Estimating the Kinetic Energy Spectrum from the Second-Order Velocity Structure Function using a Regularized Fitting Approach
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Physics. The day in general physics: instruments, methods and results across the field.
Kai: We're looking at the paper: "Estimating the Kinetic Energy Spectrum from the Second-Order Velocity Structure Function using a Regularized Fitting Approach".
Mira: Here's the gist of the paper. Estimating the Kinetic Energy Spectrum from the Second-Order Velocity Structure Function using a Regularized Fitting Approach is a novel method proposed to overcome the ill-conditioned nature of inverting relationships between kinetic energy spectra and second-order structure functions by employing a regularized fitting approach that stabilizes the inversion <ref:2604.27200#pg12>.
Kai: This framework enables robust estimation of spectral slopes from sparse Lagrangian data, extending spectral diagnostics beyond gridded Eulerian measurements <ref:2604.27200#pg6>.
Mira: Theoretical Background The paper establishes a relationship between the kinetic energy spectrum, E(k), and the second-order structure function, SF2(r), through an integral equation involving the zeroth-order Bessel function <ref:2604.27200#pg36>.
Kai: The forward relation is given by S F2 (r) = 2 ∫ ∞ 0 E(k)
1− J0 (kr): dk.
Title and authors: Mira: So, focusing on the title itself, Estimating the Kinetic Energy Spectrum from the Second-Order Velocity Structure Function using a Regularized Fitting Approach, it really tells you what they are doing: they are using structure function data to find kinetic energy spectra >
Kai: It's not just finding a spectrum; it’s dealing with this inherent ill-conditioning of the inverse relationship that exists between the two measurements >
Lev: The paper sets up a forward relation involving an integral equation with the zeroth-order Bessel function, and then they have to deal with that oscillatory Bessel kernel in the inverse transformation >
Mira: That sensitivity is what motivates them to use this regularized approach where they assume a finite number of segments for the spectrum instead of trying to invert it directly >
Kai: They validate this method first on idealized data, testing it against spectra they already know and then comparing the reconstruction of structure functions with the true ones, seeing errors on the order of ten to the minus five in oceanic and atmospheric cases > <ref:2604.27200#pg1>
Lev: That validation step is important because it shows that even in a perfect setup, this constrained fitting method can get very accurate results >
The paper's summary: Kai: Moving into what they actually did, the paper summarizes their methodology by saying they assume the kinetic energy spectrum consists of a finite number of power-law segments over equal intervals in logarithmic wavenumber space >
Mira: They perform the estimation indirectly by assuming this parametric form, then evaluating that form using the forward relation to get an estimated structure function, and then adjusting those parameters until that estimated structure function matches what they actually measured >
Lev: So it's a fitting process where they constrain the amplitudes and slopes to keep things physically sensible while solving for the spectrum parameters >
Kai: They found that for oceanic and atmospheric cases, the optimal number of spectral segments ended up being twenty-nine meaning their total set of parameters was S plus one >
Mira: That's a specific finding because it gives you a concrete idea of how many pieces you need to model the turbulence structure for those systems >
Lev: It’s a nice result for parameter counting, but Kai, what about when we try to apply this to something more complex than idealized spectra?
