day=(date-year*10000-month*100)
hour=time//100
min=(time-hour*100)
doy=[ datetime.datetime(np.int(year[i]),np.int(month[i]),np.int(day[i]),\
np.int(hour[i]),np.int(min[i]),0)- \
datetime.datetime(first_year,1,1,0,0,0) \
for i in range(len(year))]
processed_dates=[doy[i].days+doy[i].seconds/(24.*60.*60.) for i in range(len(doy))]
return processed_dates
#read obs files
all_files = glob.glob('/work/home/db876/observations/surface/%s/EMEP/*'%(species))
all_files = modules.natsorted(all_files)
#get all refs
ref_list = []
valid_refs=[]
for i in range(len(all_files)):
f = all_files[i]
f = f.replace("/work/home/db876/observations/surface/%s/EMEP/"%(species), "")
f = f[:7]
ref_list.append(f)
refs = set(ref_list)
refs = sorted([i for i in refs])
refs = np.array(refs)
print 'all refs len = ', len(refs)
elif species == "NO":
data_resolution = 1.1
mol_mass = 30.01
elif species == "NO2":
data_resolution = 1.1
mol_mass = 46.0055
elif species == "CO":
data_resolution = 100.1
mol_mass = 28.01
# read obs files
hourly_files = glob.glob("/work/home/db876/observations/surface/%s/GAW/*.hr*.dat" % (species))
hourly_files = modules.natsorted(hourly_files)
daily_files = glob.glob("/work/home/db876/observations/surface/%s/GAW/*.da.*" % (species))
daily_files = modules.natsorted(daily_files)
monthly_files = glob.glob("/work/home/db876/observations/surface/%s/GAW/*.mo.*" % (species))
monthly_files = modules.natsorted(monthly_files)
# get all refs
ref_list_hourly = []
ref_list_daily = []
ref_list_monthly = []
valid_refs = []
for i in range(len(hourly_files)):
f = hourly_files[i]
f = f.replace("/work/home/db876/observations/surface/%s/GAW/" % (species), "")
def site_iter_process(valid_refs, c):
# for r in range(len(valid_refs)):
ref = valid_refs[c]
print ref
# get site instrument for species
met_i = met_refs.index(ref)
print len(met_refs)
print len(met_species)
site_species = list(met_species[met_i])
site_instruments = list(met_instruments[met_i])
print site_species
print site_instruments
mm = site_instruments[site_species.index(species)]
site_resolutions = []
data_valid = True
s_files = insensitive_glob("/work/home/db876/observations/surface/%s/EANET/*%s.csv" % (species, ref))
site_files = []
for y in year_array:
for f in s_files:
if str(y)[-2:] in f:
site_files.append(f)
site_files = modules.natsorted(site_files)
if site_files == []:
print "No files for ref.\n"
years = []
months = []
days = []
hours = []
vals = []
last_year_index = len(site_files)
for y in year_array:
print "Processing Year %s" % y
got_year = False
for file in site_files:
last_file_split = file.split("/")[-1]
if str(y)[2:] in last_file_split:
got_year = True
break
if got_year == False:
# fill in data for missing year
timedelta_diff = datetime.date(y + 1, 1, 1) - datetime.date(y, 1, 1)
ndays_missing = timedelta_diff.days
print "ndays missing = ", ndays_missing
vals = np.append(vals, [-99999] * (ndays_missing * 24))
continue
print file
valid = True
with open(file, "rb") as f:
reader = csv.reader(f, delimiter=",")
counter = 0
# get resolution
for row in reader:
if counter == 0:
all_units = row
elif counter == 1:
file_res = "H"
try:
hour_index = row.index("Hour")
except:
file_res = "D"
try:
day_index = row.index("Day")
except:
file_res = "M"
month_index = row.index("Month")
year_index = row.index("Year")
try:
spec_index = row.index(species.upper())
units = all_units[spec_index]
except:
valid = False
break
# make sure each year units are ppb
if units != "ppb":
print "Units not ppb!"
1 + "a"
if counter == 2:
if file_res == "H":
yyyy = row[year_index]
mm = row[month_index]
dd = row[day_index]
hh = row[hour_index]
elif file_res == "D":
yyyy = row[year_index]
mm = row[month_index]
dd = row[day_index]
hh = 1
elif file_res == "M":
yyyy = row[year_index]
mm = row[month_index]
dd = 1
hh = 1
start_datetime = datetime.datetime(int(yyyy), int(mm), int(dd), int(hh))
if counter == 3:
if file_res == "H":
yyyy = row[year_index]
mm = row[month_index]
dd = row[day_index]
hh = row[hour_index]
elif file_res == "D":
yyyy = row[year_index]
mm = row[month_index]
dd = row[day_index]
hh = 1
elif file_res == "M":
yyyy = row[year_index]
mm = row[month_index]
dd = 1
hh = 1
present_datetime = datetime.datetime(int(yyyy), int(mm), int(dd), int(hh))
time_delt = present_datetime - start_datetime
hour_delt = datetime.timedelta(hours=1)
day_delt = datetime.timedelta(hours=24)
week_delt = datetime.timedelta(hours=24 * 7)
month_delt = datetime.timedelta(hours=24 * 28)
print time_delt
if time_delt < day_delt:
print "Hourly Data"
file_res = "H"
site_resolutions.append(file_res)
