LIM2 read air temperature to decide if snow or rain

CORE2で行くのでもう必要ないけどもしいつか必要になった時のために残しておくことにする。

random notes

one way to make existing "time" dimension to be "unlimited": 

cdo seltimestep,1/12 ifile ofile

where 1/12 is the first and last indices of your time record. 

Ref: https://code.zmaw.de/boards/1/topics/798

 1057  cdo seltimestep,1/1 sic_G10010_SIBT1850_v1.1-naa_icecon.nc4 sic.nc
 1058  cdo seltimestep,1/1 sit_G10010_SIBT1850_v1.1-naa_icecon.nc4 sic.nc
 1059  cdo seltimestep,1/1 sit_PIOMAS_1979_2013-naa_arlan.nc4 sit.nc 

Add all variables in sic.nc to sit.nc
 1062  ncks -h -A sic.nc sit.nc 

Remove time_counter dimension from the lists of dimensions and variables. 
 1065  ncwa -a time_counter sit.nc sit2.nc

Same names (labels) but different IDs for nodes in PyGraphviz:

add_node(1,label=A)

add_node(2,label=B)

add_node(3,label=A)

will create three nodes consisting of two As and one B. 

Ref: http://stackoverflow.com/questions/15711909/how-to-create-duplicate-nodes-in-pygraphviz

shaw memo

14 april --> internet 5.
phone disconnect --> immediately.
--> arris --> return to uptown branch.

固定電話とテレビとインターネットのプロバイダーはどこも高くてモノポリー化してるビクトリア :-(

interpolate woa13 no3 field onto NAA grid using SOSIE

I had little trouble interpolating the woa13 no3 data onto NAA grid. SOSIE was creating high values (first figure; set to the prescribed vmax of 100. in this plot) along the coast lines. I was not sure why, but what solved this issue was to change ldrown from F to T:

cf_lsm_in = 'missing_value' ! we use 'missing_value' of input field to determine

cv_lsm_in = ''                ! the land-sea-mask

ldrown    = F            ! we want to propagate sea values onto the land-sea mask

Here, the last line is modified as follows:

ldrown    = T 

Then the interpolation improved (second figure).

Furthermore, the interpolation is not too bad compared to the original file (third figure), although it may underestimate no3 in the beaufort region.

interpolated field with ldrown = F

interpolated field with ldrown = T

comparison between the original 1x1 degree data (left) and the interpolated data (right)

snow and ice melt rate

rdmsnif/rdt_ice is snow melt rate (m/s) averaged over ice covered area, and not over the entire grid cell. Otherwise the code below should include the coefficient "1-frld" in the last term:

limsbc_2.F90

!  computing freshwater exchanges at the ice/ocean interface

            zemp = + emp(ji,jj)     *         frld(ji,jj)      &   !  e-p budget over open ocean fraction 

               &   - tprecip(ji,jj) * ( 1. -  frld(ji,jj) )    &   !  liquid precipitation reaches directly the ocean

               &   + sprecip(ji,jj) * ( 1. - pfrld(ji,jj) )    &   !  change in ice cover within the time step

               &   + rdmsnif(ji,jj) * r1_rdtice                    !  freshwater flux due to snow melting