LIGHT

Lagrangian, In situ, Global, High-Performance Particle Tracking (LIGHT)

Online particle tracking, especially in the Earth System Modeling community, is still relatively rare. However, at Exascale, online particle tracking must be employed to ensure that spatio-temporal fidelity of Lagrangian particle tracks is ensured, especially given the severe performance cost of reading from large data files from disk1. To this end, the Lagrangian, in Situ, Global, High-Performance Particle Tracking (LIGHT) analysis module2 of the Model for Prediction Across Scales Ocean (MPAS-O)3 was developed to provide an online approach to resolving the Exascale challenge of online particle tracking within a high-resolution global ocean model simulation.

LIGHT computes trajectories on an individual particle basis by interpolating velocity fields vertically, depending upon desired particle advection (e.g., isopycnally-constrained), and then performing horizontal interpolations using Wachspress interpolation. Temporal integration uses generalized Runge Kutta methods. Both computation and input and output are fully parallelized, allowing use on the highest-resolution ocean simulations.

LIGHT was previously featured highlighted in an E3SM technical highlight.

New LIGHT biogeochemistry capability videos

These recent LIGHT videos by CSGF summer student Riley Brady highlight use of LIGHT’s new biogeochemsitry capabilities.

Surface BGC Floats in MPAS Ocean from Riley Brady on Vimeo.

The Flow of Dissolved Carbon in the Southern Ocean from Riley Brady on Vimeo.

Prior idealized applications of LIGHT

Applications of LIGHT on mixing as quantifid via a diffusivity have been varied, but have primarily focused on use of LIGHT to compute fluid mixing along isopycnals via an isopycnal diffusivity2, e.g., Fig. 1.

Figure 1: Idealized Gulf Stream particle statistics, where the black arrows denote the mean path of a particle cluster and the gray area is the spread of the particles. Diffusivity is the rate at which the gray line grows normal to the black line.
Figure 1: Idealized Gulf Stream particle statistics, where the black arrows denote the mean path of a particle cluster and the gray area is the spread of the particles. Diffusivity is the rate at which the gray line grows normal to the black line.

Resolution and scale dependence of diffusivity in an idealized, eddying mid-latitude basin were computed. The results demonstrat that the largest eddies most strongly contribute to mixing, but these eddies require a grid resolution of at least a half the Rossby Radius of Deformation2. Diagnosis of mean, eddy, and residual eddy-mean flow interactions on diffusivity in a zonal idealized Southern Ocean indicate that the mean flow and eddies nonlinearly enhance diffusivity4. A novel application of online-computed particle trajectories is to compute post facto Lagrangian scalar transport, which can be used to compute a Lagrangian effective diffusivity5.

LIGHT in high-resolution, global climate modeling applications

Figure 2: Global application of Lagrangian particle trajectories using LIGHT.
Figure 2: Global application of Lagrangian particle trajectories using LIGHT.

Department of Energy Leadership Class Facility computing, e.g., Theta, has been used to compute online LIGHT trajectories for the 30km to 10km global MPAS-O resolution, using the CORE-II climate forcing dataset6,7. A schematic of the global Lagrangian particle tracking using LIGHT is shown in Fig. 2. The current simulation uses 16 million particles on 8192 processors. Furture simulations with LIGHT will be performed on the highest-resolution 18km to 6 km MPAS-O mesh. LIGHT is included as an in-situ analysis member of MPAS-O.

References

1. van Sebille, E. & others. Lagrangian analysis of ocean velocity data: Fundamentals and practices. Ocean Modell. (2018). doi:10.1016/j.ocemod.2017.11.008

2. Wolfram, P.J., Ringler, T., Maltrud, M., Jacobsen, D. & Petersen, M. Diagnosing isopycnal diffusivity in an eddying, idealized mid-latitude ocean basin via Lagrangian In-situ, Global, High-performance particle Tracking (LIGHT). J. Phys. Oceanogr. 45, 2114–2133 (2015).

3. Ringler, T., Petersen, M., Higdon, R. L., Jacobsen, D., Jones, P. W. & Maltrud, M. A multi-resolution approach to global ocean modeling. Ocean Modell. 69, 211–232 (2013).

4. Wolfram, P. J. & Ringler, T. D. Quantifying residual, eddy, and mean flow effects on mixing in an idealized circumpolar current. J. Phys. Oceanogr. 47, 1897–1920 (2017).

5. Wolfram, P. J. & Ringler, T. D. Computing eddy-driven effective diffusivity using Lagrangian particles. Ocean Modell. 118, 94–106 (2017).

6. Large, W. G. & Yeager, S. The global climatology of an interannually varying air–sea flux data set. Clim. Dyn. 33, 341–364 (2009).

7. Griffies, S. M., Yin, J., Durack, P. J., Goddard, P., Bates, S. C., Behrens, E., Bentsen, M., Bi, D., Biastoch, A., Böning, C. W. & others. An assessment of global and regional sea level for years 1993–2007 in a suite of interannual CORE-II simulations. Ocean Modell. 78, 35–89 (2014).