def SWOT_observation_error(): """observation error of WSE depending on the L*W of each pixel used sigma*(1/l)*(1/w) l=k*L, w=q*W Rodrigaz et al 2017: According to CaMa k=0.25, q=0.85""" fname=pm.DA_dir()+"/out/"+pm.experiment()+"/assim_out/obs/obs_err.bin" if os.path.isfile(fname): return 0 else: nx,ny,gsize = pm.map_dimension() k=1.00 # assume nearest part to the unit catchment q=1.00 # used 1.0 -> river width variability is 30% ovs_err = 0.10 rivlen=np.fromfile(pm.CaMa_dir()+"/map/glb_15min/rivlen.bin",np.float32).reshape(ny,nx) rivwth=np.fromfile(pm.CaMa_dir()+"/map/glb_15min/rivwth_gwdlr.bin",np.float32).reshape(ny,nx) nextx=(np.fromfile(pm.CaMa_dir()+"/map/glb_15min/nextxy.bin",np.int32).reshape(2,ny,nx)[0]!=-9999)*1.0 rivlen=1.0 #rivlen*1.0e-3 #used as one kilometer rivwth=rivwth*1.0e-3 area=(k*rivlen)*(q*rivwth) obs_err=ovs_err*(1/(k*rivlen+1.0e-20))*(1/(q*rivwth+1.0e-20))*nextx #obs_err=pm.ovs_err()*(1/(area+1.0e-20))*nextx # if water area < 1.0 km2 -> 0.25 obs_err=obs_err*(area>=1.0)*1.0+0.25*(1/(k*rivlen+1.0e-20))*(1/(q*rivwth+1.0e-20))*nextx*(area<1.0)*1.0 obs_err=ma.masked_where(area<0.625,obs_err).filled(0.25) # 25cm for < 1km^2 water area obs_err=obs_err*nextx obs_err=obs_err.astype(np.float32) # obs_err0=obs_err[iy,ix] fname=pm.DA_dir()+"/out/"+pm.experiment()+"/assim_out/obs/obs_err.bin" obs_err.tofile(fname) return 0
def observation_error():
"""observation error of WSE depending on the L*W of each pixel
used sigma*(1/l)*(1/w) l=k*L, w=q*W Rodrigaz et al 2017:
According to CaMa k=0.25, q=0.85"""
k = 1.00 # assume nearest part to the unit catchment
q = 1.00 # used 1.0 -> river width variability is 30%
rivlen = np.fromfile(pm.CaMa_dir() + "/map/glb_15min/rivlen.bin",
np.float32).reshape(720, 1440)
rivwth = np.fromfile(pm.CaMa_dir() + "/map/glb_15min/rivwth_gwdlr.bin",
np.float32).reshape(720, 1440)
nextx = (np.fromfile(pm.CaMa_dir() + "/map/glb_15min/nextxy.bin",
np.int32).reshape(2, 720, 1440)[0] != -9999) * 1.0
rivlen = 1.0 #rivlen*1.0e-3 #used as one kilmeter
rivwth = rivwth * 1.0e-3
area = (k * rivlen) * (q * rivwth)
obs_err = pm.ovs_err() * (1 / (k * rivlen + 1.0e-20)) * (
1 / (q * rivwth + 1.0e-20)) * nextx
#obs_err=pm.ovs_err()*(1/(area+1.0e-20))*nextx
# if water area < 1.0 km2 -> 0.25
obs_err = obs_err * (area >= 1.0) * 1.0 + 0.25 * (
1 / (k * rivlen + 1.0e-20)) * (1 / (q * rivwth + 1.0e-20)) * nextx * (
area < 1.0) * 1.0
obs_err = ma.masked_where(area < 0.625, obs_err).filled(
0.25) # 25cm for < 1km^2 water area
obs_err = obs_err * nextx
obs_err = obs_err.astype(np.float32)
obs_err.tofile("./obs_err.bin")
return 0
def get_grdc_station_v396(name):
#--ask the river name and a or b
# all locations
#grdc = pm.CaMa_dir() + "/map/glb_15min/grdc_loc.txt"
grdc = pm.CaMa_dir() + "/map/"+pm.mapname()+"/grdc_loc.txt"
f = open(grdc,"r")
lines = f.readlines()
f.close()
station_loc = []
x_list = []
y_list = []
#---
for line in lines[1::]:
#print line
line = filter(None, re.split(";",line))
#print line
grdc_id = line[0]
