1 |
|
Example: Labrador Sea Region with Sea-Ice |
2 |
|
========================================= |
3 |
|
|
4 |
|
This example sets up a small (20x16x23) Labrador Sea experiment |
5 |
|
coupled to a dynamic thermodynamic sea-ice model. |
6 |
|
A brief description of the sea-ice model is in "seaice.ps". |
7 |
|
|
8 |
|
The domain of integration spans 280E to 320E and 46N to 78N. |
9 |
|
Horizontal grid spacing is 2 degrees. |
10 |
|
The 23 vertical levels and the bathymetry file |
11 |
|
bathyFile = 'bathy.labsea' |
12 |
|
are obtained from the the 2-degree ECCO configuration. |
13 |
|
|
14 |
|
Integration is initialized from annual-mean Levitus climatology |
15 |
|
hydrogThetaFile = 'LevCli_temp.labsea' |
16 |
|
hydrogSaltFile = 'LevCli_salt.labsea' |
17 |
|
|
18 |
|
Surface salinity relaxation is to the monthly mean Levitus climatology |
19 |
|
saltClimFile = 'SSS.labsea' |
20 |
|
|
21 |
|
Forcing files are a 1979-1999 monthly climatology computed from the |
22 |
|
NCEP reanalysis (see pkg/seaice/SEAICE_FFIELDS.h for units and signs) |
23 |
|
gairxFile = 'u10m.labsea79' # 10-m zonal wind |
24 |
|
gairyFile = 'v10m.labsea79' # 10-m meridional wind |
25 |
|
tairFile = 'tair.labsea1979' # 2-m air temperature |
26 |
|
qaFile = 'qa.labsea1979' # 2-m specific humidity |
27 |
|
floFile = 'flo.labsea1979' # longwave radiation |
28 |
|
fshFile = 'fsh.labsea1979' # shortwave radiation |
29 |
|
rainFile = 'prate.labsea1979' # precipitation |
30 |
|
evapFile = 'evap.labsea1979' # evaporation |
31 |
|
|
32 |
|
|
33 |
|
Using testscript to test sea-ice code |
34 |
|
===================================== |
35 |
|
|
36 |
|
Running the testscript experiment: |
37 |
|
cd verification |
38 |
|
./testscript -force lab_sea |
39 |
|
|
40 |
|
Note that fairly large differences in accuracy occur across different |
41 |
|
platforms. For example, testscript comparisons between g77 (Linux) |
42 |
|
and f77 (SGI) generated output gives: |
43 |
|
|
44 |
|
T S U V |
45 |
|
C D M c m s m s m s m s |
46 |
|
n p a R g m m e . m m e . m m e . m m e . |
47 |
|
f n k u 2 i a a d i a a d i a a d i a a d |
48 |
|
g d e n d n x n . n x n . n x n . n x n . |
49 |
|
|
50 |
|
Y Y Y Y 5 5 7 7 7 8 10 9 6 6 6 6 7 5 7 5 7 FAIL lab_sea |
51 |
|
|
52 |
|
|
53 |
|
Instructions for generating 1-CPU and 2-CPU executables |
54 |
|
======================================================= |
55 |
|
|
56 |
|
Generating 1-CPU executable: |
57 |
|
cd ../verification/lab_sea/input |
58 |
|
ln -sf ../code/SIZE.h . |
59 |
|
ln -sf ../code/CPP_OPTIONS.h . |
60 |
|
ln -sf ../code/CPP_EEOPTIONS.h . |
61 |
|
../../../tools/genmake -makefile |
62 |
|
==> on alhena use: |
63 |
|
../../../tools/genmake -platform=o2k_noopt -makefile |
64 |
|
make clean |
65 |
|
make depend |
66 |
|
make |
67 |
|
mv mitgcmuv mitgcmuv_1 |
68 |
|
|
69 |
|
Generating 2-CPU executable: |
70 |
|
cd ../../../verification/lab_sea/input |
71 |
|
ln -sf ../code/SIZE_2x1.h SIZE.h |
72 |
|
ln -sf ../code/CPP_OPTIONS.h . |
73 |
|
ln -sf ../code/CPP_EEOPTIONS_MPI.h CPP_EEOPTIONS.h |
74 |
|
../../../tools/genmake -mpi -makefile |
75 |
|
==> on alhena for comparison purposes use: |
76 |
|
../../../tools/genmake -mpi -platform=o2k_noopt -makefile |
77 |
|
==> on alhena for fast execution use: |
78 |
|
../../../tools/genmake -mpi -platform=o2k -makefile |
79 |
|
make clean |
80 |
|
make depend |
81 |
|
make |
82 |
|
