メモリアル大での研究のメモ

Dropboxを整理しておくといろんなものが出てくる。

中にNEMOモデルを使ってラブラドール海のモデルシミュレーションをやっていた頃の過程ノートがテキストファイルであったので削除する前にここに載せておく。

以下stn_note.txt

USE:
namelist_pisces
voddmavs or votkeavt
atcco2=280 (349ppm in 1987, but use 280 for equilibrium state)
nrdttrc=1
ln_pisdmp=true (relaxation) ---> currently works only with "orca" configuration and NOT 1D. Thus, this does nothing for my case.
nn_pisdmp=end of year
namelist_top
ln_trcdta=true
ln_trcdmp=true (damping)!
nn_hdmp_tr=89
namelist
rn_rdt<86400 (1day)
nn_msh=1 (mesh file)
ln_dynwzv=false (read w from file)
stn15/

case3+: dt=6hr
case3333&33333:no initialization
---> unrealistic value of NO3 (x2 greater)

case10+:dt=1day
10years simulation

10&10_1:no initialization
10:no damp or relax
10_1:damp & relax on every 365days (whole column)
10_2&10_3:seems to be the same...
10_3: same as 10_1 except initialization used.
10_4:same as 10_3 except constant & weak eddy diffusivity (10^-5)
10_5:same as 10_3 except no damping or relax.

case30+:30 years simulation
30_1:same as 10_3 except 280ppm & 30 years

Tracer damping or relaxation do not work in 1D...

DONE:
deleted: 10_2, 10_4,10_5


Above setups are all WRONG UP UNTIL NOW because the mesh_mask.nc was wrong.

case30_2: dt=1day; 30 years; 280ppm
case10_6: dt_1hr; 10 years; 280ppm (hopefully this will improve case30_2 since including diurnal cycle)


case1_1: dt=1hr; 1yr;
case1_2: dt=1hr; 2yr; <-- I want to see if model results improve after 1st year. --> can't see it very clearly :(
case1_3: dt=1hr; 10yr; --> it drifts away (same as case2_1).
case2_1: dt=1day; 10yr; <-- I want to see if model results improve with time resolution. --> no, rather it drifts away.

With the new physical data:

case3_1: dt=1hr; 2yr; (same as case2_1, but new physical data)

run0: This is the old physical data, which I might use?
run1: This is what I use for my thesis. conditions are:

280ppm
monthly-mean offline dynamics for 1987


damping???

以下LABSEA_note.txt

case1_3: new offline dynamics: NaN at kt=93; dt=1hr
case1_4: same as case1_3 except damping 1day

==========
beta2:
the model works with dt=100sec
instability at bottom left corner, as well as reduction of DIC in the whole domain (unrealistic)
suggesions: reduce horizontal eddy diffusivity (currently 2000)

beta3:
same as beta2 except reduced horizontal eddy diffusivity in namelist_top (changed to 90)
--> still instability. --> needs to modify the code to make it OBC instead of CBC.


20minutes work!
must use 32 CPUs.