river = line[1].strip()
station = line[2].strip()
ix1 = int(line[3])-1
iy1 = int(line[4])-1
ix2 = int(line[5])
#print river
if ix2 != -9999:
ix2 = ix2-1
iy2 = int(line[6])
if iy2 != -9999:
iy2 = iy2-1
if station == name:
return ix1,iy1,ix2,iy2
def get_grdc_loc_v396(name):
#--ask the river name and a or b
# all locations
#grdc = pm.CaMa_dir() + "/map/glb_15min/grdc_loc.txt"
grdc = pm.CaMa_dir() + "/map/"+pm.mapname()+"/grdc_loc.txt"
f = open(grdc,"r")
lines = f.readlines()
f.close()
#--
id_list = []
station_loc = []
x_list = []
y_list = []
#---
for line in lines[1::]:
#print line
line = filter(None, re.split(";",line))
#print line
grdc_id = line[0]
river = line[1].strip()
station = line[2].strip()
ix = int(line[3])-1
iy = int(line[4])-1
ix2 = int(line[5])
#print river
if ix2 != -9999:
continue
if river == name:
id_list.append(grdc_id)
station_loc.append(station)
x_list.append(ix)
y_list.append(iy)
return id_list,station_loc,x_list,y_list
def get_loc_v394(gid):
#--ask the river name and a or b
# a - most downsream loc
# b - all locations
#grdc = pm.CaMa_dir() + "/map/glb_15min/grdc_loc.txt"
#grdc = "../data/GRDC_alloc.txt"
grdc = pm.CaMa_dir() + "/map/"+pm.mapname()+"/grdc_loc.txt"
#----
f = open(grdc,"r")
lines = f.readlines()
f.close()
ix=0
iy=0
gid=int(gid)
#---
for line in lines[1::]:
line = filter(None, re.split(" ",line))
grdc_id = int(line[0])
u_info = line[7]
#--
if gid==grdc_id:
#print "get_loc_v394", grdc_id
ix = int(line[8])-1
iy = int(line[9])-1
return ix,iy
def get_grdc_loc_dic(name,info = "a"):
#--aske the river name and a or b
# a - most downsream loc
# b - all locations
grdc = pm.CaMa_dir() + "/map/"+pm.mapname()+"/grdc_loc.txt"
#grdc = pm.CaMa_dir() + "/map/glb_15min/grdc_loc.txt"
#grdc = "../data/grdc_loc.txt"
f = open(grdc,"r")
lines = f.readlines()
f.close()
station_loc = {}
# print station_loc
for line in lines:
line = filter(None, re.split(" ",line))
grdc_id = line[0]
river = line[1]
station = line[2]
d_info = line[3]
station_loc[station] = ()
#--
lon = float(line[4])
lat = float(line[5])
#--
ix = int(line[6])-1
iy = int(line[7])-1
if info == d_info and info == "a":
station_loc[station] = (ix,iy)
else:
station_loc[station] = (ix,iy)
# print station_loc
return station_loc
def swot_data(yyyy,mm,dd):
# prepare sythetic observations using
# pre-simulated data
# river width thershold
rivwdth_thr=50.0 #m
nx,ny,gsize = pm.map_dimension()
ny_swot = min(ny,640)
day=SWOT_day(yyyy,mm,dd)
SWOTDD="%02d"%(day)
fname="../sat/mesh_day"+SWOTDD+".bin" # for glb_15min
mesh_in=np.fromfile(fname,np.float32).reshape([ny_swot,nx])
mesh=(mesh_in>=10)*(mesh_in<=60)
meshP=mesh-1000*(mesh<0.1)
SWOTmesh=np.zeros([ny,nx],np.float32)
SWOTmesh[40:680,:]=meshP
fname=pm.CaMa_dir()+"/map/"+pm.mapname()+"/rivwth_gwdlr.bin"
rivwth=np.fromfile(fname,np.float32).reshape(ny,nx)
obs=(SWOTmesh>=1.0)*(rivwth>=rivwdth_thr)*1.0
lname =[]
xlist =[]
ylist =[]
l_wse =[]
m_wse =[]
s_wse =[]
l_sat =[]
# leledif, lEGM08, lEGM96, satellite
#===================================
