Ocean Biogeochemical Modelling Environment

Description

OceanBioME is a flexible biogeochemical modelling environment written in Julia for modelling the coupled interactions between ocean biology, carbonate chemistry, and physics. OceanBioME can be run as a stand-alone box model, or coupled with Oceananigans.jl to run as a 1D column model or with 2 and 3D physics.

OceanBioME was developed with generous support from the Centre for Climate Repair CCR and the Gordon and Betty Moore Foundation as a tool to study the effectiveness and impacts of ocean carbon dioxide removal (CDR) strategies.

Installation:

First, download and install Julia

From the Julia prompt (REPL), type:

julia> using Pkg
julia> Pkg.add("OceanBioME")

Running your first model

As a simple example lets run a Nutrient-Phytoplankton-Zooplankton-Detritus (NPZD) model in a two-dimensional simulation of a buoyancy front. This example requires Oceananigans, so we install that first:

using Pkg; Pkg.add("Oceananigans")

using OceanBioME, Oceananigans
using Oceananigans.Units

grid = RectilinearGrid(CPU(), size = (160, 32), extent = (10000meters, 500meters), topology = (Bounded, Flat, Bounded))

biogeochemistry = NutrientPhytoplanktonZooplanktonDetritus(; grid) 

model = NonhydrostaticModel(; grid, biogeochemistry,
                              advection = WENO(),
                              closure = AnisotropicMinimumDissipation(),
			      buoyancy = SeawaterBuoyancy(constant_salinity = true))

@inline front(x, z, μ, δ) = μ + δ * tanh((x - 7000 + 4 * z) / 500)

Pᵢ(x, z) = ifelse(z > -50, 0.03, 0.01)
Nᵢ(x, z) = front(x, z, 2.5, -2)
Tᵢ(x, z) = front(x, z, 9, 0.05)

set!(model, N = Nᵢ, P = Pᵢ, Z = Pᵢ, T = Tᵢ)

simulation = Simulation(model; Δt = 50, stop_time = 4days)

simulation.output_writers[:tracers] = JLD2Writer(model, model.tracers,
                                                 filename = "buoyancy_front.jld2",
                                                 schedule = TimeInterval(24minute),
                                                 overwrite_existing = true)

run!(simulation)

<details> <summary>We can then visualise this:</summary>

# Before running the visualization code below, make sure CairoMakie is installed:
using Pkg; Pkg.add("CairoMakie")

T = FieldTimeSeries("buoyancy_front.jld2", "T")
N = FieldTimeSeries("buoyancy_front.jld2", "N")
P = FieldTimeSeries("buoyancy_front.jld2", "P")

xc, yc, zc = nodes(T)

times = T.times

using CairoMakie

n = Observable(1)

T_lims = (8.94, 9.06)
N_lims = (0, 4.5)
P_lims = (0.007, 0.02)

Tₙ = @lift interior(T[$n], :, 1, :)
Nₙ = @lift interior(N[$n], :, 1, :)
Pₙ = @lift interior(P[$n], :, 1, :)

fig = Figure(size = (1000, 520), fontsize = 20)

title = @lift "t = $(prettytime(times[$n]))"
Label(fig[0, :], title)

axis_kwargs = (xlabel = "x (m)", ylabel = "z (m)", width = 770, yticks = [-400, -200, 0])
ax1 = Axis(fig[1, 1]; title = "Temperature (°C)", axis_kwargs...)
ax2 = Axis(fig[2, 1]; title = "Nutrients concentration (mmol N / m³)",axis_kwargs...)
ax3 = Axis(fig[3, 1]; title = "Phytoplankton concentration (mmol N / m³)", axis_kwargs...)

hm1 = heatmap!(ax1, xc, zc, Tₙ, colorrange = T_lims, colormap = Reverse(:lajolla), interpolate = true)
hm2 = heatmap!(ax2, xc, zc, Nₙ, colorrange = N_lims, colormap = Reverse(:bamako), interpolate = true)
hm3 = heatmap!(ax3, xc, zc, Pₙ, colorrange = P_lims, colormap = Reverse(:bamako), interpolate = true)

Colorbar(fig[1, 2], hm1, ticks = [8.95, 9.0, 9.05])
Colorbar(fig[2, 2], hm2, ticks = [0, 2, 4])
Colorbar(fig[3, 2], hm3, ticks = [0.01, 0.02, 0.03])

rowgap!(fig.layout, 0)

record(fig, "buoyancy_front.gif", 1:length(times)) do i
    n[] = i
end

</details>

https://github.com/OceanBioME/OceanBioME.jl/assets/26657828/d4a5dbc9-ffff-4ef0-8431-b9afc951142f

In this example OceanBioME is providing the biogeochemistry and the remainder is taken care of by Oceananigans. For comprehensive documentation of the physics modelling see Oceananigans' Documentation, and for biogeochemistry and other features we provide read below.

Using GPU

To run the same example on a GPU we just need to construct the grid on the GPU; the rest is taken care of!

Just replace CPU() with GPU() in the grid construction with everything else left unchanged:

grid = RectilinearGrid(GPU(), size = (256, 32), extent = (500meters, 100meters), topology = (Bounded, Flat, Bounded))

Documentation

See the documentation for full description of the software package and more examples, as well as full descriptions of the included models and parametrisations.

Contributing

If you're interested in contributing to the development of OceanBioME we would appreciate your help!

If you'd like to work on a new feature, or if you're new to open source and want to crowd-source projects that fit your interests, please start a discussion.

For more information check out our contributor's guide.

Citing

If you use OceanBioME as part of your research, teaching, or other activities, we would be grateful if you could cite our work below and mention the package by name.

@article{OceanBioMEJOSS,
  doi = {10.21105/joss.05669},
  url = {https://doi.org/10.21105/joss.05669},
  year = {2023},
  publisher = {The Open Journal},
  volume = {8},
  number = {90},
  pages = {5669},
  author = {Jago Strong-Wright and Si Chen and Navid C. Constantinou and Simone Silvestri and Gregory LeClaire Wagner and John R. Taylor},
  title = {{OceanBioME.jl: A flexible environment for modelling the coupled interactions between ocean biogeochemistry and physics}},
  journal = {Journal of Open Source Software}
}

If on top of citing the JOSS paper above, you need to cite a specific version of the package then please cite its corresponding version from the Zenodo archive.