skip to main content


Title: Machine Learning for Improving Surface-Layer-Flux Estimates
Abstract

Flows in the atmospheric boundary layer are turbulent, characterized by a large Reynolds number, the existence of a roughness sublayer and the absence of a well-defined viscous layer. Exchanges with the surface are therefore dominated by turbulent fluxes. In numerical models for atmospheric flows, turbulent fluxes must be specified at the surface; however, surface fluxes are not known a priori and therefore must be parametrized. Atmospheric flow models, including global circulation, limited area models, and large-eddy simulation, employ Monin–Obukhov similarity theory (MOST) to parametrize surface fluxes. The MOST approach is a semi-empirical formulation that accounts for atmospheric stability effects through universal stability functions. The stability functions are determined based on limited observations using simple regression as a function of the non-dimensional stability parameter representing a ratio of distance from the surface and the Obukhov length scale (Obukhov in Trudy Inst Theor Geofiz AN SSSR 1:95–115, 1946),$$z/L$$z/L. However, simple regression cannot capture the relationship between governing parameters and surface-layer structure under the wide range of conditions to which MOST is commonly applied. We therefore develop, train, and test two machine-learning models, an artificial neural network (ANN) and random forest (RF), to estimate surface fluxes of momentum, sensible heat, and moisture based on surface and near-surface observations. To train and test these machine-learning algorithms, we use several years of observations from the Cabauw mast in the Netherlands and from the National Oceanic and Atmospheric Administration’s Field Research Division tower in Idaho. The RF and ANN models outperform MOST. Even when we train the RF and ANN on one set of data and apply them to the second set, they provide more accurate estimates of all of the fluxes compared to MOST. Estimates of sensible heat and moisture fluxes are significantly improved, and model interpretability techniques highlight the logical physical relationships we expect in surface-layer processes.

 
more » « less
NSF-PAR ID:
10372408
Author(s) / Creator(s):
; ; ; ; ; ;
Publisher / Repository:
Springer Science + Business Media
Date Published:
Journal Name:
Boundary-Layer Meteorology
Volume:
185
Issue:
2
ISSN:
0006-8314
Page Range / eLocation ID:
p. 199-228
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Abstract

    The pre-ALPACA (Alaskan Layered Pollution And Chemical Analysis) 2019 winter campaign took place in Fairbanks, Alaska, in November–December 2019. One objective of the campaign was to study the life-cycle of surface-based temperature inversions and the associated surface energy budget changes. Several instruments, including a 4-component radiometer and sonic anemometer were deployed in the open, snow-covered University of Alaska Fairbanks (UAF) Campus Agricultural Field. A local flow from a connecting valley occurs at this site. This flow is characterized by locally elevated wind speeds (greater than 3 m s$$^{-1}$$-1) under clear-sky conditions and a north-westerly direction. It is notably different to the wind observed at the airport more than 3.5 km to the south-west. The surface energy budget at the UAF Field site exhibits two preferential modes. In the first mode, turbulent sensible heat and net longwave fluxes are close to 0 W m$$^{-2}$$-2, linked to the presence of clouds and generally low winds. In the second, the net longwave flux is around − 50 W m$$^{-2}$$-2and the turbulent sensible heat flux is around 15 W m$$^{-2}$$-2, linked to clear skies and elevated wind speeds. The development of surface-based temperature inversions at the field is hindered compared to the airport because the local flow sustains vertical mixing. In this second mode the residual of the surface energy budget is large, possibly due to horizontal temperature advection.

     
    more » « less
  2. Abstract

    The oceanic bottom mixed layer (BML) is a well mixed, weakly stratified, turbulent boundary layer. Adjacent to the seabed, the BML is of intrinsic importance for studying ocean mixing, energy dissipation, particle cycling and sediment-water interactions. While deep-seabed mining of polymetallic nodules is anticipated to commence in the Clarion-Clipperton Zone (CCZ) of the northeastern tropical Pacific Ocean, knowledge gaps regarding the form of the BML and its potentially key influence on the dispersal of sediment plumes generated by deep-seabed mining activities are yet to be addressed. Here, we report recent field observations from the German mining licence area in the CCZ that characterise the structure and variability of the BML locally. Quasi-uniform profiles of potential temperature extending from the seafloor reveal the presence of a spatially and temporally variable BML with an average local thickness of approximately 250 m. Deep horizontal currents in the region have a mean speed of 3.5 cm s$$^{-1}$$-1and a maximum speed of 12 cm s$$^{-1}$$-1at 18.63 ms above bottom over an 11 month record. The near-bottom currents initially have a net southeastward flow, followed by westward and southward flows with the development of complex, anticyclonic flow patterns. Theoretical predictions and historical data show broad consistency with mean BML thickness but cannot explain the observed heterogeneity of local BML thickness. We postulate that deep pressure anomalies induced by passing surface mesoscale eddies and abyssal thermal fronts could affect BML thickness, in addition to local topographic effects. A simplified transport model is then used to study the influence of the BML on the interplay between turbulent diffusion and sediment settling in the transport of deep-seabed mining induced sediment plumes. Over a range of realistic parameter values, the effects of BML on plume evolution can vary significantly, highlighting that resolving the BML will be a crucial step for accurate numerical modelling of plume dispersal.