# odir="/cluster/data6/menaka/ensemble_org/CaMa_out/E2O003"
odir=pm.obs_dir()
fname=odir+"/sfcelv"+yyyy+".bin"
year=int(yyyy)
mon=int(mm)
day=int(dd)
if calendar.isleap(year):
nt=366
else:
nt=365
orgfile=np.fromfile(fname,np.float32).reshape([nt,ny,nx])
obs_err=SWOT_observation_error()
#-----------------------
start_dt=datetime.date(year,1,1)
target_dt=datetime.date(year,mon,day)
it=(target_dt-start_dt).days
for ix in np.arange(nx):
for iy in np.arange(ny):
if obs[iy,ix] == 1.0:
wse=orgfile[it,iy,ix] + err_rand(ix,iy)
# print (ix,iy,wse[0])
l_wse.append(wse[0])
xlist.append(ix+1)
ylist.append(iy+1)
m_wse.append(-9999.0)
s_wse.append(-9999.0)
l_sat.append("SWOT")
return xlist, ylist, l_wse, m_wse, s_wse, l_sat
def grdc_river_name_v396():
#grdc = pm.CaMa_dir() + "/map/glb_15min/grdc_loc.txt"
grdc = pm.CaMa_dir() + "/map/"+pm.mapname()+"/grdc_loc.txt"
f = open(grdc,"r")
lines = f.readlines()
f.close()
rivername = []
for line in lines[1::]:
line = filter(None, re.split(";",line))
grdc_id = line[0]
river = line[1].strip()
#d_info = line[3]
#if d_info == "a":
if river not in rivername:
rivername.append(river)
return rivername
def grdc_river_name():
# get river names
grdc = pm.CaMa_dir() + "/map/"+pm.mapname()+"/grdc_loc.txt"
#grdc = pm.CaMa_dir() + "/map/glb_15min/grdc_loc.txt"
#grdc = "../data/grdc_loc.txt"
f = open(grdc,"r")
lines = f.readlines()
f.close()
rivername = []
for line in lines:
line = filter(None, re.split(" ",line))
grdc_id = line[0]
river = line[1]
d_info = line[3]
if d_info == "a":
rivername.append(river)
return rivername
def get_grdc_loc(name,info = "a"):
#--ask the river name and a or b
# a - most downsream loc
# b - all locations
grdc = pm.CaMa_dir() + "/map/"+pm.mapname()+"/grdc_loc.txt"
#grdc = "../data/grdc_loc.txt"
f = open(grdc,"r")
lines = f.readlines()
f.close()
station_loc = []
x_list = []
y_list = []
#---
for line in lines:
line = filter(None, re.split(" ",line))
grdc_id = line[0]
river = line[1]
station = line[2]
d_info = line[3]
#--
lon = float(line[4])
lat = float(line[5])
#--
ix = int(line[6])-1
iy = int(line[7])-1
if name == river:
if info == "a":
if d_info=="a":
station_loc.append(station)
x_list.append(ix)
y_list.append(iy)
else:
station_loc.append(station)
x_list.append(ix)
y_list.append(iy)
return station_loc,x_list,y_list
def get_id(name):
#--get GRDC id
#grdc = pm.CaMa_dir() + "/map/glb_15min/grdc_loc.txt"
#grdc = "../data/grdc_loc.txt"
grdc = pm.CaMa_dir() + "/map/"+pm.mapname()+"/grdc_loc.txt"
f = open(grdc,"r")
lines = f.readlines()
f.close()
gid=-9999
#---
for line in lines:
line = filter(None, re.split(" ",line))
grdc_id = int(line[0])
river = line[1]
station = line[2]
d_info = line[3]
#--
if name == station:
gid=grdc_id
return gid
input:
s: simulated
o: observed
output:
RMSE: Root Mean Squre Error
"""
o=ma.masked_where(o<=0.0,o).filled(0.0)
s=ma.masked_where(o<=0.0,s).filled(0.0)
s,o = filter_nan(s,o)
# return np.sqrt(np.mean((s-o)**2))
return np.sqrt(np.ma.mean(np.ma.masked_where(o<=0.0,(s-o)**2)))
