Difference between 'TIDE_START' and 'zero_phase_date'

Message

Dan_chan · #1 Unread post by **Dan_chan** » Mon Mar 04, 2024 8:51 am

Hi,

I use OTPS to extract my TIDEFILE and have some confusion about "tide_start". As OTPS toolbox couldn't set zero_phase_date variable directly, I set TIDE_START in roms.in. As shown in #896 https://www.myroms.org/projects/src/ticket/896, it returns to same result compared with zero_phase_date.

1. simulation time period: 1993.01.01-2020.12.31
2. Reftime in ININAME:

Code: Select all

   
   Time coordinate :
                       ocean_time : 2 steps
     RefTime =  1990-01-01 00:00:00  Units = seconds  Calendar = proleptic_gregorian
  YYYY-MM-DD hh:mm:ss  YYYY-MM-DD hh:mm:ss  YYYY-MM-DD hh:mm:ss  YYYY-MM-DD hh:mm:ss

3. time settings in otps2frc_wrapper_tpxo9_atlas30.m:

Code: Select all

base_date=datenum(1990,1,1);
pred_date=datenum(1990,1,1);
otps2frc_v5_tpxo9_atlas30(gfile,base_date,pred_date,ofile,model_dir,'bysea')

4. time settings in roms.in:

Code: Select all

      DSTART =  1096.0d0                      ! days
  TIDE_START =  0.0d0                      ! days
    TIME_REF =  19900101.0d0                      ! yyyymmdd.dd

Based on the above settings, it could successfully run simulations for the period 1993-2020 without restart. However, my ongoing work requires running simulations for a more extended period, prompting me to implement restart. Initially, I simulated ocean states for the years 1993-2012 and attempted to continue the simulation using the generated RSTFILE. Unfortunately, I encountered an issue "The MPI error".

Code: Select all

  ROMS I/O variables Metadata File:  ../../input/varinfo.yaml
[b3301r5n4:128192] *** An error occurred in MPI_Bcast
[b3301r5n4:128192] *** reported by process [47033195167745,4294967416]
[b3301r5n4:128192] *** on communicator MPI_COMM_WORLD
[b3301r5n4:128192] *** MPI_ERR_TRUNCATE: message truncated
[b3301r5n4:128192] *** MPI_ERRORS_ARE_FATAL (processes in this communicator will now abort,
[b3301r5n4:128192] ***    and potentially your MPI job)

  TIDE_DATE - Checking tidal reference date for zero phase:
                 'zero_phase_date' variable not found in: ../../input/bysea5km_tide_tpxo9_atlas30_19900101_8p.nc
                 given tide_start =        0.0000 (days)
                 tide_DateNumber  =   726834.0000 (1990-01-01 00:00:00)

 Tile partition information for Grid 01:  216x220x31  tiling: 16x12

     tile     Istr     Iend     Jstr     Jend     Npts

        0        1       10        1       15     4650
        1       11       24        1       15     6510
        2       25       38        1       15     6510
        3       39       52        1       15     6510
        ......
        66       25       38       7forrtl: error (78): process killed (SIGTERM)
Image              PC                Routine            Line        Source
romsM              0000000000966F2A  Unknown               Unknown  Unknown
......
       67       39       52       73       91     8246
       68       53       66       73       91     8246
       ......

With the assistance of the ROMS community and fellow users, I addressed the issue by adding the zero_phase_date variable to the TIDEFILE.

Code: Select all

add_tide_date('bysea5km_tide_tpxo9_atlas30_19900101_8p.nc', datenum(1990,1,1))

I am still curious about what makes this bug. Will the simulation results obtained by these two methods be very different?

Many thanks in advance for any advice.