elif (time_delt > hour_delt) & (time_delt < week_delt):
print "Daily Data"
file_res = "D"
site_resolutions.append(file_res)
elif time_delt > week_delt:
print "Monthly Data"
file_res = "M"
site_resolutions.append(file_res)
# break
# limit files by timeres return if not suitable for output res
if output_res == "H":
if (file_res == "D") or (file_res == "M"):
print "File resolution has to be Minimum Hourly. Skipping"
data_valid = False
return c, vals, data_valid, -999, -999, -999, "na", "na", "na", "na", "na", -999
elif output_res == "D":
if file_res == "M":
print "File resolution has to be Minimum Daily. Skipping"
data_valid = False
return c, vals, data_valid, -999, -999, -999, "na", "na", "na", "na", "na", -999
counter += 1
# READ IN DATA
if valid == True:
with open(file, "rb") as f:
reader = csv.reader(f, delimiter=",")
counter = 0
for row in reader:
if counter >= 2:
yyyy = row[year_index]
mm = row[month_index]
if file_res == "H":
try:
vals = np.append(vals, np.float64(row[spec_index]))
except:
vals = np.append(vals, -99999)
elif file_res == "D":
try:
vals = np.append(vals, [np.float64(row[spec_index])] * 24)
except:
vals = np.append(vals, [-99999] * 24)
elif file_res == "M":
month_days = monthrange(int(yyyy), int(mm))[1]
try:
vals = np.append(vals, [np.float64(row[spec_index])] * (month_days * 24))
except:
vals = np.append(vals, [-99999] * (month_days * 24))
counter += 1
else:
print "Species is not in file header. Skipping Year"
timedelta_diff = datetime.date(y + 1, 1, 1) - datetime.date(y, 1, 1)
ndays_missing = timedelta_diff.days
print "ndays missing = ", ndays_missing
vals = np.append(vals, [-99999] * (ndays_missing * 24))
valid_refs_rev.append(ref)
i_ref = met_refs.index(ref)
tz = np.float64(met_tz[i_ref])
lat = np.float64(met_lats[i_ref])
lon = np.float64(met_lons[i_ref])
alt = np.float64(met_alts[i_ref])
raw_class_name = met_class[i_ref]
anthrome_class_name = class_name[i_ref]
# check tz is whole number else skip site
if (tz % 1) != 0:
print "Timezone is not even. Skipping"
data_valid = False
tz = int(tz)
# correct time to UTC
if tz < 0:
# get rid of values at start and append -99999's at end
cut = vals[:tz]
for num in range(np.abs(tz)):
cut = np.insert(cut, 0, -99999)
vals = cut
elif tz > 0:
# put -99999's at start and get rid of values at end
cut = vals[tz:]
for num in range(tz):
cut = np.append(cut, -99999)
vals = cut
# do data quality checks
full_data, data_valid, data_complete = modules.quality_check_periodic(
vals, data_valid, data_resolution, np.float64(alt), grid_dates, start_year, end_year
)
# if all site resolutions are same continue, make program exit
all_same = all(x == site_resolutions[0] for x in site_resolutions)
if all_same == True:
pass
else:
print "Not all files for site have same resolution. Skipping."
data_valid = False
return c, full_data, data_valid, -999, -999, -999, "na", "na", "na", "na", "na", -999
# set sampling as average
st = "average"
return (
c,
full_data,
data_valid,
lat,
lon,
alt,
raw_class_name,
anthrome_class_name,
mm,
st,
file_res,
data_complete,
)
elif (species == 'NO'):
data_resolution = 1.1
param_code = 42601
elif (species == 'NO2'):
data_resolution = 1.1
param_code = 42602
elif species == 'ISOP':
data_resolution = 0.6
param_code = 43243
files = glob.glob('/work/home/db876/observations/surface/%s/AQS/*'%(species))
files=modules.natsorted(files)
year_array = np.arange(start_year,end_year+1)
valid_files = []
for i in year_array:
for f in files:
if str(i) in f:
valid_files.append(f)
print valid_files
#create ref lists
all_site_refs = []
test_site_refs = []
uniq_refs = []
doy=[ datetime.datetime(np.int(year[i]),np.int(month[i]),np.int(day[i]),\
np.int(hour[i]),np.int(min[i]),0)- \
datetime.datetime(first_year,1,1,0,0,0) \
for i in range(len(year))]
processed_dates = [
doy[i].days + doy[i].seconds / (24. * 60. * 60.)
for i in range(len(doy))
]
return processed_dates
#read obs files
all_files = glob.glob('/work/home/db876/observations/surface/%s/EMEP/*' %
(species))
all_files = modules.natsorted(all_files)
#get all refs
ref_list = []
valid_refs = []
for i in range(len(all_files)):
f = all_files[i]
f = f.replace("/work/home/db876/observations/surface/%s/EMEP/" % (species),
"")
f = f[:7]
ref_list.append(f)
refs = set(ref_list)
refs = sorted([i for i in refs])
refs = np.array(refs)
start_year = start_years[y]
end_year = end_years[y]
print start_year, end_year
year_array = range(start_year, end_year + 1)
if (species == "O3") or (species == "NO") or (species == "NO2"):
data_resolution = 1.1
n_years = (end_year - start_year) + 1