The paper's improvements: Kai: Well, one of the improvements they bring in is how they handle real-world conditions by testing the framework using high-resolution velocity fields from the MITgcm llc4320 simulation >
Mira: That simulation is key because it gives a realistic picture of upper-ocean turbulence and they found that their reconstructed spectra closely follow the directly computed ones across all wavenumbers except at the two highest wavenumber points near the Nyquist limit >
Lev: It's good that they can resolve changes in spectral slope across scales there, capturing distinct regimes like quasi-geostrophic and submesoscale dynamics >
Kai: They also applied this framework to sparse surface drifter observations from GLAD and LASER experiments in the Gulf of Mexico, where the mean estimated structure function closely followed the mean observed one across most separations >
Mira: That's good because it shows stability even when you have sampling variability, and they confirmed that their estimated slopes satisfy the expected power-law relationship between structure functions and kinetic energy spectra >
Lev: But there’s a caveat they mention too, which is that uncertainties in the second-order structure function at large separations propagate into the estimated spectra, limiting accuracy at low wavenumbers >
Conclusion: Kai: So to wrap up on the paper Estimating the Kinetic Energy Spectrum from the Second-Order Velocity Structure Function using a Regularized Fitting Approach, this method is robust because it stabilizes that ill-conditioned inversion by turning it into a constrained fitting problem >
Mira: It allows us to estimate spectral slopes over a broad range, specifically from negative four to one, which is more general than relying solely on classical power-law arguments >
Lev: And when applied to realistic data like the MITgcm simulation or drifter data, it manages to capture different regimes of motion across scales >
Kai: But we have to remember the limitations they state: accuracy at low wavenumbers is limited by uncertainties in the structure function at large separations, and performance drops with increasing noise >
Mira: So this framework is really good for recovering broad spectral structure under low noise, but if you have high noise or need fine detail at the very small scales, it struggles a bit >
Lev: If we look ahead, they plan to use smoother basis functions instead of those Heaviside functions and incorporate noise-aware fitting strategies to handle scale-dependent uncertainty better >
Kai: Right. It’s a solid method for bridging that gap between what you measure in the structure function and what you want to know about the energy distribution across scales >
Ayantika Bhattacharjee, C Spencer Jones, Dhruv Balwada, Shane Elipot, Manuel O. Gutierrez-Villanueva
Texas A&M University · Lamont-Doherty Earth Observatory Columbia University · Rosenstiel School of Marine and Atmospheric Science University of Miami
physics.ao-ph
Submitted: 2026-04-29
Updated: 2026-10-28
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Importance score: 77/100
The gist: Estimating the Kinetic Energy Spectrum from the Second-Order Velocity Structure Function using a Regularized Fitting Approach is a novel method proposed to overcome the ill-conditioned nature of…
Key concepts
- Kinetic Energy Spectrum (E(k))
- This represents how much kinetic energy is present at different spatial scales (wavenumbers) in the fluid. The paper seeks to estimate this spectrum from indirect measurements like velocity structure functions, which are easier to obtain from observational data.
- Second-Order Structure Function (SF2(r))
- This function measures how kinetic energy is distributed across different separation distances (r) between points in the fluid. It is related to the kinetic energy spectrum through an integral equation involving Bessel functions, providing a key observable for estimation.
- Regularized Fitting Approach
- Because inverting the relationship between SF2(r) and E(k) is mathematically unstable due to noise, this method avoids direct inversion. Instead, it assumes a simple parametric form for the energy spectrum and fits parameters to match observed structure functions, stabilizing the estimation process.
- Piecewise Power-Law Segments
- The paper models the kinetic energy spectrum as several segments of power laws defined over equal intervals in logarithmic wavenumber space. This assumption regularizes the fitting problem by limiting the number of free parameters, preventing overfitting and ensuring physically plausible spectral slopes.
Terminology
Summary
Estimating the Kinetic Energy Spectrum from the Second-Order Velocity Structure Function using a Regularized Fitting Approach is a novel method proposed to overcome the ill-conditioned nature of inverting relationships between kinetic energy spectra and second-order structure functions by employing a regularized fitting approach that stabilizes the inversion <ref:2604. p. 027200#pg12>. This framework enables robust estimation of spectral slopes from sparse Lagrangian data, extending spectral diagnostics beyond gridded Eulerian measurements <ref:2604. p. 127200#pg6>.
Theoretical Background
The paper establishes a relationship between the kinetic energy spectrum, E(k), and the second-order structure function, SF2(r), through an integral equation involving the zeroth-order Bessel function <ref:2604.27200#pg36>. The forward relation is given by S F2 (r) = 2 ∫ ∞ 0 E(k) [1− J0 (kr)] dk. p. 6 The inverse relation, which is highly sensitive to noise and discretization errors, involves an oscillatory Bessel kernel. p. 4 This sensitivity motivates the proposed regularized approach to stabilize the ill-conditioned inversion <ref:2604. p. 427200#pg6>.
