skip to main content


Title: Physics-informed deep-learning parameterization of ocean vertical mixing improves climate simulations
Abstract Uncertainties in ocean-mixing parameterizations are primary sources for ocean and climate modeling biases. Due to lack of process understanding, traditional physics-driven parameterizations perform unsatisfactorily in the tropics. Recent advances in the deep-learning method and the new availability of long-term turbulence measurements provide an opportunity to explore data-driven approaches to parameterizing oceanic vertical-mixing processes. Here, we describe a novel parameterization based on an artificial neural network trained using a decadal-long time record of hydrographic and turbulence observations in the tropical Pacific. This data-driven parameterization achieves higher accuracy than current parameterizations, demonstrating good generalization ability under physical constraints. When integrated into an ocean model, our parameterization facilitates improved simulations in both ocean-only and coupled modeling. As a novel application of machine learning to the geophysical fluid, these results show the feasibility of using limited observations and well-understood physical constraints to construct a physics-informed deep-learning parameterization for improved climate simulations.  more » « less
Award ID(s):
2048631
NSF-PAR ID:
10403346
Author(s) / Creator(s):
; ; ; ; ;
Date Published:
Journal Name:
National Science Review
Volume:
9
Issue:
8
ISSN:
2095-5138
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. null (Ed.)
    Systems exhibiting nonlinear dynamics, including but not limited to chaos, are ubiquitous across Earth Sciences such as Meteorology, Hydrology, Climate and Ecology, as well as Biology such as neural and cardiac processes. However, System Identification remains a challenge. In climate and earth systems models, while governing equations follow from first principles and understanding of key processes has steadily improved, the largest uncertainties are often caused by parameterizations such as cloud physics, which in turn have witnessed limited improvements over the last several decades. Climate scientists have pointed to Machine Learning enhanced parameter estimation as a possible solution, with proof-of-concept methodological adaptations being examined on idealized systems. While climate science has been highlighted as a "Big Data" challenge owing to the volume and complexity of archived model-simulations and observations from remote and in-situ sensors, the parameter estimation process is often relatively a "small data" problem. A crucial question for data scientists in this context is the relevance of state-of-the-art data-driven approaches including those based on deep neural networks or kernel-based processes. Here we consider a chaotic system - two-level Lorenz-96 - used as a benchmark model in the climate science literature, adopt a methodology based on Gaussian Processes for parameter estimation and compare the gains in predictive understanding with a suite of Deep Learning and strawman Linear Regression methods. Our results show that adaptations of kernel-based Gaussian Processes can outperform other approaches under small data constraints along with uncertainty quantification; and needs to be considered as a viable approach in climate science and earth system modeling. 
    more » « less
  2. Abstract Microstructure observations in the Pacific cold tongue reveal that turbulence often penetrates into the thermocline, producing hundreds of watts per square meter of downward heat transport during nighttime and early morning. However, virtually all observations of this deep-cycle turbulence (DCT) are from 0°, 140°W. Here, a hierarchy of ocean process simulations, including submesoscale-permitting regional models and turbulence-permitting large-eddy simulations (LES) embedded in a regional model, provide insight into mixing and DCT at and beyond 0°, 140°W. A regional hindcast quantifies the spatiotemporal variability of subsurface turbulent heat fluxes throughout the cold tongue from 1999 to 2016. Mean subsurface turbulent fluxes are strongest (∼100 W m −2 ) within 2° of the equator, slightly (∼10 W m −2 ) stronger in the northern than Southern Hemisphere throughout the cold tongue, and correlated with surface heat fluxes ( r 2 = 0.7). The seasonal cycle of the subsurface heat flux, which does not covary with the surface heat flux, ranges from 150 W m −2 near the equator to 30 and 10 W m −2 at 4°N and 4°S, respectively. Aseasonal variability of the subsurface heat flux is logarithmically distributed, covaries spatially with the time-mean flux, and is highlighted in 34-day LES of boreal autumn at 0° and 3°N, 140°W. Intense DCT occurs frequently above the undercurrent at 0° and intermittently at 3°N. Daily mean heat fluxes scale with the bulk vertical shear and the wind stress, which together explain ∼90% of the daily variance across both LES. Observational validation of the scaling at 0°, 140°W is encouraging, but observations beyond 0°, 140°W are needed to facilitate refinement of mixing parameterization in ocean models. Significance Statement This work is a fundamental contribution to a broad community effort to improve global long-range weather and climate forecast models used for seasonal to longer-term prediction. Much of the predictability on seasonal time scales is derived from the slow evolution of the upper eastern equatorial Pacific Ocean as it varies between El Niño and La Niña conditions. This study presents state-of-the-art high-resolution regional numerical simulations of ocean turbulence and mixing in the eastern equatorial Pacific. The results inform future planning for field work as well as future efforts to refine the representation of ocean mixing in global forecast models. 
    more » « less
  3. Abstract