# read obs files
all_files = glob.glob("/work/home/db876/observations/surface/%s/EANET/AT*" % (species))
all_files = modules.natsorted(all_files)
ref_list = []
valid_refs = []
valid_refs_rev = []
for i in range(len(all_files)):
f = all_files[i].replace(".csv", "")
f = f.replace("/work/home/db876/observations/surface/%s/EANET/" % (species), "")
f = f[4:]
f = f.lower()
ref_list.append(f)
refs = set(ref_list)
refs = [i for i in refs]
refs = sorted(refs)
def site_iter_process(valid_refs, c):
#set local counts
inv_nometa = 0
inv_anyvaliddata = 0
inv_nokeymeta = 0
inv_resolution = 0
inv_badmeasurementmethod = 0
n_all = 0
n_after_nometa = 0
n_after_flagsandlod = 0
n_after_duplicate = 0
n_after_anyvaliddata = 0
n_after_nokeymeta = 0
n_after_resolution = 0
n_after_badmeasurementmethod = 0
#set local unknown lists
unknown_mm_list = []
unknown_mm_refs_list = []
unknown_local_tz_list = []
#for ref_i in range(len(valid_refs)):
data_valid = True
site_ref = valid_refs[c]
print 'Current Ref is = ', site_ref, c
s_files = glob.glob(
'/work/home/db876/observations/surface/%s/CAPMON/ozon_smpls_%s*' %
(species, site_ref))
site_files = []
for y in year_array:
for f in s_files:
if str(y) in f:
site_files.append(f)
site_files = modules.natsorted(site_files)
yymmdd = []
hhmm = []
vals = []
for file_i in range(len(site_files)):
count = 0
meta_start = -99999
start_read_1 = False
start_read_2 = False
with open(site_files[file_i], 'rb') as f:
reader = csv.reader(f, delimiter=',')
print site_files[file_i]
for row in reader:
#print count
#break out of loop at bottom of file
if (start_read_2 == True) & (row[0] == '*TABLE ENDS'):
break
#get metadata
try:
if (row[0] == '*TABLE NAME') & (row[1]
== 'Site information'):
meta_start = count + 2
except:
pass
if count == meta_start:
siteid_i = row.index('Site ID: standard')
sitename_i = row.index('Description')
lat_i = row.index('Latitude: decimal degrees')
lon_i = row.index('Longitude: decimal degrees')
try:
alt_i = row.index(
'Ground elevation: above mean sea level')
except:
alt_i = row.index('Ground altitude')
class_i = row.index('Site land use')
if count == (meta_start + 6):
latitude = row[lat_i]
longitude = row[lon_i]
altitude = row[alt_i]
raw_class_name = row[class_i]
site_name = row[sitename_i]
#get data
if start_read_2 == True:
#read dates, times, and vals
date = row[8]
time = row[9]
yymmdd.append(date[:4] + date[5:7] + date[8:])
hhmm.append(time[:2] + time[3:])
quality_code = row[13]
#if flag not equal to V0 then make -99999
if quality_code == 'V0':
vals = np.append(vals, np.float64(row[12]))
else:
vals = np.append(vals, -99999)
try:
if (row[0] == '*TABLE NAME') & (row[1] == 'OZONE_HOURLY'):
start_read_1 = True
except:
pass
if (start_read_1 == True) & (row[0] == '*TABLE COLUMN UNITS'):
unit = row[12]
if (start_read_1 == True) & (row[0] == '*TABLE BEGINS'):
start_read_2 = True
count += 1
#add to n_obs_all
n_all += len(vals)
n_after_nometa += len(vals)
#convert data less < 0 to -99999
test_inv = vals < 0
vals[test_inv] = -99999
#create max possible grid
full_data = np.empty(n_hours)
full_data_after_flagsandlod = np.empty(n_hours)
big_n_dup_array = np.zeros(n_hours)
full_data[:] = -99999
full_data_after_flagsandlod[:] = -99999
#put vals into full grid
date_con = np.array(yymmdd).astype(int)
time_con = np.array(hhmm).astype(int)
#remove data < 1970 and >= 2015
test_inds = (date_con >= 19700101) & (date_con < 20150101)
date_con = date_con[test_inds]
time_con = time_con[test_inds]
vals = vals[test_inds]
#set st_big and mm_big
st_big = ['continuous'] * len(vals)
mm_big = ['ultraviolet photometry'] * len(vals)
#get obs valid
test = vals != -99999
n_obs_valid = len(vals[test])
n_after_flagsandlod += n_obs_valid
#find matching times between actual times and grid of times, return big array of indices of matched indices in grid
converted_time = modules.date_process(date_con, time_con, start_year)
converted_time = np.round(converted_time, decimals=5)
syn_grid_time = np.arange(0, n_days, 1. / 24)
syn_grid_time = np.round(syn_grid_time, decimals=5)
raw_indices = np.searchsorted(syn_grid_time, converted_time, side='left')
vals = np.array(vals)
full_data_after_flagsandlod[raw_indices] = vals
raw_st = np.copy(st_big)
raw_mm = np.copy(mm_big)
# test and remove duplicate and overlap points
converted_time, vals, mm_big, st_big, na = modules.remove_duplicate_points(
site_ref, converted_time, vals, mm_big, st_big, 'blank', output_res)
test = vals >= 0
n_obs_valid = int(len(vals[test]))
n_after_duplicate += n_obs_valid
#find matching times between actual times and grid of times, return big array of indices of matched indices in grid