Regularized Estimation Methodology
The core of the proposed method assumes a parametric form for E(k) consisting of piecewise power-law segments defined over successive equal intervals in logarithmic wavenumber space. The estimation is performed indirectly by assuming this parametric form, evaluating the corresponding SF2(r) using the forward relation, and adjusting parameters until they match the estimated structure function. p. 8 This process involves fitting a limited set of parameters, which regularizes the inversion. p. 8 The amplitude parameters (b i) are constrained to be non-negative, and spectral slopes (alpha i) are restricted to the interval [-4, 1] for physical plausibility. p. 8
Validation Against Idealized Data
The method's accuracy is first tested on idealized KE spectra with prescribed spectral slopes and energy levels. The reconstruction of the corresponding structure functions using the regularized approach closely overlaps with the true structure functions, achieving best-fit errors on the order of 10-5 for oceanic and atmospheric cases. p. 0 The optimal number of spectral segments was found to be 29 for oceanic and atmospheric cases, resulting in S + 1 parameters. p. 8
Performance Under Realistic Conditions
The framework was evaluated using high-resolution velocity fields from the MITgcm llc4320 simulation, which provides a realistic representation of upper-ocean turbulence. The reconstructed spectra closely follow the directly computed spectra across all wavenumbers except at the two highest-wavenumber points near the Nyquist limit. p. 8 The method successfully resolves changes in spectral slope across scales, capturing distinct regimes such as quasi-geostrophic and submesoscale dynamics. p. 2
Application to Observational Data
The framework was applied to sparse surface drifter observations from GLAD and LASER experiments in the Gulf of Mexico. The mean optimized SF2(r) closely follows the mean observed SF2(r) across most separations, demonstrating stability despite sampling variability. p. 7 The estimated slopes satisfy the expected power-law relationship between SF2(r) and E(k), supporting the consistency of the method with theoretical expectations. p. 7 However, uncertainties in SF2(r) at large separations propagate into the estimated spectra and limit accuracy at low wavenumbers. p. 4 The method is robust in recovering broad spectral structure under low noise but its ability to resolve slope transitions decreases with increasing noise. p. 1
Future Directions
Future work includes applying the framework to a wider range of flow fields, such as those decomposed through wave–vortex or Helmholtz decompositions, to characterize balanced and unbalanced motions separately. Further refinements are planned to replace the non-differentiable Heaviside functions with smoother basis functions and incorporate noise-aware fitting strategies that explicitly account for scale-dependent uncertainty. p. 1 This would enhance performance at large separations where independent pair counts are limited. p. 4 The chosen slope bounds [-4, 1] serve a regularizing role by preventing the optimizer from exploiting unconstrained values to overfit local features. p. 9 The framework can be applied directly to a wide range of existing datasets and flow partitions. p. 4
Acknowledgments
This work was funded by NSF Physical Oceanography grant number OCE2242109. Data availability statement indicates that the MITgcm LLC4320 data can be accessed using the code available at https://doi. p. 4org/10.5281/zenodo.11373789. The surface drifter data are available at https://data. p. 1gulfresearchinitiative.org/. The code for the method and figure generation is available at https://doi. p. 7org/10.5281/zenodo.19704378.
References
Babiano, A., C. Basdevant, and R. Sadourny, 1985: Structure functions and dispersion laws in two-dimensional turbulence <ref:2604. p. 627200#pg29>.
Balwada, D. p. 0 p. 0, J. H. LaCasce, and K. p. 3 p. 3 G. Speer, 2016: Scale-dependent distribution of kinetic energy from surface drifters in the gulf of mexico. p. 3
Balwada, D., J.-H. Xie, R. Marino, and F. Feraco, 2022: Direct observational evidence of an oceanic dual kinetic energy cascade and its seasonality. p. 2
Barkan, R., K. B. Winters, and J. C. McWilliams, 2017: Stimulated imbalance and the enhancement of eddy kinetic energy dissipation by internal waves. p. 2
Bennett, A. p. 6, 1984: Relative dispersion: Local and nonlocal dynamics. p. 3
Branch, M. A., T. F. Coleman, and Y. p. 8 Li, 1999: A subspace, interior, and conjugate gradient method for large-scale bound-constrained minimization problems.
