search

gher-uliege / DIVAnd.jl

DIVAnd performs an n-dimensional variational analysis of arbitrarily located observations
GNU General Public License v2.0
70 stars 11 forks source link

Fix typo in DIVAndrun.jl docstring #69

Closed briochemc closed 4 years ago

coveralls commented 4 years ago

Coverage Status

Coverage decreased (-0.03%) to 78.605% when pulling 78a7ddff06fe0e141f66c0b632d2f2ed0ff5e2bf on briochemc:patch-1 into ab2e9125c9e6e6e8c847acd0e36d27d12894981b on gher-ulg:master.

Alexander-Barth commented 4 years ago

Thanks a lot @briochemc !

briochemc commented 4 years ago

No worries, I just tried DIVAnd for interpolating a global geological map and it worked pretty nicely! BTW, I have a question regarding this. If I use DIVAnd_metric to weight observations using the distance on the globe, as in

pm, pn = DIVAnd.DIVAnd_metric(xi, yi)

Do I need to impose longitudinal cyclicality via, e.g., moddim=[360,0]? It seems to me the results are different with and without moddim but I fail to see why in this case, since the weights are constructed with cyclical longitude embedded in DIVAnd_metric. Could you clarify this for me? (Maybe they are the same and my plotting is bad though).

jmbeckers commented 4 years ago

_metric ensures distances gridboxes have the correct sizes and which are later use in derivative estimates.

the role of moddim is different: it ensures the domain is actually periodic in the direction you specify by connecting the right points during the derivative calculations.

So, yes, you need to activate moddim to force a periodic domain.

briochemc commented 4 years ago

Thanks @jmbeckers for your explanation!