To-do list

[ggebbie]

• [x] extract timeseries at any grid point

  1. using https://github.com/ggebbie/TMI.jl
  2. read just one time snapshot at any time
  3. save as a Vector{}: (can be Float32?)

• [x] extract timeseries at any location, nothing if land, interpolate to point if wet

       1. read one snapshot at a time
       2. use `TMI.observe` to pick a sample point https://github.com/ggebbie/TMI.jl/blob/9ef4ce01f89713a757b6e83ce7ea12f9c700f735/src/TMI.jl#L1329
       3. use observe like this function https://github.com/ggebbie/TMI.jl/blob/9ef4ce01f89713a757b6e83ce7ea12f9c700f735/src/TMI.jl#L1273 

• [x] handle (-t)-z-y-x convention for 4D file contents

• [ ] compute global ocean heat content using some TMI utilities

            1.         get volume from `volumefilled` function https://github.com/ggebbie/TMI.jl/blob/9ef4ce01f89713a757b6e83ce7ea12f9c700f735/src/TMI.jl  
            2.   can approximate density
            3. later think about adding GibbsSeaWater.jl for density
            4. use `\gamma.wet` mask to make sums without searching for NaNs    \
            5. example script `ex5.surfaceorigin.jl`:   https://github.com/ggebbie/TMI.jl/blob/9ef4ce01f89713a757b6e83ce7ea12f9c700f735/src/TMI.jl

• [ ] compute regional ocean heat content using masks/basin info, etc.

  1. TMI has surface boundary masks, use for interior ocean?
  2. example is function surfaceregion https://github.com/ggebbie/TMI.jl/blob/9ef4ce01f89713a757b6e83ce7ea12f9c700f735/src/TMI.jl#L163

• [ ] transfer ocean heat content to heat convergence and effective planetary heat flux

• [ ] ...

[anthony-meza]

Addressed Items 1-3 in my branch here

One issue with my solution is that it relies on the Field struct, which only accepts "tracer" fields of Array{T, 3}. Likely inefficient. Can we extend tracers to Array{T, N}?

[ggebbie]

thank you, @anthony-meza! I'm happy to make an update. Is this a change to be made in a dependency somewhere? What is the case with 4+ dimensional arrays? When time is added as a dimension?

[anthony-meza]

@ggebbie The change to the Field struct would have to be made in TMI.jl. Yes, the case with 4D arrays arises when time is added as a dimension as in GH19.jl.