TMItransient

struct EvolvingField

This structure permits the grid to be 
automatically passed to functions with
the evolving tracer field.

This structure assumes the Tracer type to be 
three-dimensional with an additional vector dimension for time.

tracer::Vector{Array{T,3}}
γ::Grid
name::Symbol
longname::String
units::String

function agedistribution

age distribution refers to distribution over lags, not space
sometimes called a transit time distribution

function ces_ncwrite(γ,time,sol_array)
Write .nc file output for commonerasim.jl

Arguments

• γ:
• time: vector of time values
• sol_array: solution array in form time x lat x lon x depth - must match γ + time

Output

• saves .nc file titled "ces_output.nc" in data array

function deltaresponse

Take CDF and turn it into PDF

function globalmean_impulseresponse

Ḡ: satisfied ∫₀^∞ Ḡ(τ) dτ = 1

`alg`: centered or leapfrog

Does leapfrog satisfy normalization?

function globalmean_impulseresponse

based on a globalmean_stepresponse D̄, compute the impulse response
can be done via centered or leapfrog difference

function globalmean_stepresponse

calculate the global mean response to "turning on" some region

function gooddata
a useful one-liner

 read surface layer

function  s_array(sol, γ)
Converts from DE.jl output to time x lat x lon x depth

function setupODE(γ, u0,tsfc, dsfc,bc,L, t_int)
Setup ODEFunction for LC+Bf with no Jacobian

function stepresponse

calculate the response to "turning on" some region
can compute some statistics on output by providing a function to f 

# Arguments
- TMIversion
- b: BoundaryCondition
- γ
- L
- B
- τ: evenly spaced vector
- f: some function f(u) where u is a vector of all wet points

function varying!(du, u, p, t)
ODE function for varying boundary cond
Sets up dc/dt = L*C + B*f to be solved

Arguments

• du: dc/dt (must have this name for DifferentialEquations.jl to work
• u: C, what we are solving for
• p: parameters for diffeq - must hold specified vars
• t: time we are solving for (automatically determined by DE.jl)

Output

• du: numerical value of LC+Bf, vector of size 74064 for 4°

function vintagedistribution(t₀,tf,Δ,τ,tmodern=2022,interp="linear")

percentage of water in the modern ocean
from a vintage defined by the calendar year interval
t₀ [cal yr CE] => tf [cal yr CE]

Arguments

• t₀: Starting calendar year of vintage, e.g., 1450 CE
• tf: final calendar year of vintage, e.g., 1850 CE
• Δ::Vector{Field}: Step function response
• τ = Vector{Float64}: time lags associated with step response
• tmodern=2022: modern calendar year
• interp=linear: or can be "spline"

Output

• g::Field: 3D distribution of vintage contribution

Warning

• should be a way to make Δ argument more general (more types)