#====================================================================
mk_dir(assim_out+"/figures")
mk_dir(assim_out+"/figures/disgraph")
#----
fname=pm.CaMa_dir()+"/map/"+pm.mapname()+"/params.txt"
f=open(fname,"r")
lines=f.readlines()
f.close()
#-------
nx = int(filter(None, re.split(" ",lines[0]))[0])
ny = int(filter(None, re.split(" ",lines[1]))[0])
gsize = float(filter(None, re.split(" ",lines[3]))[0])
#----
syear,smonth,sdate=pm.starttime()#2004#1991 #2003,1,1 # 2009,1,1 #
eyear,emonth,edate=pm.endtime() #2005,1,1 #2004,1,1 # 2004,1,1 # 2010,1,1 # 2012,1,1 # 2011,1,1 #
#month=1
#date=1
start_dt=datetime.date(syear,smonth,sdate)
end_dt=datetime.date(eyear,emonth,edate)
size=60
output:
RMSE: Root Mean Squre Error
"""
o = ma.masked_where(o == -9999.0, o).filled(0.0)
s = ma.masked_where(o == -9999.0, s).filled(0.0)
o = np.compress(o > 0.0, o)
s = np.compress(o > 0.0, s)
s, o = filter_nan(s, o)
return np.sqrt(np.mean((s - o)**2))
#==========================================================
mk_dir(assim_out + "/figures")
mk_dir(assim_out + "/figures/anomaly")
#----
fname = pm.CaMa_dir() + "/map/" + pm.mapname() + "/params.txt"
f = open(fname, "r")
lines = f.readlines()
f.close()
#-------
nx = int(filter(None, re.split(" ", lines[0]))[0])
ny = int(filter(None, re.split(" ", lines[1]))[0])
gsize = float(filter(None, re.split(" ", lines[3]))[0])
#---
# year=2004
# month=1
# date=1
# start_dt=datetime.date(year,month,date)
# size=60
syear, smonth, sdate = pm.starttime() #2004#1991 #2003,1,1 #
llcrnrlon=west,
urcrnrlon=east,
lat_ts=0,
resolution='c')
#m.drawcoastlines( linewidth=0.3, color='k' )
m.fillcontinents(color=land, lake_color=water)
m.drawmapboundary(fill_color=water)
m.drawparallels(np.arange(south, north + 0.1, 20),
labels=[1, 0, 0, 0],
fontsize=10,
linewidth=0.1)
m.drawmeridians(np.arange(west, east + 0.1, 40),
labels=[0, 0, 0, 1],
fontsize=10,
linewidth=0.1)
fname = pm.CaMa_dir() + "/map/glb_15min/outclm.bin"
trueforo = np.fromfile(fname, np.float32).reshape([2, 720, 1440])[0]
river = (trueforo > 100.) * 1.0
ratio = np.ma.fix_invalid(ratio).data
ratio = ratio * river
data = ratio[npix:spix, wpix:epix]
im = m.imshow(np.flipud(data),
vmin=1e-20,
vmax=1,
interpolation="nearest",
cmap=cmap,
zorder=100)
#im = m.imshow(np.flipud(ai),vmin=1e-20, vmax=1,interpolation="nearest",cmap=cmap,zorder=100)
cbar = m.colorbar(im, "right", size="2%")
cbar.set_label("annual mean AI")
plt.title("annual mean Assimilation Index ") #+yyyy+"-"+mm+"-"+dd)
def latlon_river(rivername,ix,iy,mapname="glb_06min",nYY=1800,nXX=3600):
global lllat, urlat, lllon, urlon
lonlat = pm.CaMa_dir()+"/map/glb_"+mapname[-5::]+"/lonlat.bin"
lonlat = np.fromfile(lonlat,np.float32).reshape(2,nYY,nXX)
llon=lonlat[0,iy-1,ix-1]
llat=lonlat[1,iy-1,ix-1]
adj=20.0
lllat, urlat, lllon, urlon = max(llat-adj,-90.),min(llat+adj,90.),max(llon-adj,-180.),min(llon+adj,180.)
if rivername=="LENA":
lllat = 50.
urlat = 80.
lllon = 100.
urlon = 145.