mv mitgcmuv mitgcmuv_2x1 |
83 |
|
|
84 |
|
|
85 |
|
Instructions for running Experiment 1 |
86 |
|
===================================== |
87 |
|
|
88 |
|
This is a 1-cpu, 10-hour integration used to make sure that all the |
89 |
|
files are available and that the model compiles and integrates. It |
90 |
|
is the default experiment of "verification/testscript lab_sea". |
91 |
|
|
92 |
|
To run Experiment 1: |
93 |
|
cd ../../../verification/lab_sea/input |
94 |
|
ln -sf data.10hours data |
95 |
|
ln -sf data.seaice.adi data.seaice |
96 |
|
mitgcmuv_1 >&! output.txt |
97 |
|
mkdir exp1 |
98 |
|
mv *tave.0000000010.data exp1 |
99 |
|
|
100 |
|
Use matlab script lookat_exp1.m to compare the output |
101 |
|
of exp1 with that from release1_patch5 sea-ice code: |
102 |
|
cd ../../../verification/lab_sea/matlab |
103 |
|
matlab |
104 |
|
lookat_exp1 |
105 |
|
|
106 |
|
|
107 |
|
Instructions for running Experiment 2 |
108 |
|
===================================== |
109 |
|
|
110 |
|
This is a 1-cpu test of the LSR solver. The solution is |
111 |
|
compared to that of experiment 1, which used the ADI solver. |
112 |
|
|
113 |
|
To run Experiment 2: |
114 |
|
cd ../../../verification/lab_sea/input |
115 |
|
ln -sf data.10hours data |
116 |
|
ln -sf data.seaice.lsr data.seaice |
117 |
|
mitgcmuv_1 >&! output.txt |
118 |
|
mkdir exp2 |
119 |
|
mv *tave.0000000010.data exp2 |
120 |
|
|
121 |
|
Use matlab script lookat_exp2.m to compare |
122 |
|
the output of exp2 to that of exp1: |
123 |
|
cd ../../../verification/lab_sea/matlab |
124 |
|
matlab |
125 |
|
lookat_exp2 |
126 |
|
|
127 |
|
|
128 |
|
Instructions for running Experiment 3 |
129 |
|
===================================== |
130 |
|
|
131 |
|
This is a test of periodic boundary conditions for LSR |
132 |
|
and ADI solvers. The domain has a flat bottom and |
133 |
|
is periodic both in the x and the y directions. |
134 |
|
All forcing files are null or constant (u10m = v10m = 5 m/s). |
135 |
|
|
136 |
|
To run Experiment 3: |
137 |
|
cd ../../../verification/lab_sea/input |
138 |
|
ln -sf data.1hour data |
139 |
|
ln -sf data.seaice.testadi data.seaice |
140 |
|
mitgcmuv_1 >&! output.txt |
141 |
|
mkdir exp3a |
142 |
|
mv *tave.0000000001.data exp3a |
143 |
|
ln -sf data.seaice.testlsr data.seaice |
144 |
|
mitgcmuv_1 >&! output.txt |
145 |
|
mkdir exp3b |
146 |
|
mv *tave.0000000001.data exp3b |
147 |
|
|
148 |
|
Use matlab script lookat_exp3.m to compare |
149 |
|
the output of exp2 to that of exp1: |
150 |
|
cd ../../../verification/lab_sea/matlab |
151 |
|
matlab |
152 |
|
lookat_exp3 |
153 |
|
|
154 |
|
If the dynamic ice solvers are correct, they should converge to a |
155 |
|
constant solution. The tests above show that both the ADI and the LSR |
156 |
|
solver are unable to handle periodic domains and therefore that they |
157 |
|
cannot be "correctly" parallelized. |
158 |
|
|
159 |
|
|
160 |
|
Instructions for running Experiment 4 |
161 |
|
===================================== |
162 |
|
|
163 |
|
This is a 2-cpu, 10-hour integration used to test |
164 |
|
tile edges for sea-ice dynamic solvers. |
165 |
|
|
166 |
|
To run Experiment 4: |
167 |
|
cd ../../../verification/lab_sea/input |
168 |
|
ln -sf data.10hours data |
169 |
|
ln -sf data.seaice.adi data.seaice |
170 |
|
mpirun -np 2 mitgcmuv_2x1 |
171 |
|
mkdir exp4 |
172 |
|
mv *tave.0000000010.data exp4 |
173 |
|
|
174 |
|
