search

DOI-USGS / COAWST

COAWST modeling system git repository
Other
104 stars 49 forks source link

WRF+SWAN WIND input grids #261

Open DiegoBindoni opened 4 months ago

DiegoBindoni commented 4 months ago

Good afternoon, i'm running COAWST simulations with WRF+SWAN coupling, on different grids. I'm trying to compare the results with WRF standalone simulations to assess the importance of wind wave interactions. I'm having issues with the WIND grid of SWAN. If i don't select it, the simulation runs smoothly, but when i postprocess the results, there is zero difference with WRF standalone simulations. When i add only INPGRID, without READINP, the simulation crashes. I attach here my swan input files, hoping to figure out the problem. Thanks in advance for the help.

swan_Belgium.txt swan_Belgium_ref3.txt

jcwarner-usgs commented 4 months ago

when it 'crashes' what actually happens?

DiegoBindoni commented 4 months ago

WRF starts normally. They should exchange data every 27 seconds. WRF works with 27,9,3 s. SWAN with 90 and 30. At 4:30, COAWST crashes. Errfile are empty, but PRINT (attached) doesn't show the continuation of the computation of SWAN. I address hereby this problem to SWAN. Let me know if i can provide additional files. Thanks a lot for the help

coupling_belgium.txt swan_Belgium.txt swan_Belgium_ref3.txt PRINT01-001.txt PRINT02-001.txt

jcwarner-usgs commented 4 months ago

ok first the models need to time step evenly into the coupling interval. the coupling interval is set at 27s, but swan is on a 90 sec time step. so swan can not provide that data.

how about WRF at like 30 10 and 5, and SWaN at 90 and 30, with a coupling at 90 sec?

DiegoBindoni commented 4 months ago

I did 45,15,5 in order to keep the same time step ratio of WRF, SWAN at 90,30 and coupling interval at 90. Now it stops after 90 seconds. With this PRINT (Always no Errf)

WRF grid 1 sent data to WAVE grid 1

Timing for main: time 2022-03-29_00:01:30 on domain 3: 0.17093 elapsed seconds Timing for main: time 2022-03-29_00:01:30 on domain 2: 1.11155 elapsed seconds == SWAN grid 1 recv data from WRF grid 1 WRFtoSWAN Min/Max U10 (ms-1): -8.002854E+00 8.638475E+00 WRFtoSWAN Min/Max V10 (ms-1): -9.743976E+00 1.073547E+00

WRF grid 2 sent data to WAVE grid 1

WRF grid 3 sent data to WAVE grid 1

WRF grid 1 sent data to WAVE grid 2

== SWAN grid 1 recv data from WRF grid 2 WRFtoSWAN Min/Max U10 (ms-1): -9.916405E+00 8.954644E+00 WRFtoSWAN Min/Max V10 (ms-1): -1.222754E+01 9.448338E+00 == SWAN grid 1 recv data from WRF grid 3 WRFtoSWAN Min/Max U10 (ms-1): -6.579018E+00 0.000000E+00 WRFtoSWAN Min/Max V10 (ms-1): -3.851140E+00 1.964051E+00 == SWAN grid 2 recv data from WRF grid 1 WRFtoSWAN Min/Max U10 (ms-1): 0.000000E+00 0.000000E+00 WRFtoSWAN Min/Max V10 (ms-1): 0.000000E+00 0.000000E+00

WRF grid 2 sent data to WAVE grid 2

WRF grid 3 sent data to WAVE grid 2

== SWAN grid 2 recv data from WRF grid 2 WRFtoSWAN Min/Max U10 (ms-1): -9.926502E+00 2.860514E+00 WRFtoSWAN Min/Max V10 (ms-1): -4.630641E+00 9.434210E+00 == SWAN grid 2 recv data from WRF grid 3 WRFtoSWAN Min/Max U10 (ms-1): -6.583903E+00 0.000000E+00 WRFtoSWAN Min/Max V10 (ms-1): -3.938122E+00 2.019151E+00 [ealin35:6015 :0:6015] Caught signal 11 (Segmentation fault: address not mapped to object at address 0x150c7c1c)

PRINT01-001.txt PRINT02-001.txt

jcwarner-usgs commented 4 months ago

can you send the full output? seems odd to have stepped for 90s and then stop.

DiegoBindoni commented 4 months ago

log.coawst.txt Here it is

jcwarner-usgs commented 4 months ago

looks like SWAN did not really statrt . I see INITIAL HOTSTART MULTIPLE '/data/nobackup/bindoni/COAWST/Projects/SWANSP/HOTFILE_W2/Belgium_d01_swanrst.hot' INIT that is 2 init calls. can you comment out the second one, for both the swan_belgium input files and try again?

