

Section: Insight Guided By Machine Learningmentioning

confidence: 99%

See 2 more Smart Citations

The Southern Ocean supergyre: a unifying dynamical framework identified by machine learning

Sonnewald

Reeve²,

Lguensat³

2023

Preprint

View full text Add to dashboard Cite

The Southern Ocean closes the global overturning circulation and is key to the regulation of carbon and heat, biological production, and sea level. However, the dynamics of the general circulation and upwelling pathways remain poorly understood. Here, a unifying framework is proposed invoking a semi-circumpolar `supergyre' south of the Antarctic circumpolar current: a massive series of ‘leaking’ sub-gyres spanning the Weddell and Ross seas that are connected and maintained via rough topography that acts as scaffolding. The supergyre framework challenges the conventional view of having separate circulation structures in the Weddell and Ross seas and suggests a limited utility for climate applications of idealized models and conventional zonal averaged frameworks. Machine learning was used to reveal areas of coherent driving forces within a vorticity-based analysis. Predictions from the supergyre framework are supported by available observations and could aid observational and modelling efforts of the climatically key region undergoing rapid change.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Understanding Wind Gyre Circulation Using a Barotropic Vorti...mentioning

confidence: 99%

Section: Insight Guided By Machine Learningmentioning

confidence: 99%

See 1 more Smart Citation

The Southern Ocean supergyre: a unifying dynamical framework identified by machine learning

Sonnewald

Reeve²,

Lguensat³

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Specifically, we used GMM to classify profiles from a historical experiment and two future climate experiments, namely SSP1-2.6 and SSP5-8.5. We used GMM as a "hypothesis generation tool", generating ideas for further exploration and analysis (Kaiser et al, 2022). Note that the detailed exploration of this hypothesis is beyond the scope of this technical note; further analysis of the ideas presented here would be a welcome addition to the literature.…”

Section: Discussionmentioning

confidence: 99%

Technical note: Unsupervised classification of ozone profiles in UKESM1

Fahrin

Harle

et al. 2023

Atmos. Chem. Phys.

View full text Add to dashboard Cite

Abstract. The vertical distribution of ozone in the atmosphere, which features complex spatial and temporal variability set by a balance of production, loss, and advection, is relevant for both surface air pollution and climate via its role in radiative forcing. At present, the way in which regions of coherent ozone structure are defined relies on somewhat arbitrarily drawn boundaries. Here we consider a more general, data-driven method for defining coherent regimes of ozone structure. We apply an unsupervised classification technique called Gaussian mixture modeling (GMM), which represents the underlying distribution of ozone profiles as a linear combination of multi-dimensional Gaussian functions. In doing so, GMM identifies coherent groups or subpopulations of the ozone profile distribution. As a proof-of-concept study, we apply GMM to ozone profiles from three subsets of the UKESM1 coupled climate model runs carried out for CMIP6: specifically, the seasonal mean of a historical subset (2009–2014) and two subsets from two different future climate projections (i.e., SSP1-2.6 and SSP5-8.5). Despite not being given any spatiotemporal information, GMM identifies several spatially coherent regions of ozone structure. Using a combination of statistical guidance and post hoc judgment, we select a six-class representation of global ozone, consisting of two tropical classes and four mid-to-high-latitude classes. The tropical classes feature a relatively high-altitude tropopause, while the higher-latitude classes feature a lower-altitude tropopause and low values of tropospheric ozone, as expected based on broad patterns observed in the atmosphere. Both of the future projections feature lower ozone concentrations at 850 hPa than the historical benchmark, with signatures of ozone hole recovery. We find that the area occupied by the tropical classes is expanded in both future projections, which are most prominent during austral summer. Our results suggest that GMM may be a useful method for identifying coherent ozone regimes, particularly in the context of model analysis.

show abstract

“…A hypothesis considered here is that deviations of the spectral shape or other properties of the internal wave spectrum from the assumed GM form may be relevant (Müller & Liu, 2000;Polzin & Lvov, 2011), or variability in other individual parameters of the parameterization themselves, based on the local geography, topographic conditions and the presence of external forcing to the local internal wave field (Waterman et al, 2014;Chinn et al, 2016;Pollmann, 2020). Recently, both supervised and unsupervised learning approaches have been used across a variety of fluid mechanical applications to provide new insight into fundamental relationships and patterns of variability in our oceans (Giglio et al, 2018;Brunton et al, 2020;Callaham et al, 2021;Kaiser et al, 2022;Mashayek et al, 2022). In particular, clustering techniques have proven useful in generating insights and exploring existing oceanographic data such as categorizing datasets of temperature-salinity profiles (e.g., Rosso et al, 2020;Jones et al, 2019;Boehme & Rosso, 2021), classifying global ecological marine provinces (Sonnewald et al, 2020) and identifying dominant dynamical balances in global ocean circulation models (Sonnewald et al, 2019).…”

Section: Introductionmentioning

confidence: 99%

Global Patterns of Bias in Ocean Mixing Parameterization Identified Through Unsupervised Machine Learning

Lele

Purkey

MacKinnon

et al. 2023

Preprint

View full text Add to dashboard Cite

Turbulent mixing in the ocean is often parameterized in terms of the downscale energy transfer by internal waves. Expressed in terms of the vertical wavenumber spectrum of velocity shear and isopycnal strain, the “finescale parameterization” relies on several parameters that set the shape of the spectrum. Here, we use an unsupervised learning model to identify dominant spectral shapes and their geographical dependencies in data from 1875 global hydrographic profiles. The model reveals clustered regions with underlying spectra deviating significantly from the canonical Garrett-Munk (GM) spectrum, suggesting potential departures from the underlying assumption about the downscale energy cascade. These clusters are further found to be associated with regions where parameterized estimates of mixing differ significantly from temperature microstructure observations. Extending the methodology to other hydrographic datasets has the potential to reveal reasons for parameterization bias and to identify the dynamical underpinnings leading to more robust parameterizations of oceanic turbulent mixing.

show abstract

Automated identification of dominant physical processes

Cited by 10 publications

References 18 publications

The Southern Ocean supergyre: a unifying dynamical framework identified by machine learning

The Southern Ocean supergyre: a unifying dynamical framework identified by machine learning

Technical note: Unsupervised classification of ozone profiles in UKESM1

Global Patterns of Bias in Ocean Mixing Parameterization Identified Through Unsupervised Machine Learning

Contact Info

Product

Resources

About