System has switched to CentOS.

There's no display attached to it, so I modified the ploting utility to fit it.
The cuda library is located at different location as well.
And I change the simulation time/output_time to 7200/1800, which comsumes less time thus better for testing.

Makefile

@@ -4,14 +4,14 @@
 #####################################################################
 
 ### Plese specify your fortran, cuda compiler and GPU's compute capabilities #####
-FC          = x86_64-pc-linux-gnu-gfortran-4.9
+FC          = gfortran
 FC_FLAGS    = -O2
 
 
 NVCC        = nvcc
 NVFLAGS     =
 NVCC_OPT    = -arch=sm_61 -lineinfo
-NVLIB       = -L/opt/cuda/lib64
+NVLIB       = -L/usr/local/cuda-9.2/lib64
 ##################################################################################
 
 

comcot.ctl

@@ -8,8 +8,8 @@
 # General Parameters for Simulation             : Value Field                  |
 #===============================================:===============================
 #Job Description: NZ30sec bathymetry, Spherical Coordinates for code testing
- Total run time (Wall clock, seconds)           :   86400.0
- Time interval to Save Data    ( unit: sec )    :   3600.0
+ Total run time (Wall clock, seconds)           :   7200.0
+ Time interval to Save Data    ( unit: sec )    :   1800.0
  Output Zmax & TS (0-Max Z;1-Timeseries;2-Both) :     2
  Start Type (0-Cold start; 1-Hot start)         :     0
  Resuming Time If hot start (Seconds)           :  100000000000000.00

plot_dat.py

@@ -159,4 +159,5 @@ def Global_MAP(self):
         return m
 
 if __name__ == '__main__':
+    plt.switch_backend('agg') # no display mode 
     test = COMCOTGlobalmap(argv[1])