if rivername=="NIGER":
lllat = 0.
urlat = 25.
lllon = -10.
urlon = 15.
if rivername=="AMAZONAS" or rivername=="AMAZON":
lllat = -20.
urlat = 10.
lllon = -80.
urlon = -45.
if rivername=="MEKONG":
lllat = 10.
urlat = 35.
lllon = 90.
urlon = 120.
if rivername=="MISSISSIPPI":
lllat = 20.
urlat = 50.
lllon = -115.
urlon = -75.
if rivername=="OB":
lllat = 40.
urlat = 70.
lllon = 55.
urlon = 95.
if rivername=="CONGO":
lllat = -15.
urlat = 10.
lllon = 10.
urlon = 35.
if rivername=="INDUS":
lllat = 20.
urlat = 40.
lllon = 60.
urlon = 80.
if rivername=="VOLGA":
lllat = 40.
urlat = 65.
lllon = 30.
urlon = 70.
if rivername=="NILE":
lllat = -5.
urlat = 30.
lllon = 20.
urlon = 40.
if rivername=="YUKON":
lllat = 55.
urlat = 75.
lllon = -165.
urlon = -130.
#if rivername not in ["LENA","NIGER","CONGO","OB","MISSISSIPPI","MEKONG","AMAZONAS","INDUS"]:
# adj=20.
# lllat, urlat, lllon, urlon = max(llat-adj,-90.),min(llat+adj,90.),max(llon-adj,-180.),min(llon+adj,180.)
return lllat, urlat, lllon, urlon
RMSE: Root Mean Squre Error
"""
o = ma.masked_where(o == -9999.0, o).filled(0.0)
s = ma.masked_where(o == -9999.0, s).filled(0.0)
o = np.compress(o > 0.0, o)
s = np.compress(o > 0.0, s)
s, o = filter_nan(s, o)
# return np.sqrt(np.mean((s-o)**2))
return np.sqrt(np.ma.mean(np.ma.masked_where(o <= 0.0, (s - o)**2)))
#==========================================================
mk_dir(assim_out + "/figures")
mk_dir(assim_out + "/figures/RMSE")
#----
fname = pm.CaMa_dir() + "/map/" + pm.mapname() + "/params.txt"
f = open(fname, "r")
lines = f.readlines()
f.close()
#-------
nx = int(filter(None, re.split(" ", lines[0]))[0])
ny = int(filter(None, re.split(" ", lines[1]))[0])
gsize = float(filter(None, re.split(" ", lines[3]))[0])
lon0 = float(filter(None, re.split(" ", lines[4]))[0])
lat0 = float(filter(None, re.split(" ", lines[7]))[0])
west = float(filter(None, re.split(" ", lines[4]))[0])
east = float(filter(None, re.split(" ", lines[5]))[0])
south = float(filter(None, re.split(" ", lines[6]))[0])
north = float(filter(None, re.split(" ", lines[7]))[0])
#----
nextxy = pm.CaMa_dir() + "/map/" + pm.mapname() + "/nextxy.bin"
os.remove(dst)
os.symlink(src,dst)
else:
raise
#--
def make_yearlist(syear,smon,sday,N=365):
start_dt=datetime.date(syear,smon,sday)
fname="year_day.txt"
f=open(fname,"w")
for day in np.arange(0,N):
nw_dt = start_dt + datetime.timedelta(days=day)
yyyy,mm,dd = nw_dt.year, nw_dt.month, nw_dt.day
yyyymmdd = "%04d%02d%02d\n"%(yyyy,mm,dd)
f.write(yyyymmdd)
f.close()
#--
slink("../gosh/params.py", "params.py")
import params as pm
# define values
N=366 #2004
exp="E2O_womc_anomalyDA_if3.0"
outdir=pm.DA_dir()+"/out/"+exp
# mkdir stat
mk_dir(outdir+"/assim_out/stat")
# make list of days
#make_yearlist(2004,1,1,N=366)
# compile
#os.system("ifort calc_stat.f90 -o calc_stat -O3 -assume byterecl -heap-arrays -g -traceback -free -parallel")
# run calc_stat
os.system("./calc_stat "+str(N)+" "+str(pm.ens_mem())+" "+outdir+" "+pm.CaMa_dir())