indices = np.searchsorted(syn_grid_time, converted_time, side='left')
full_data[indices] = vals
#get metadata
try:
lat = np.float32(latitude)
except:
lat = 'na'
try:
lon = np.float32(longitude)
except:
lon = 'na'
try:
alt = np.float32(altitude)
except:
alt = 'na'
unit = str(unit)
raw_class_name = str(raw_class_name)
site_name = str(site_name)
country = 'Canada'
contact = 'Dave MacTavish, 4905 Dufferin St., Toronto ON, CANADA, M3H 5T4, [email protected]'
#set data tz - all CAPMON times are UTC
data_tz = 0
all_tz = [data_tz]
key_meta = [lat, lon, alt]
#set site file resolution
file_res = 'H'
#get sampling/instrument grids
raw_st_grid, p_st_grid, p_st_grid_after_flagsandlod, raw_mm_grid, p_mm_grid, p_mm_grid_after_flagsandlod, unknown_mm_list, unknown_mm_refs_list = modules.sampling_and_instruments_by_processgroup(
site_ref, process_group, species, raw_st, raw_mm,
full_data_after_flagsandlod, full_data, raw_indices, unknown_mm_list,
unknown_mm_refs_list, no2_type)
#do quality checks
data_valid, full_data, valid_hours_dup, p_st_grid, p_mm_grid, n_all, inv_nometa, n_after_nometa, n_after_flagsandlod, n_after_duplicate, inv_anyvaliddata, n_after_anyvaliddata, inv_nokeymeta, n_after_nokeymeta, inv_resolution, n_after_resolution, inv_badmeasurementmethod, n_after_badmeasurementmethod, exit_r = modules.primary_quality_control(
site_ref, species, file_res, no2_type, grid_dates, full_data,
big_n_dup_array, 0, raw_st_grid, p_st_grid,
p_st_grid_after_flagsandlod, raw_mm_grid, p_mm_grid,
p_mm_grid_after_flagsandlod, data_resolution, n_obs_valid, key_meta,
n_all, inv_nometa, n_after_nometa, n_after_flagsandlod,
n_after_duplicate, inv_anyvaliddata, n_after_anyvaliddata,
inv_nokeymeta, n_after_nokeymeta, inv_resolution, n_after_resolution,
inv_badmeasurementmethod, n_after_badmeasurementmethod)
if data_valid == False:
exit_c_list = np.array([
inv_nometa, inv_anyvaliddata, inv_nokeymeta, inv_resolution,
inv_badmeasurementmethod
])
n_c_list = np.array([
n_all, n_after_nometa, n_after_flagsandlod, n_after_duplicate,
n_after_anyvaliddata, n_after_nokeymeta, n_after_resolution,
n_after_badmeasurementmethod
])
unknown_list = [
unknown_mm_list, unknown_mm_refs_list, unknown_local_tz_list
]
meta = [
lat, lon, alt, 'na', 'na', 'na', 'na', 'na', 'na', 'na', 'na', 'na'
]
return c, ['na'], ['na'], [
'na'
], False, meta, exit_c_list, n_c_list, unknown_list, exit_r, np.zeros(
1)
#set processed unit
p_unit = 'pbbv'
#get local timezone
try:
local_tz_name = tz_root.tzNameAt(lat, lon, forceTZ=True)
pytz_obj = pytz.timezone(local_tz_name)
datetime_offset = pytz_obj.utcoffset(datetime.datetime(2000, 1, 1))
if datetime_offset < datetime.timedelta(0):
local_tz = -(24 - int(datetime_offset.seconds / 60 / 60))
else:
local_tz = int(datetime_offset.seconds / 60 / 60)
except:
local_tz = 'na'
print 'TIMEZONE NOT KNOWN, SITE IS %s' % (site_ref)
unknown_local_tz_list.append(site_ref)
#pack meta
meta = [
lat, lon, alt, raw_class_name, file_res, unit, p_unit, data_tz,
local_tz, site_name, country, contact
]
#if blank strings in meta then convert to 'na'
for i in range(len(meta)):
try:
if meta[i].strip() == '':
meta[i] = 'na'
except:
pass
print set(raw_st_grid)
print set(raw_mm_grid)
print set(p_st_grid)
print set(p_mm_grid)
print meta
exit_c_list = np.array([
inv_nometa, inv_anyvaliddata, inv_nokeymeta, inv_resolution,
inv_badmeasurementmethod
])
n_c_list = np.array([
n_all, n_after_nometa, n_after_flagsandlod, n_after_duplicate,
n_after_anyvaliddata, n_after_nokeymeta, n_after_resolution,
n_after_badmeasurementmethod
])
print 'exit counts = ', exit_c_list
print 'n obs counts = ', n_c_list
unknown_list = [
unknown_mm_list, unknown_mm_refs_list, unknown_local_tz_list
]
return c, full_data, p_st_grid, p_mm_grid, data_valid, meta, exit_c_list, n_c_list, unknown_list, 'na', big_n_dup_array
#setup netcdf output
root_grp = Dataset(
'%s_RADIOSONDES_SHADOZ_%s_%s.nc' % (species, start_year, end_year), 'w')
root_grp.description = 'SHADOZ Radiosondes of %s at sites in ppb - Program written by Dene Bowdalo' % (
species)
site_count = 0
#---------------------------------------------------------------------------------
#process SHADOZ data
print '\nProcessing SHADOZ data\n'
files = glob.glob('/work/home/db876/observations/ozonesonde/SHADOZ/*')
files = modules.natsorted(files)
site_names = []
#separate out sites into each location
for i in files:
i = i.replace('/work/home/db876/observations/ozonesonde/SHADOZ/', '')
split = i.split('_')
site_names.append(split[0])
site_names = np.sort(list(set(site_names)))
for site_name in site_names:
print '\n'
print site_name
