|
| str | dir1 = "./sampledata/scale-5.4.5/scale-rm/test/tutorial/real/experiment/run/" |
| |
| int | PRC_NUM_X = 2 |
| |
| int | PRC_NUM_Y = 2 |
| |
| str | ftype = "history" |
| |
| str | domainlabel = "_d01" |
| |
| str | timelabel = "" |
| |
| str | varname1 = "T2" |
| | Value (e.g., PREC) 2D (spatial) + time.
|
| |
| str | varunit1 = "(K)" |
| |
| int | varfact1 = 1 |
| |
| | varlevels1 = np.linspace(265,305,41) |
| |
| str | varcmap1 = "jet" |
| |
| int | blct = 1 |
| |
| int | blcy = 15 |
| |
| int | blcx = 15 |
| |
| | args = sys.argv |
| |
| | PRC_NUM_X_ANL = int(args[1]) |
| |
| | PRC_NUM_Y_ANL = int(args[2]) |
| |
| str | dir_out = "./fig/" |
| |
| str | horaxis = "LATLON" |
| |
| bool | withmap = True |
| |
| | lonticks = np.arange(80,180+1,10) |
| |
| | latticks = np.arange(10,60+1,10) |
| |
| int | slon = 120 |
| |
| int | elon = 150 |
| |
| int | slat = 25 |
| |
| int | elat = 45 |
| |
| bool | savefig = True |
| |
| | comm |
| |
| | size |
| |
| | rank |
| |
| | PRC_NUM_X_PER_ANL |
| |
| | PRC_NUM_Y_PER_ANL |
| |
| | fpathlist = get_fpathlist_mpi(dir1,ftype,domainlabel,timelabel,PRC_NUM_X,PRC_NUM_Y,PRC_NUM_X_ANL,PRC_NUM_Y_ANL,size,rank) |
| |
| | PRC_NUM_ARR0 = check_prcnum_mpi(PRC_NUM_X,PRC_NUM_Y,PRC_NUM_X_ANL,PRC_NUM_Y_ANL,size,rank) |
| |
| | xrvar = get_xrvar(fpathlist) |
| |
| | var2d_avg = get_blcavg(xrvar,varname1,blct,blcy,blcx) |
| |
| | sendbuf = np.array(var2d_avg,"d") |
| |
| | recvbuf = None |
| |
| | root |
| |
| | var2d_avg_combined = combine_var2d_blockavg(sendbuf,recvbuf,PRC_NUM_Y_ANL,PRC_NUM_X_ANL) |
| |
| | time = xrvar.coords["time"] |
| |
| | tsize = len(time) |
| |
| str | tlabel = f"{t*blct:04d}-{(t+1)*blct-1:04d}" |
| |
| | lon = xrvar.coords["lon"] |
| |
| | lat = xrvar.coords["lat"] |
| |
| | fig = plt.figure(figsize=(8,5)) |
| |
| | ax = fig.add_subplot(111,projection=ccrs.PlateCarree()) |
| |
| | color |
| |
| | crs |
| |
| | xlocs |
| |
| | ylocs |
| |
| | figsize |
| |
| | cax = ax.contourf(lon[blcy//2::blcy,blcx//2::blcx],lat[blcy//2::blcy,blcx//2::blcx],varfact1*var2d_avg_combined[t,:,:],levels=varlevels1,cmap=varcmap1) |
| |
| str | figtitle = f"mpi_xarray_coarsened_ll_{varname1}_blct{blct:04d}blcy{blcy:04d}blcx{blcx:04d}_t{tlabel}.png" |
| |
| | x = xrvar.coords["x"] |
| |
| | y = xrvar.coords["y"] |
| |
| | cbar = plt.colorbar(cax,ax=ax,fraction=0.02) |
| |
MPI 並列を用いた水平粗視化・描画
MPI並列(MPI Gather)を用いた二次元変数の粗視化分布
- Author
- Tomoro Yanase, Team SCALE
- Note
- 実行例: $ mpirun -n 2 python mpi_xarray_coarsened_xy.py 1 2
