Ticket #1628: correspondence_mohit.txt

File correspondence_mohit.txt, 6.2 KB (added by ros, 4 years ago)

Email thread

Line 
1Hi Garry,
2
3
4
5I am not familiar with other parts of the hg66.x builds, but NETCDF links w=
6ill certainly needed to be added explicitly. From what I see for a vn8.4 bu=
7ild:
8
9
10
11- umui_jobs/<job-id>/FCM_BLD_COMMAND file has the line =93module load cray-=
12netcdf=94.
13
14- The compile commands have =93-INetCDFmodule=94 flag, while the link comma=
15nd has =93-LNetCDFmodule =96lnetcdf=94 flags.
16
17
18
19The first one can be added via a hand_edit, but not sure how additional fla=
20gs are added (- probably machine overrides, but what syntax?), so it is wor=
21th emailing NCAS-CMS.
22
23
24
25---
26
27Mohit
28
29
30
31From: Hayman, Garry D. [mailto:garr@ceh.ac.uk]
32Sent: 18 August 2015 14:55
33To: Dalvi, Mohit
34Subject: RE: HadGEM2 and nudging
35
36
37
38Mohit
39
40
41
42OK, I will give it a try.  It is rather unfortunate as the run was starting=
43 to cover the period of the GOSAT measurements.
44
45
46
47I have some another questions regarding the runs that I am setting up on th=
48e Cray.  I have successfully run the example atmosphere-ocean and atmospher=
49e-only jobs (respectively, xlrla=3Dxlqja and xlrlb=3Dxlpoi), these had the =
50UKCA switched off.  I have upgraded my Std Trop job (xldkn) to HG6.6.6.  Al=
51l the nudging-related options were switched off.  The run has completed one=
52 month after I recreated the binary files containing the offline photolysis=
53 rates.
54
55
56
57I am now switching on the nudging =96 handedits, branches, overrides.  The =
58code has failed to compile (see my job xlrld).  It failed with no rule to m=
59ake =91banner_output=92 (see /home/ghayma/output/xlrld000.xlrld.d15230.t121=
60306.comp.leave on Cray).  Where does this come from?  Would it be better to=
61 duplicate my branches using HG6.6.6?  Also, have the netCDF libraries been=
62 sent up?
63
64
65
66Again, any assistance welcome.
67
68
69
70Regards
71
72
73
74Garry
75
76
77
78From: Dalvi, Mohit [mailto:mohit.dalvi@metoffice.gov.uk]
79Sent: 18 August 2015 09:58
80To: Hayman, Garry D. <garr@ceh.ac.uk<mailto:garr@ceh.ac.uk>>
81Subject: RE: HadGEM2 and nudging
82
83
84
85Hi Garry,
86
87
88
89I am afraid I do not have any idea about the cause of the error. We conside=
90r this as =91model blowing up=92, implying there is/ has been a numerical p=
91roblem somewhere. The latest dump also does not seem to show one of the usu=
92al culprits i.e. non-uniform polar rows.
93
94
95
96At this stage I can only suggest trying to perturb the run in some way e.g.=
97 changing the convection sub-steps for a few days and then reverting back a=
98fter model passes the failure point.
99
100
101
102---
103
104Mohit
105
106
107
108From: Hayman, Garry D. [mailto:garr@ceh.ac.uk]
109Sent: 18 August 2015 09:32
110To: Dalvi, Mohit
111Subject: RE: HadGEM2 and nudging
112
113
114
115Mohit
116
117
118
119Many thanks.  I see that the transfer of the ERA-40 data is now complete.  =
120This will allow to setup and test a nudged run on the Cray.
121
122
123
124On another matter, I had a run on the IBM (xldko: Std Trop chemical scheme =
125with nudging) for comparison with satellite observations.  The run crashed =
126with an error in an iinterpolation routine (at 16th June 2010).  I restarte=
127d from December 2009 but it crashed again at the same point with the same e=
128rror.  The leave files are:
129
130
131
132/home/ghayma/output/xldko038.xldko.d15223.t212412.leave
133
134/home/ghayma/output/xldko006.xldko.d15229.t201455.leave
135
136
137