data_valid = True
exit_resolution_lats = []
exit_resolution_lons = []
exit_resolution_pg = []
exit_badmeasurementmethod_refs = []
exit_badmeasurementmethod_lats = []
exit_badmeasurementmethod_lons = []
exit_badmeasurementmethod_pg = []
unknown_mm = []
unknown_mm_refs = []
unknown_local_tz = []
#read obs files
all_files = glob.glob('/work/home/db876/observations/surface/%s/CAPMON/*' %
(species))
all_files = modules.natsorted(all_files)
#get all refs
ref_list = []
valid_refs = []
for i in range(len(all_files)):
f = all_files[i]
f = f.replace(
"/work/home/db876/observations/surface/%s/CAPMON/" % (species), "")
f = f[11:14]
ref_list.append(f)
refs = set(ref_list)
refs = [i for i in refs]
refs = sorted(refs)
def site_iter_process(valid_refs,c):
#set local counts
inv_nometa = 0
inv_anyvaliddata = 0
inv_nokeymeta = 0
inv_resolution = 0
inv_badmeasurementmethod = 0
n_all = 0
n_after_nometa = 0
n_after_flagsandlod = 0
n_after_duplicate = 0
n_after_anyvaliddata = 0
n_after_nokeymeta = 0
n_after_resolution = 0
n_after_badmeasurementmethod = 0
#set local unknown lists
unknown_mm_list = []
unknown_mm_refs_list = []
unknown_local_tz_list = []
#read files site at a time
#for ref_i in range(len(valid_refs)):
site_ref = valid_refs[c]
all_latitudes = []
all_longitudes = []
all_altitudes = []
all_unit = []
all_site_name = []
all_country = []
all_contact = []
mm_big = []
meta_valid_list = []
data_valid = True
print 'Current Ref is = ', site_ref,c
#find if sites have full valid range from start year and finishing in end year
s_files = glob.glob('/work/home/db876/observations/surface/%s/EMEP/%s*'%(fname_species,site_ref))
year_files = [file.replace("/work/home/db876/observations/surface/%s/EMEP/"%(fname_species), "") for file in s_files]
cut_year_files = [file[8:12] for file in year_files]
site_files = []
for y in year_array:
for i in range(len(s_files)):
if str(y) in cut_year_files[i]:
site_files.append(s_files[i])
site_files = modules.natsorted(site_files)
#test for duplicate file years, if duplicates break processing
file_years = []
for file in site_files:
last_file_split = file.split('/')[-1]
file_years=np.append(file_years,last_file_split[8:12])
for y in year_array:
test = file_years == str(y)
if len(file_years[test]) > 1:
print 'Site has duplicate files for %s. Breaking processing'%(y)
1+'a'
if site_files == []:
print 'No valid files for site\n'
return
#remove daily/monthly files if necessary
if output_res == 'H':
del_i = []
for i in range(len(site_files)):
if '.1d.' in site_files[i]:
del_i.append(i)
elif '.1mo.' in site_files[i]:
del_i.append(i)
site_files=np.delete(site_files,del_i)
elif output_res == 'HD':
del_i = []
for i in range(len(site_files)):
if '.1mo.' in site_files[i]:
del_i.append(i)
site_files=np.delete(site_files,del_i)
for y in year_array:
bad_meta = False
got_year = False
for file in site_files:
last_file_split = file.split('/')[-1]
if str(y) in last_file_split[8:12]:
got_year = True
break
if got_year == False:
#fill in data for missing year
timedelta_diff = datetime.date(y+1, 1, 1) - datetime.date(y, 1, 1)
ndays_missing = timedelta_diff.days
continue
count = 0
with open(file, 'rb') as f:
reader = csv.reader(f,delimiter=' ')
print file
for row in reader:
try:
row = filter(lambda a: a != '', row)
except:
pass
try:
row = filter(lambda a: a != ',', row)
except:
pass
#get start date of file
if row[0] == 'Startdate:':
data = row[1]
s_yyyy = data[:4]
s_mm = data[4:6]
s_dd = data[6:8]
s_hh = data[8:10]
s_min = data[10:12]
start_datetime = datetime.datetime(int(s_yyyy),1,1,0,0)
#get unit
if row[0] == 'Unit:':
try:
if len(row) == 3:
unit_part1 = row[1]
unit_part2 = row[2]
unit = unit_part1+'_'+unit_part2
elif len(row) == 2:
unit = row[1]
all_unit.append(unit)
except:
bad_meta = True
#get resolution
if row[0] == 'Resolution':
if row[1] == 'code:':
file_res = row[2]
print 'Resolution = %s'%file_res
#get latitude
if row[0] == 'Station':
if row[1] == 'latitude:':
latitude = row[2]
all_latitudes.append(latitude)
#get longitude
if row[0] == 'Station':
if row[1] == 'longitude:':
longitude = row[2]
all_longitudes.append(longitude)
#get altitude
if row[0] == 'Station':
if row[1] == 'altitude:':
altitude = row[2][:-1]
all_altitudes.append(altitude)
#get site name
if row[0] == 'Station':
if row[1] == 'name:':
site_name = row[2]
all_site_name.append(site_name)
#get period
if row[0] == 'Period':
period_code = row[2]
#get stats method
if row[0] == 'Statistics:':
try:
st = row[1] + row[2]
if st != 'arithmeticmean':
print 'Not Arithmetic Mean!'
print row[1]
print 1+'a'
except:
print 'Not Arithmetic Mean!'