    Most machine learning applications in Earth system modeling currently rely on gradient‐based supervised learning. This imposes stringent constraints on the nature of the data used for training (typically, residual time tendencies are needed), and it complicates learning about the interactions between machine‐learned parameterizations and other components of an Earth system model. Approaching learning about process‐based parameterizations as an inverse problem resolves many of these issues, since it allows parameterizations to be trained with partial observations or statistics that directly relate to quantities of interest in long‐term climate projections. Here, we demonstrate the effectiveness of Kalman inversion methods in treating learning about parameterizations as an inverse problem. We consider two different algorithms: unscented and ensemble Kalman inversion. Both methods involve highly parallelizable forward model evaluations, converge exponentially fast, and do not require gradient computations. In addition, unscented Kalman inversion provides a measure of parameter uncertainty. We illustrate how training parameterizations can be posed as a regularized inverse problem and solved by ensemble Kalman methods through the calibration of an eddy‐diffusivity mass‐flux scheme for subgrid‐scale turbulence and convection, using data generated by large‐eddy simulations. We find the algorithms amenable to batching strategies, robust to noise and model failures, and efficient in the calibration of hybrid parameterizations that can include empirical closures and neural networks.

     
    more » « less
  4. Abstract

    Because of their limited spatial resolution, numerical weather prediction and climate models have to rely on parameterizations to represent atmospheric turbulence and convection. Historically, largely independent approaches have been used to represent boundary layer turbulence and convection, neglecting important interactions at the subgrid scale. Here we build on an eddy‐diffusivity mass‐flux (EDMF) scheme that represents all subgrid‐scale mixing in a unified manner, partitioning subgrid‐scale fluctuations into contributions from local diffusive mixing and coherent advective structures and allowing them to interact within a single framework. The EDMF scheme requires closures for the interaction between the turbulent environment and the plumes and for local mixing. A second‐order equation for turbulence kinetic energy (TKE) provides one ingredient for the diffusive local mixing closure, leaving a mixing length to be parameterized. Here, we propose a new mixing length formulation, based on constraints derived from the TKE balance. It expresses local mixing in terms of the same physical processes in all regimes of boundary layer flow. The formulation is tested at a range of resolutions and across a wide range of boundary layer regimes, including a stably stratified boundary layer, a stratocumulus‐topped marine boundary layer, and dry convection. Comparison with large eddy simulations (LES) shows that the EDMF scheme with this diffusive mixing parameterization accurately captures the structure of the boundary layer and clouds in all cases considered.

     
    more » « less
  5. Gas exchange between the atmosphere and ocean interior profoundly impacts global climate and biogeochemistry. However, our understanding of the relevant physical processes remains limited by a scarcity of direct observations. Dissolved noble gases in the deep ocean are powerful tracers of physical air-sea interaction due to their chemical and biological inertness, yet their isotope ratios have remained underexplored. Here, we present high-precision noble gas isotope and elemental ratios from the deep North Atlantic (~32°N, 64°W) to evaluate gas exchange parameterizations using an ocean circulation model. The unprecedented precision of these data reveal deep-ocean undersaturation of heavy noble gases and isotopes resulting from cooling-driven air-to-sea gas transport associated with deep convection in the northern high latitudes. Our data also imply an underappreciated and large role for bubble-mediated gas exchange in the global air-sea transfer of sparingly soluble gases, including O 2 , N 2 , and SF 6 . Using noble gases to validate the physical representation of air-sea gas exchange in a model also provides a unique opportunity to distinguish physical from biogeochemical signals. As a case study, we compare dissolved N 2 /Ar measurements in the deep North Atlantic to physics-only model predictions, revealing excess N 2 from benthic denitrification in older deep waters (below 2.9 km). These data indicate that the rate of fixed N removal in the deep Northeastern Atlantic is at least three times higher than the global deep-ocean mean, suggesting tight coupling with organic carbon export and raising potential future implications for the marine N cycle. 
    more » « less