     
    more » « less
  3. Abstract

    Given a suitable solutionV(tx) to the Korteweg–de Vries equation on the real line, we prove global well-posedness for initial data$$u(0,x) \in V(0,x) + H^{-1}(\mathbb {R})$$u(0,x)V(0,x)+H-1(R). Our conditions onVdo include regularity but do not impose any assumptions on spatial asymptotics. We show that periodic profiles$$V(0,x)\in H^5(\mathbb {R}/\mathbb {Z})$$V(0,x)H5(R/Z)satisfy our hypotheses. In particular, we can treat localized perturbations of the much-studied periodic traveling wave solutions (cnoidal waves) of KdV. In the companion paper Laurens (Nonlinearity. 35(1):343–387, 2022.https://doi.org/10.1088/1361-6544/ac37f5) we show that smooth step-like initial data also satisfy our hypotheses. We employ the method of commuting flows introduced in Killip and Vişan (Ann. Math. (2) 190(1):249–305, 2019.https://doi.org/10.4007/annals.2019.190.1.4) where$$V\equiv 0$$V0. In that setting, it is known that$$H^{-1}(\mathbb {R})$$H-1(R)is sharp in the class of$$H^s(\mathbb {R})$$Hs(R)spaces.

     
    more » « less
  4. Abstract

    We revisit the longstanding problem of grid sensitivity, i.e., the lack of grid convergence in large-eddy simulations (LES) of the stable boundary layer. We use a comprehensive set of LES of the well-known Global Energy and Water Cycle Experiment Atmospheric Boundary Layer Study 1 (GABLS1) case with varying grid spacings between 12.5 m and 1 m to investigate several physical processes and numerical features that are possible causes of grid sensitivity. Our results demonstrate that there are two resolution regimes in which grid sensitivity manifests differently. We find that changing the numerical advection schemes and the subgrid-scale models alters the simulation results, but the options tested do not fully address the grid-sensitivity issue. Moreover, sensitivity runs suggest that the surface boundary condition and the interaction of the surface with the near-surface flow, as well as the mixing with the free atmosphere, are unlikely to be the causes of the observed grid sensitivity. One interesting finding is that the grid sensitivity in the fine grid-spacing regime (grid spacings$$\le 2\,\mathrm{m}$$2m) is closely related to the reduction in the energy content of large-scale turbulence, leading to less turbulence kinetic energy and hence lower boundary-layer heights. The present work demonstrates that there is still an urgent need to address this grid-sensitivity issue in order to perform reliable LES of the stable boundary layer.

     
    more » « less
  5. Abstract

    We study the classical Hénon family$$f_{a,b}:(x,y)\mapsto (1-ax^2+y,bx)$$fa,b:(x,y)(1-ax2+y,bx),$$00<a<2,$$00<b<1, and prove that given an integer$$k\ge 1$$k1, there is a set of parameters$$E_k$$Ekof positive two-dimensional Lebesgue measure so that$$f_{a,b}$$fa,b, for$$(a,b)\in E_k$$(a,b)Ek, has at leastkattractive periodic orbits and one strange attractor of the type studied in Benedicks and Carleson (Ann Math (2) 133(1):73–169, 1991). A corresponding statement also holds for the Hénon-like families of Mora and Viana (Acta Math 171:1–71, 1993), and we use the techniques of Mora and Viana (1993) to study homoclinic unfoldings also in the case of the original Hénon maps. The final main result of the paper is the existence, within the classical Hénon family, of a positive Lebesgue measure set of parameters whose corresponding maps have two coexisting strange attractors.

     
    more » « less