print row[1]
print 1+'a'
#get instrument method and name
if row[0] == 'Instrument':
if row[1] == 'type:':
mm_list = row[2:]
if len(mm_list) > 1:
site_mm = ''
for x in range(len(mm_list)):
site_mm = site_mm+mm_list[x]+' '
site_mm = site_mm.strip()
else:
site_mm = mm_list[0]
if row[1] == 'name:':
mn_list = row[2:]
if len(mn_list) > 1:
site_mn = ''
for x in range(len(mn_list)):
site_mn = site_mn+mn_list[x]+' '
site_mn = site_mn.strip()
else:
site_mn = mn_list[0]
#get method ref
if row[0] == 'Method':
if row[1] == 'ref:':
try:
mf_list = row[2:]
if len(mf_list) > 1:
site_mf = ''
for x in range(len(mf_list)):
site_mf = site_mf+mf_list[x]+' '
site_mf = site_mf.strip()
else:
site_mf = mf_list[0]
except:
site_mf = ''
#put together intrument type+instrument_name+method_ref
mm = site_mm+site_mn+site_mf
#get contact
if row[0] == 'Originator:':
try:
contact_list = row[1:]
if len(contact_list) > 1:
site_contact = ''
for x in range(len(mf_list)):
site_contact = site_contact+contact_list[x]+' '
site_contact = site_contact.strip()
else:
site_contact = site_contact[0]
except:
site_contact = ''
all_contact.append(site_contact)
#get country
site_country = EMEP_COUNTRIES(file.split('/')[-1][:2])
all_country.append(site_country)
if row[0] == 'starttime':
skip_n = count+1
if species == 'ISOP':
spec_ind = row.index('C5H8')
try:
flag_ind = row.index('flag_C5H8')
except:
flag_ind = row.index('flag')
else:
spec_ind = row.index(species)
try:
flag_ind = row.index('flag_'+species)
except:
flag_ind = row.index('flag')
count+=1
read = np.loadtxt(file,dtype="f8,f8,f8,f8",skiprows=skip_n,usecols=(0,1,spec_ind,flag_ind),unpack=True)
read = np.array(read)
times_since_start = read[0,:]
endtimes_since_start = read[1,:]
conc = read[2,:]
conc = np.array(conc).astype('float64')
flags = read[3,:]
dates = []
times = []
enddates = []
endtimes = []
times_since_start = np.float64(times_since_start)
endtimes_since_start = np.float64(endtimes_since_start)
for x in range(len(times_since_start)):
days_since_start = math.trunc(times_since_start[x])
enddays_since_start = math.trunc(endtimes_since_start[x])
remainder = times_since_start[x] - days_since_start
remainder_end = endtimes_since_start[x] - enddays_since_start
unrounded_hour = remainder*24
unrounded_hour_end = remainder_end*24
hour = np.round(unrounded_hour)
hour_end = np.round(unrounded_hour_end)
time_delta = datetime.timedelta(days = days_since_start,hours = hour)
time_delta_end = datetime.timedelta(days = enddays_since_start,hours = hour_end)
calc_datetime = start_datetime + time_delta
calc_datetime_end = start_datetime + time_delta_end
calc_yyyymmdd = calc_datetime.strftime("%Y%m%d")
calc_hhmm = calc_datetime.strftime("%H%M")
end_calc_yyyymmdd = calc_datetime_end.strftime("%Y%m%d")
end_calc_hhmm = calc_datetime_end.strftime("%H%M")
dates.append(calc_yyyymmdd)
times.append(calc_hhmm)
enddates.append(end_calc_yyyymmdd)
endtimes.append(end_calc_hhmm)
conc = np.float64(conc)
flags = np.float64(flags)
#add to n_obs_all
n_all += len(conc)
#IF bad_meta == True then set all file vals as nans
if bad_meta == True:
conc[:] = np.NaN
meta_valid_list.append(bad_meta)
#DO INLINE INVALID AND FLAG CONVERT to NaN
test = conc < 0
conc[test] = np.NaN
test = flags != 0
conc[test] = np.NaN
#convert units by line (only if value is >= than 0
try:
if (unit.lower() != 'ppb') & (unit.lower() != 'ppbv'):
if unit == 'ug/m3':
#calculate conversion factor from mg/m3 assuming 293K and 1013 hPa - in EU LAW
#R/MW*(TEMP0C(K)*TEMP(degC)/P(hPa)/10
conv_fact = 8.3144/mol_mass*(293.)/(1013./10.)
conc = conv_fact*conc
elif unit == 'ug_N/m3':
conv_fact = 8.3144/14.00674*(293.)/(1013./10.)
conc = conv_fact*conc
elif (unit == 'ppm') or (unit == 'ppmv'):
conc = conc*1e3
#print 'Converting Units from ppmv to ppbv'
elif (unit == 'ppt') or (unit == 'pptv'):
conc = conc/1e3
#print 'Converting Units from pptv to ppbv'
else:
print 'Unknown Unit'
1+'a'
except:
pass
#remove 9.999 from ISOP dataset
if species == 'ISOP':
test = conc == 9.999
conc[test] = np.NaN
#if file resolution is daily or monthly then replicate times after point, to fill hourly data array.
count=0
if file_res == '1h':
n_dups = np.zeros(len(conc))
elif file_res == '1d':
n_dups = []
#if measurement method is flask, then put leave flask measurement in as hourly measurement, the first hour of month
file_hours = len(dates)
for i in range(file_hours):
current_year = int(dates[count][:4])
current_month = int(dates[count][4:6])
current_day = int(dates[count][6:])
current_hh = int(times[count][:2])
current_mm = int(times[count][2:])
next_year = int(enddates[i][:4])
next_month = int(enddates[i][4:6])
next_day = int(enddates[i][6:])
next_hh = int(endtimes[i][:2])
next_mm = int(endtimes[i][2:])
s = datetime.datetime(year = current_year, month = current_month, day = current_day, hour = current_hh, minute = current_mm)
e = datetime.datetime(year = next_year, month = next_month, day = next_day, hour = next_hh, minute = next_mm)
day_dates = [d.strftime('%Y%m%d') for d in pd.date_range(s,e,freq='H')][1:-1]
day_hours = [d.strftime('%H%M') for d in pd.date_range(s,e,freq='H')][1:-1]
dates = np.insert(dates,count+1,day_dates)