B¨uhler, O. p. 2, J. Callies, and R. p. 2 Ferrari, 2014: Wave–vortex decomposition of one-dimensional ship-track data. p. 2
Callies, J. p. 2 p. 2, and R. Ferrari, 2013: Interpreting energy and tracer spectra of upper-ocean turbulence in the submesoscale range (1–200 km). p. 2
Callies, J., R. Ferrari, J. p. 2 M. Klymak, and J. Gula, 2015: Seasonality in submesoscale turbulence. p. 2
Cao, H., B. Fox-Kemper, Z. Jing, X. Song, and Y. Liu, 2023: Towards the upper-ocean unbalanced submesoscale motions in the oleander observations. p. 3
Chereskin, T. p. 3 K., C. B. Rocha, S. T. Gille, D. Menemenlis, and M. Passaro, 2019: Characterizing the transition from balanced to unbalanced motions in the southern california current. p. 3
Coleman, T. p. 8 F., and Y. Li, 1996: An interior trust region approach for nonlinear minimization subject to bounds. p. 8
Cree, M., and P. Bones, 1993: Algorithms to numerically evaluate the hankel transform. p. 7
Davidson, P. p. 3, 2015: Turbulence: an introduction for scientists and engineers. p. 2
Dufour, C. O., and Coauthors, 2015: Role of mesoscale eddies in cross-frontal transport of heat and biogeochemical tracers in the southern ocean. p. 2
D’Asaro, E. A., and Coauthors, 2018: Ocean convergence and the dispersion of flotsam. p. 2
Elipot, S., R. Lumpkin, R. p. 3 p. 3 C. Perez, J. M. Lilly, J. J. Early, and A. M. Sykulski, 2016: A global surface drifter data set at hourly resolution. p. 4
Elipot, S., and J. Wenegrat, 2021: Vertical structure of near-surface currents: Importance, state of knowledge, and measurement challenges. p. 2
Ferrari, R. p. 2, and C. Wunsch, 2009: Ocean circulation kinetic energy: Reservoirs, sources, and sinks. p. 1
Gutierrez-Villanueva, M. p. 0 O., B. D. Cornuelle, S. T. Gille, M. R. Mazloff, and D. Balwada, 2026: An improved methodology to estimate cross-scale kinetic energy transfers from third order structure functions using regularized least squares. p. 0
Koszalka, I. p. 3, J. LaCasce, and K. Orvik, 2009: Relative dispersion in the nordic seas. p. 3
LaCasce, J. p. 3 p. 3 p. 3, 2002: On turbulence and normal modes in a basin. p. 7
LaCasce, J., 2016: Estimating eulerian energy spectra from drifters. p. 3
LaCasce, J., and T. Meunier, 2022: Relative dispersion with finite inertial ranges. p. 3
Lindborg, E. p. 6 p. 6, 1999: Can the atmospheric kinetic energy spectrum be explained by two-dimensional turbulence? p. 0 Journal of Fluid Mechanics, 388, 259–288. p. 0
Lindborg, E., 2007: Horizontal wavenumber spectra of vertical vorticity and horizontal divergence in the upper troposphere and lower stratosphere. p. 5
Lumpkin, R., M. Pazos, and Coauthors, 2007: Measuring surface currents with surface velocity program drifters: the instrument, its data, and some recent results. p. 2
Marshall, J., C. Hill, L. Perelman, and A. Adcroft, 1997: Hydrostatic, quasi-hydrostatic, and nonhydrostatic ocean modeling. p. 3
McCaffrey, K. p. 3, B. Fox-Kemper, and G. Forget, 2015: Estimates of ocean macroturbulence: Structure function and spectral slope from argo profiling floats. p. 2
McWilliams, J. p. 2 p. 2 C., 2008: The nature and consequences of oceanic eddies. p. 2
McWilliams, J. C., 2016: Submesoscale currents in the ocean. p. 2
Novelli, G. p. 3, C. M. Guigand, C. Cousin, E. H. Ryan, N. J. Laxague, H. Dai, B. K. Haus, and T.-M Ozg¨okmen, 2017: A biodegradable surface drifter for ocean sampling on a massive scale. p. 3
Poje, A., C., and Coauthors, 2014: Submesoscale dispersion in the vicinity of the deepwater horizon spill. p. 3
Qian, Y.-K., J. H. LaCasce, and S. p. 3 Peng, 2025: Inferring submesoscale energy spectra in the gulf of mexico from surface drifters. p. 3
Qiu, B., T. Nakano, S. Chen, and P. Klein, 2017: Submesoscale transition from geostrophic flows to internal waves in the northwestern pacific upper ocean. p. 3
Rocha, C. B., T. K. Chereskin, S. p. 3 T. Gille, and D. Menemenlis, 2016a: Mesoscale to submesoscale wavenumber spectra in drake passage. p. 3
Rocha, C. B., S. T. Gille, T. K. Chereskin, and D. p. 3 Menemenlis, 2016b: Seasonality of submesoscale dynamics in the kuroshio extension. p. 3
Scott, R. B., and F. Wang, 2005: Direct evidence of an oceanic inverse kinetic energy cascade from satellite altimetry. p. 2