DiegoBindoni commented 4 months ago

It still crashed. I attach the outputs. I'm running SWAN in parallel, it shouldn't be a problem as long as the hotfiles are the same number of processors no?

log.coawst1.txt PRINT01-001.txt PRINT02-001.txt

jcwarner-usgs commented 4 months ago

not sure. try this: 1- try both swan grids with just INIT swan does not seem to be starting. 2- try swan time steps to be 30 and 15.

DiegoBindoni commented 4 months ago

Still the same problem, with the same output files. Unfortunately i cannot pinpoint where exactly it stops and why

DiegoBindoni commented 4 months ago

belgium.h.txt scrip_coawst_belgium.in.txt For clarity i also attach the .h file and the scrip file

jcwarner-usgs commented 4 months ago

have you run swan nested by itself? can you try with just 1 wrf grid or maybe 2 wrf grids?

DiegoBindoni commented 4 months ago

Yes i ran both swan and wrf standalone models and they work. I could try, but i would like to understand why is the wind grid that creates problem. Without it works perfectly fine

jcwarner-usgs commented 4 months ago

i would like to understand also. But i need your help. The Sandy case works with just wrf+swan and that is my reference. I am not sure why your setup does not work. Can you try the Sandy case with just wrf+swan? that only has 2 grids, but lets see if that works for you. If it does, then lets try your case with just 2 wrf and 2 swan. If that works, then i would probably need to get your inputs and try it all here.

DiegoBindoni commented 4 months ago

log.sandy.txt PRINT01.txt PRINT02.txt

Ok sandy test case now crashes. Could be directly a problem with the installation, but that's odd. I was pretty sure back then to have the tutorial running. Thank you again for your time

jcwarner-usgs commented 4 months ago

i dont see the error. when you run it, have all the output go to an out file

mpirun -np 2 ./coawstM Projects/Sandy/coupling_sandy.in &> outfile.txt

jcwarner-usgs commented 4 months ago

ok. So i did a git pull, made a tar file, copied it to a new directory, untarred, set the build_coawst.sh to build the sandy case, edited the Projects/Sandy/sandy.h to be

undef ROMS_MODEL

define NESTING

define WRF_MODEL

define SWAN_MODEL

undef WW3_MODEL

define MCT_LIB

undef MCT_INTERP_OC2AT

define MCT_INTERP_WV2AT

undef MCT_INTERP_OC2WV

/.... rest of sandy.h is unchanged

./build_coawst.sh -j 14 selected 15 for ifort, select the default build for no movig nest, it built and created a coawstM cp Projects Sandy/namelist . cp Projects Sandy/wrf* . mpirun -np 2 ./coawstM Projects/Sandy/coupling_sandy.in and it is running.

....

WRF grid 1 recv data from WAVE grid 2

WAVtoWRF Min/Max HSIGN (m): 0.000000E+00 8.345883E+00 WAVtoWRF Min/Max WLENP (m): 0.000000E+00 2.552078E+02 WAVtoWRF Min/Max RTP (m): 0.000000E+00 1.462009E+01 == SWAN grid 2 sent data to WRF grid 2

WRF grid 2 recv data from WAVE grid 2

WAVtoWRF Min/Max HSIGN (m): 0.000000E+00 8.363349E+00 WAVtoWRF Min/Max WLENP (m): 0.000000E+00 2.552078E+02 WAVtoWRF Min/Max RTP (m): 0.000000E+00 1.462009E+01 Timing for main: time 2012-10-28_13:30:00 on domain 1: 22.06583 elapsed seconds Timing for main: time 2012-10-28_13:30:00 on domain 1: 22.06583 elapsed seconds +time 20121028.133300 , step 31; iteration 1; sweep 1 grid 1 Timing for Writing wrfout_d01_2012-10-28_13:30:00 for domain 1: 0.40529 elapsed seconds Timing for Writing wrfout_d01_2012-10-28_13:30:00 for domain 1: 0.40529 elapsed seconds +time 20121028.133300 , step 31; iteration 1; sweep 2 grid 1 +time 20121028.133300 , step 31; iteration 1; sweep 3 grid 1 +time 20121028.133300 , step 31; iteration 1; sweep 4 grid 1 +time 20121028.133130 , step 61; iteration 1; sweep 1 grid 2 +time 20121028.133130 , step 61; iteration 1; sweep 2 grid 2 Timing for Writing wrfout_d02_2012-10-28_13:30:00 for domain 2: 0.52084 elapsed seconds Timing for Writing wrfout_d02_2012-10-28_13:30:00 for domain 2: 0.52084 elapsed seconds +time 20121028.133130 , step 61; iteration 1; sweep 3 grid 2 +time 20121028.133130 , step 61; iteration 1; sweep 4 grid 2 Timing for main: time 2012-10-28_13:31:00 on domain 2: 2.32662 elapsed seconds Timing for main: time 2012-10-28_13:31:00 on domain 2: 2.32662 elapsed ......

Can you try those exact steps? is that what you did? I am not sure what to try next.