138gc_abort (Processor  63 ): over-writing due to dim_e_out size
139
140
141
142  Traceback:
143
144    Offset 0x00000010 in procedure xl__trbk_
145
146    Offset 0x000000f8 in procedure gc_abort_, near line 180 in file /projec=
147ts/um1/gcom/gcom3.5/meto_ibm_pwr6_mpp/ppsrc/gcom/gc/gc_abort.f
148
149    Offset 0x00000458 in procedure ereport_, near line 382 in file /project=
150s/jules/ghayma/xldko/ummodel/ppsrc/UM/control/misc/ereport.f90
151
152    Offset 0x000093f8 in procedure interpolation_, near line 1155 in file /=
153projects/jules/ghayma/xldko/ummodel/ppsrc/UM/atmosphere/dynamics_advection/=
154interpolat
155
156ion.f90
157
158    Offset 0x00008384 in procedure sl_thermo_, near line 980 in file /proje=
159cts/jules/ghayma/xldko/ummodel/ppsrc/UM/atmosphere/dynamics_advection/sl_th=
160ermo.f90
161
162    Offset 0x00000fd0 in procedure ni_sl_thermo_, near line 703 in file /pr=
163ojects/jules/ghayma/xldko/ummodel/ppsrc/UM/atmosphere/dynamics_advection/ni=
164_sl_thermo
165
166.f90
167
168    Offset 0x00017790 in procedure atm_step_, near line 10310 in file /proj=
169ects/jules/ghayma/xldko/ummodel/ppsrc/UM/control/top_level/atm_step.f90
170
171    Offset 0x0001abc8 in procedure u_model_, near line 5270 in file /projec=
172ts/jules/ghayma/xldko/ummodel/ppsrc/UM/control/top_level/u_model.f90
173
174    Offset 0x00001ec8 in procedure um_shell, near line 4237 in file /projec=
175ts/jules/ghayma/xldko/ummodel/ppsrc/UM/control/top_level/um_shell.f90
176
177    --- End of call chain ---
178
179
180
181Are you able to provide any insight as to the cause of the problem?
182
183
184
185Many thanks
186
187
188
189Garry
190
191
192
193>-----Original Message-----
194
195>From: Dalvi, Mohit [mailto:mohit.dalvi@metoffice.gov.uk]
196
197>Sent: 17 August 2015 17:22
198
199>To: Hayman, Garry D. <garr@ceh.ac.uk<mailto:garr@ceh.ac.uk>>; Luke Abraham
200
201><luke.abraham@atm.ch.cam.ac.uk<mailto:luke.abraham@atm.ch.cam.ac.uk>>
202
203>Subject: RE: HadGEM2 and nudging
204
205>
206
207>Hi Garry,
208
209>
210
211>The plan is for the files to reside under /projects/ukca-admin/analyses/ o=
212n
213
214>the Cray. I have already transferred the ERA-Interim data, but did not tra=
215nsfer
216
217>ERA-40 in view of planned system work. I will start the transfer now and l=
218et
219
220>you know once this is complete.
221
222>
223
224>Cheers
225
226>---
227
228>Mohit
229
230>
231
232>
233
234>-----Original Message-----
235
236>From: Hayman, Garry D. [mailto:garr@ceh.ac.uk]
237
238>Sent: 17 August 2015 17:13
239
240>To: Dalvi, Mohit; Luke Abraham
241
242>Subject: HadGEM2 and nudging
243
244>
245
246>Mohit, Luke
247
248>
249
250>I am in the process of getting my HadGEM2 jobs to run on the new CRAY.  In
251
252>most of my jobs, the dynamics and temperatures of the climate model were
253
254>"nudged" towards ECMWF ERA-40 reanalyses of the atmospheric state of
255
256>temperature, surface pressure and the horizontal wind components (as
257
258>described in Telford et al., 2008).
259
260>
261
262>The ECMWF ERA-40 reanalysis files are currently located at
263
264>/nerc/ukca/analyses/era-in and /nerc/ukca/analyses/era-40.  There will be =
265no
266
267>/nerc directory on the CRAY.   The directories are too large for me to cop=
268y to
269
270>the JULES project area with the current JULES allocation (era-40, 375 Gbyt=
271es;
272
273>era-in, 253 GBytes).  Have these files been copied to the CRAY?  If so, wh=
274ere?
275
276>If not, when is this likely to happen?  I would need access to these to te=
277st the
278
279>runs with the nudging on.
280
281>
282
283>Thanks and regards
284
285>
286
287>Garry
288
289>
290
291>Dr Garry Hayman
292
293>