Soares, S. M., S. T. Gille, T. K. Chereskin, E. p. 3 Firing, J. Hummon, and C. B. Rocha, 2022: Transition from balanced to unbalanced motion in the eastern tropical pacific. p. 3
Steinberg, J. p. 2, S. T., Cole, K., Drushka, and R. P. Abernathey, 2022: Seasonality of the mesoscale inverse cascade as inferred from global scale-dependent eddy energy observations. p. 3
Su, Z., H. Torres, P. Klein, A. F. Thompson, L. Siegelman, J. Wang, D. Menemenlis, and C. Hill, 2020: High-frequency submesoscale motions enhance the upward vertical heat transport in the global ocean. p. 2
Torres, H., S., P.-K., D.-M., B.-Q., Z.-S., J. Wang, S. Chen, and L.-L. Fu, 2018: Partitioning ocean motions into balanced motions and internal gravity waves: A modeling study. p. 3
Torres, H. S., and Coauthors, 2022: Separating energetic internal gravity waves and small-scale frontal dynamics. p. 3
Torres, H. S., and Coauthors, 2025: Submesoscale eddy contribution to ocean vertical heat flux diagnosed from airborne observations. p. 2
Vallis, G. K., 2017: Atmospheric and oceanic fluid dynamics. p. 0
Virtanen, P. p. 8, and Coauthors, 2020: Scipy 1. p. 20: Fundamental algorithms for scientific computing in python. p. 8
Wang, Y., S. Zhang, and Y. Jia, 2025: Enhanced resolution capability of swot sea surface height measurements and their application in monitoring ocean dynamics variability. p. 2
Appendix A1
Figures 2 and 6 show that the error spike occurs when the number of spectral segments equals the total number of data points, N. The non-linear least-squares fitting in this study is performed using the Trust Region Reflective (TRF) algorithm, which operates on the Jacobian matrix J ∈ R N×(S+1). p. 8 Since rank(J) ≤ min(N, S + 1), three regimes arise. p. 3 For S ≥ N, J becomes a fat matrix (30×31), and the rank–nullity theorem guarantees a nontrivial null space, (S +1) −rank(J) ≥ 1. p. 7
2604.272001pg1
2604.272005pg3
34pg9
35pg8
35pg8
36pg9
44pg11
46pg19
--- Page 5 --- (Note: The provided text has page numbers 1-27, but the token list only goes up to page 27. I will use the provided numbering for consistency. p. 6)
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Regularized estimation of kinetic energy spectra from sparse Lagrangian data allows AI systems to estimate KE spectra from
sparse Lagrangian data
by assuming afinite number of segments with distinct slopes and amplitudes.
This enables AI to perform spectral diagnostics on observational data like GLAD and LASER experiments, extending thembeyond gridded Eulerian measurements.
-
Robust reconstruction of KE spectra from second-order structure functions stabilizes the ill-conditioned inversion by employing a
regularized approach
that reduces the problem to fitting a limited set of parameters. This allows AI systems to estimate underlying spectral slopes over a broader range, specifically recovering slopesfrom -4 to +1,
which is more general than classical power-law arguments. -
Performance under realistic conditions includes testing against scale-dependent noise, allowing AI systems to handle observational uncertainties by applying multiplicative chi-squared noise models. This enables the system to maintain accuracy even when data exhibits
scale-dependent uncertainty
and improves the recovery of spectral structure in high noise scenarios compared to direct inversion. -
Application to model data allows AI systems to validate their reconstruction capabilities against dynamically consistent settings, such as the MITgcm llc4320 simulation. This ensures that the inferred KE spectra are physically plausible and captures
distinct spectral slopes and partially capture slope transitions associated with different theoretical regimes.
Related papers
- Anthropogenic Heat in Urban Climate Systems: Forcing, Sensitivity, and Feedback
- NORi: An ML-Augmented Ocean Boundary Layer Parameterization
- A Checklist to assess the energy and carbon impacts of ML/AI applications in Earth System Modeling
- Skillful Thunderstorm Forecasts Learned Directly from Observations
- Wind vector retrieval from miniaturized wave-enabled sea-surface drifters
- Stochastic coupling of climate variables and ice volume over the Late Pleistocene glacial cycles