def netcdf_netcdf(ncfile1,ncfile2):
nc1_ids, nc1_pos = MOM.read_netcdf_drifters(ncfile1)
nc2_ids, nc2_pos = MOM.read_netcdf_drifters(ncfile2)
dist = np.zeros(nc1_ids.shape)
for i in range(0,nc1_ids.shape[0]):
for j in range(0,nc2_ids.shape[0]):
if nc1_ids[i] == nc2_ids[j]:
dist[i] = great_circle(nc1_pos[i,0:2],nc2_pos[i,0:2]).km
return nc1_ids,dist
def txt_netcdf(txtfile,ncfile):
txt_ids,txt_pos = MOM.read_txt_drifters(txtfile)
nc_ids,nc_pos = MOM.read_netcdf_drifters(ncfile)
dist = np.zeros(txt_ids.shape)
for i in range(0,txt_ids.shape[0]):
for j in range(0,nc_ids.shape[0]):
if txt_ids[i] == nc_ids[j]:
dist[i] = great_circle(txt_pos[i,0:2],nc_pos[j,0:2]).km
return txt_ids, dist
lons,lats,levs = MOM.read_netcdf_grids(gfile)
nlon = np.size(lons)
nlat = np.size(lats)
nlev = np.size(levs)
dr_id = 32
dpos_ens_g = np.zeros([2,num_ens])
dpos_ens_g2 = np.zeros([2,num_ens])
# read drifter ensemble data based on the given drifter id
for k in range(1,num_ens+1):
ens_dir = os.path.dirname(root_dir + '/OUTPUT/' + date_str + '/model/' + '%03.d'%(k) + '/RESTART/')
edrfile_g = os.path.join(ens_dir,'drifters_inp.nc')
tmp1,tmp2 = MOM.read_netcdf_drifters(edrfile_g)
dpos_ens_g[:,k-1] = tmp2[dr_id-1,0:2]
#ens_dir2 = os.path.dirname(root_dir + '/OUTPUT/' + date_str + '/model/' + '%03.d'%(k) + '/INPUT/')
#edrfile_g = os.path.join(ens_dir2,'drifters_inp.nc')
#tmp1,tmp2 = MOM.read_netcdf_drifters(edrfile_g)
#dpos_ens_g2[:,k-1] = tmp2[dr_id-1,0:2]
#dpos_ens_g[:,k-1] = dpos_ens_g[:,k-1] - dpos_ens_g2[:,k-1]
# read true drifter position as mean
tdrfile = os.path.join(true_dir,'drifters_inp.nc')
tmp1,tmp2 = MOM.read_netcdf_drifters(tdrfile)
# code the andr_sp and gsdr_sp later
for j in range(1,num_ens+1):
an_ens = os.path.join(letkf_dir,'anal' + '{:03d}'.format(j) + '.drifters_inp.nc')
gs_ens = os.path.join(letkf_dir,'gs'+ '{:02d}'.format(iday) + '{:03d}'.format(j) + '.drifters_inp.txt')
tmp_id,tmp_an_dist = txt_netcdf(anal,an_ens)
tmp_id,tmp_gs_dist = txt_gstxt(gues,gs_ens)
tmp_an_sp[:,0] = tmp_an_sp[:,0] + np.square(tmp_an_dist[:])
tmp_gs_sp[:,0] = tmp_gs_sp[:,0] + np.square(tmp_gs_dist[:])
spa_dist[:,i-1] = np.sqrt(tmp_an_sp[:,0]/(num_ens-1))
spg_dist[:,i-1] = np.sqrt(tmp_gs_sp[:,0]/(num_ens-1))
# code for plotting drifter trajectories: true, gues_me and anal_me
tid, tpos[:,:,i-1] = MOM.read_netcdf_drifters(true)
gid, gpos[:,:,i-1] = MOM.read_txt_drifters(gues)
aid, apos[:,:,i-1] = MOM.read_txt_drifters(anal)
cid, cpos[:,:,i-1] = MOM.read_netcdf_drifters(ctrl)
f.write(str(np.mean(tmg_dist[:,i-1])) + " " + str(np.mean(tma_dist[:,i-1])) + " " + str(np.mean(spg_dist[:,i-1])) + " " + str(np.mean(spa_dist[:,i-1])) + " " + str(np.mean(tmc_dist[:,i-1])) + "\n")
f.close()
# plotting the average drifters distance error in the testing time interval
plt.figure(1)
plt.plot(time,np.mean(tmg_dist[:,:],axis=0),label='true - forecast',c='b')
plt.plot(time,np.mean(tma_dist[:,:],axis=0),label='true - analysis',c='r')
plt.plot(time,np.mean(tmc_dist[:,:],axis=0),label='true - control', c='g')