Use the matlab script lookat_exp4.m to compare |
175 |
|
the 2-cpu output to that of exp1: |
176 |
|
cd ../../../verification/lab_sea/matlab |
177 |
|
matlab |
178 |
|
lookat_exp4 |
179 |
|
|
180 |
|
The inaccuracy at the tile boundary can be decreased by increasing |
181 |
|
NPSEUDO in data.seaice at the expense of computation time. |
182 |
|
Also as sea-ice ages (thickens) the difference at the tiles becomes |
183 |
|
increasingly small. For solver pkg/seaice/adi.F a value of NPSEUDO=10 |
184 |
|
appears adequate for forward integrations but cannot be used for sea-ice |
185 |
|
adjoint model. Work is underway to fix this problem. |
186 |
|
|
187 |
|
|
188 |
|
Instructions for running Experiment 5 |
189 |
|
===================================== |
190 |
|
|
191 |
|
This is a 2-cpu, 2-year, test integration. It illustrates |
192 |
|
convention used by seaice_get_forcing.F for multi-year forcing. |
193 |
|
|
194 |
|
To run Experiment 5: |
195 |
|
cd ../../../verification/lab_sea/input |
196 |
|
ln -sf data.2years data |
197 |
|
ln -sf data.seaice.adi data.seaice |
198 |
|
ln -sf evap.labsea1979 evap.labsea1980 |
199 |
|
ln -sf flo.labsea1979 flo.labsea1980 |
200 |
|
ln -sf fsh.labsea1979 fsh.labsea1980 |
201 |
|
ln -sf prate.labsea1979 prate.labsea1980 |
202 |
|
ln -sf qa.labsea1979 qa.labsea1980 |
203 |
|
ln -sf tair.labsea1979 tair.labsea1980 |
204 |
|
ln -sf u10m.labsea79 u10m.labsea80 |
205 |
|
ln -sf v10m.labsea79 v10m.labsea80 |
206 |
|
mpirun -np 2 mitgcmuv_2x1 |
207 |
|
|
208 |
|
Use the matlab script lookat_exp5.m to compare |
209 |
|
the 2-cpu output to SMMR-SSM/I data: |
210 |
|
cd ../../../verification/lab_sea/matlab |
211 |
|
matlab |
212 |
|
lookat_exp5 |
213 |
|
|
214 |
|
Disclaimer: |
215 |
|
The comparison here is just for fun, not really supposed |
216 |
|
to look anything like the data. |
217 |
|
Otherwise it would put a lot of people out of business :-) |
218 |
|
|
219 |
|
Instructions for running Experiment 6 |
220 |
|
===================================== |
221 |
|
|
222 |
|
This is a 1-cpu test of sea-ice thermodynamics (no dynamics). |
223 |
|
The solution is compared to that of experiment 1. |
224 |
|
|
225 |
|
To run Experiment 6: |
226 |
|
cd ../../../verification/lab_sea/input |
227 |
|
ln -sf data.10hours data |
228 |
|
ln -sf data.seaice.nodynamics data.seaice |
229 |
|
mitgcmuv_1 >&! output.txt |
230 |
|
mkdir exp6 |
231 |
|
mv *tave.0000000010.data exp6 |
232 |
|
|
233 |
|
Use the matlab script lookat_exp6.m to compare |
234 |
|
the output of exp6 to that of exp1: |
235 |
|
cd ../verification/lab_sea/matlab |
236 |
|
matlab |
237 |
|
lookat_exp6 |
238 |
|
|
239 |
|
Instructions for running Experiment 7 |
240 |
|
===================================== |
241 |
|
|
242 |
|
This is a 2-cpu, 10-hour integration used to test |
243 |
|
tile edges for sea-ice thermodynamics (no dynamics). |
244 |
|
2-CPU executable from experiment 4 is required. |
245 |
|
|
246 |
|
To run Experiment 7: |
247 |
|
cd ../../../verification/lab_sea/input |
248 |
|
ln -sf data.10hours data |
249 |
|
ln -sf data.seaice.nodynamics data.seaice |
250 |
|
mpirun -np 2 mitgcmuv_2x1 |
251 |
|
mkdir exp7 |
252 |
|
mv *tave.0000000010.data exp7 |
253 |
|
|
254 |
|
Use the matlab script lookat_exp7.m to compare |
255 |
|
the output of exp7 to that of exp6: |
256 |
|
cd ../verification/lab_sea/matlab |
257 |
|
matlab |
258 |
|
lookat_exp7 |