times = np.insert(times,count+1,day_hours)
conc = np.insert(conc,count+1,[conc[count]]*len(day_dates))
#append to n duplicated array
n_dups=np.append(n_dups,0)
n_dups=np.append(n_dups,[1]*len(day_dates))
count +=(len(day_dates)+1)
elif file_res == '1mo':
n_dups = []
#if measurement method is flask, then put leave flask measurement in as hourly measurement, the first hour of month
file_hours = len(dates)
for i in range(file_hours):
current_year = int(dates[count][:4])
current_month = int(dates[count][4:6])
current_day = int(dates[count][6:])
current_hh = int(times[count][:2])
current_mm = int(times[count][2:])
next_year = int(enddates[i][:4])
next_month = int(enddates[i][4:6])
next_day = int(enddates[i][6:])
next_hh = int(endtimes[i][:2])
next_mm = int(endtimes[i][2:])
s = datetime.datetime(year = current_year, month = current_month, day = current_day, hour = current_hh, minute = current_mm)
e = datetime.datetime(year = next_year, month = next_month, day = next_day, hour = next_hh, minute = next_mm)
day_dates = [d.strftime('%Y%m%d') for d in pd.date_range(s,e,freq='H')][1:-1]
day_hours = [d.strftime('%H%M') for d in pd.date_range(s,e,freq='H')][1:-1]
dates = np.insert(dates,count+1,day_dates)
times = np.insert(times,count+1,day_hours)
conc = np.insert(conc,count+1,[conc[count]]*len(day_dates))
#append to n duplicated array
n_dups=np.append(n_dups,0)
n_dups=np.append(n_dups,[1]*len(day_dates))
count += (len(day_dates)+1)
data = [dates,times,conc,n_dups]
#put measurnement methods and into big list len of times
mm_big=np.append(mm_big,[mm]*len(dates))
try:
big_list = np.hstack((big_list,data))
except:
big_list = np.array(data)
if (y == year_array[-1]):
#get dates and times
date_con = big_list[0,:]
time_con = big_list[1,:]
#get vals
vals = np.array(big_list[2,:]).astype('float64')
#get n dup array
n_dup_array = np.array(big_list[3,:]).astype(float).astype(int)
#if all files have missing key meta then exit
if all(i == True for i in meta_valid_list) == True:
inv_nometa += 1
print 'Site Invalid. No Metadata for ref'
if no2_type == 'MOLYBDENUM':
n_all,inv_nometa,n_after_nometa,n_after_flagsandlod,n_after_duplicate,inv_anyvaliddata,n_obs_after_anyvaliddata,inv_nokeymeta,n_obs_after_nokeymeta,inv_resolution,n_obs_after_resolution,inv_badmeasurementmethod,n_obs_after_badmeasurementmethod = 0,0,0,0,0,0,0,0,0,0,0,0,0
exit_c_list = np.array([inv_nometa,inv_anyvaliddata,inv_nokeymeta,inv_resolution,inv_badmeasurementmethod])
n_c_list = np.array([n_all,n_after_nometa,n_after_flagsandlod,n_after_duplicate,n_after_anyvaliddata,n_after_nokeymeta,n_after_resolution,n_after_badmeasurementmethod])
unknown_list = [unknown_mm_list,unknown_mm_refs_list,unknown_local_tz_list]
meta = ['na','na','na','na','na','na','na','na','na','na','na','na']
exit_r = 'nometa'
return c,['na'],['na'],['na'],False,meta,exit_c_list,n_c_list,unknown_list,exit_r,np.zeros(1)
valid_hours_dup = np.sum(n_dup_array)
n_after_nometa += (len(vals)-valid_hours_dup)
#delete big list
del big_list
date_con = np.array(date_con).astype(int)
time_con = np.array(time_con).astype(int)
#remove data < 1970 and >= 2015
test_inds = (date_con >= 19700101) & (date_con < 20150101)
date_con = date_con[test_inds]
time_con = time_con[test_inds]
vals = vals[test_inds]
mm_big = mm_big[test_inds]
n_dup_array = n_dup_array[test_inds]
#set st_big as 'continuous'
st_big = ['continuous']*len(vals)
#convert all Nans back to -99999
test = np.isnan(vals)
vals[test] = -99999
#get obs valid
test = vals >= 0
valid_hours_dup = np.sum(n_dup_array[test])
n_obs_valid = int(len(vals[test]) - valid_hours_dup)
n_after_flagsandlod += n_obs_valid
#create max possible species grid, measurement method and sampling type grids
full_data = np.empty(n_hours)
full_data_after_flagsandlod = np.empty(n_hours)
big_n_dup_array = np.zeros(n_hours)
full_data[:] = -99999
full_data_after_flagsandlod[:] = -99999
#find matching times between actual times and grid of times, return big array of indices of matched indices in grid
converted_time = modules.date_process(date_con,time_con,start_year)
converted_time = np.round(converted_time,decimals=5)
syn_grid_time = np.arange(0,n_days,1./24)
syn_grid_time = np.round(syn_grid_time,decimals=5)
#find matching times between actual times and grid of times, return big array of indices of matched indices in grid
raw_indices = np.searchsorted(syn_grid_time, converted_time, side='left')
vals = np.array(vals)
full_data_after_flagsandlod[raw_indices] = vals
raw_st = np.copy(st_big)
raw_mm = np.copy(mm_big)
# test and remove duplicate and overlap points
converted_time,vals,mm_big,st_big,n_dup_array = modules.remove_duplicate_points(site_ref,converted_time,vals,mm_big,st_big,n_dup_array,output_res)
test = vals >= 0
valid_hours_dup = np.sum(n_dup_array[test])
n_obs_valid = int(len(vals[test]) - valid_hours_dup)
n_after_duplicate += n_obs_valid
#find matching times between actual times and grid of times, return big array of indices of matched indices in grid
indices = np.searchsorted(syn_grid_time, converted_time, side='left')
full_data[indices] = vals
big_n_dup_array[indices] = n_dup_array
#get mode of metadata
try:
lat = np.float32(stats.mode(all_latitudes)[0][0])
except:
lat = 'na'
try:
lon = np.float32(stats.mode(all_longitudes)[0][0])
except:
lon = 'na'
try:
alt = np.float32(stats.mode(all_altitudes)[0][0])
except:
alt = 'na'
unit = stats.mode(all_unit)[0][0]
#remove empty strings from extra meta before mode test
try:
site_name = stats.mode(filter(None, all_site_name))[0][0]
except:
site_name = 'na'
try:
country = stats.mode(filter(None, all_country))[0][0]
except:
country = 'na'
try:
contact = stats.mode(filter(None, all_contact))[0][0]
except:
contact = 'na'
#set data tz - all EMEP times are UTC
data_tz = 0
all_tz = [data_tz]
key_meta = [lat,lon,alt]
#convert file res to standard format
if file_res == '1h':
file_res = 'H'
elif file_res == '1d':
file_res = 'D'
elif file_res == '1mo':
file_res = 'M'
#get sampling/instrument grids
raw_st_grid,p_st_grid,p_st_grid_after_flagsandlod,raw_mm_grid,p_mm_grid,p_mm_grid_after_flagsandlod,unknown_mm_list,unknown_mm_refs_list = modules.sampling_and_instruments_by_processgroup(site_ref,process_group,species,raw_st,raw_mm,full_data_after_flagsandlod,full_data,raw_indices,unknown_mm_list,unknown_mm_refs_list,no2_type)
#do quality checks
data_valid,full_data,valid_hours_dup,p_st_grid,p_mm_grid,n_all,inv_nometa,n_after_nometa,n_after_flagsandlod,n_after_duplicate,inv_anyvaliddata,n_after_anyvaliddata,inv_nokeymeta,n_after_nokeymeta,inv_resolution,n_after_resolution,inv_badmeasurementmethod,n_after_badmeasurementmethod,exit_r = modules.primary_quality_control(site_ref,species,file_res,no2_type,grid_dates,full_data,big_n_dup_array,valid_hours_dup,raw_st_grid,p_st_grid,p_st_grid_after_flagsandlod,raw_mm_grid,p_mm_grid,p_mm_grid_after_flagsandlod,data_resolution,n_obs_valid,key_meta,n_all,inv_nometa,n_after_nometa,n_after_flagsandlod,n_after_duplicate,inv_anyvaliddata,n_after_anyvaliddata,inv_nokeymeta,n_after_nokeymeta,inv_resolution,n_after_resolution,inv_badmeasurementmethod,n_after_badmeasurementmethod)
if data_valid == False:
exit_c_list = np.array([inv_nometa,inv_anyvaliddata,inv_nokeymeta,inv_resolution,inv_badmeasurementmethod])
n_c_list = np.array([n_all,n_after_nometa,n_after_flagsandlod,n_after_duplicate,n_after_anyvaliddata,n_after_nokeymeta,n_after_resolution,n_after_badmeasurementmethod])
unknown_list = [unknown_mm_list,unknown_mm_refs_list,unknown_local_tz_list]
meta = [lat,lon,alt,'na','na','na','na','na','na','na','na','na']
return c,['na'],['na'],['na'],False,meta,exit_c_list,n_c_list,unknown_list,exit_r,np.zeros(1)
#set metadata not available as na
raw_class_name = 'na'
#set processed unit
p_unit = 'ppbv'
#get local timezone
try:
local_tz_name = tz_root.tzNameAt(lat,lon,forceTZ=True)
pytz_obj = pytz.timezone(local_tz_name)
datetime_offset = pytz_obj.utcoffset(datetime.datetime(2000,1,1))
if datetime_offset < datetime.timedelta(0):
local_tz = -(24-int(datetime_offset.seconds/60/60))
else:
local_tz = int(datetime_offset.seconds/60/60)
except:
local_tz = 'na'
print 'TIMEZONE NOT KNOWN, SITE IS %s'%(site_ref)
unknown_local_tz_list.append(site_ref)
#pack meta
meta = [lat,lon,alt,raw_class_name,file_res,unit,p_unit,data_tz,local_tz,site_name,country,contact]
#if blank strings in meta then convert to 'na'
for i in range(len(meta)):
try:
if meta[i].strip() == '':
meta[i] = 'na'
except:
pass
print set(raw_st_grid)
print set(raw_mm_grid)
print set(p_st_grid)
print set(p_mm_grid)
print meta
exit_c_list = np.array([inv_nometa,inv_anyvaliddata,inv_nokeymeta,inv_resolution,inv_badmeasurementmethod])
n_c_list = np.array([n_all,n_after_nometa,n_after_flagsandlod,n_after_duplicate,n_after_anyvaliddata,n_after_nokeymeta,n_after_resolution,n_after_badmeasurementmethod])
print 'exit counts = ', exit_c_list
print 'n obs counts = ', n_c_list
unknown_list = [unknown_mm_list,unknown_mm_refs_list,unknown_local_tz_list]
return c,full_data,p_st_grid,p_mm_grid,data_valid,meta,exit_c_list,n_c_list,unknown_list,'na',big_n_dup_array
def site_iter_process(valid_refs,c):
#for ref_i in range(len(valid_refs)):
data_valid = True
site_ref = valid_refs[c]
print 'Current Ref is = ', site_ref
s_files = glob.glob('/work/home/db876/observations/surface/%s/CAPMON/ozon_smpls_%s*'%(species,site_ref))
site_files = []
for y in year_array:
for f in s_files:
if str(y) in f:
site_files.append(f)
site_files = modules.natsorted(site_files)
yymmdd = []
hhmm = []
vals = []
#create max possible o3 grid
full_data = np.empty(n_hours)
full_data[:] = -99999
for file_i in range(len(site_files)):
count = 0
meta_start = -99999
start_read_1 = False
start_read_2 = False
with open(site_files[file_i], 'rb') as f:
reader = csv.reader(f,delimiter=',')
print site_files[file_i]
for row in reader:
#print count
#break out of loop at bottom of file
if (start_read_2 == True) & (row[0] == '*TABLE ENDS'):
break
#get metadata
try:
if (row[0] =='*TABLE NAME') & (row[1] == 'Site information'):
meta_start = count+2
except:
pass
if count == meta_start:
lat_i = row.index('Latitude: decimal degrees')
lon_i = row.index('Longitude: decimal degrees')
try:
alt_i = row.index('Ground elevation: above mean sea level')
except:
alt_i = row.index('Ground altitude')
class_i = row.index('Site land use')
if count == (meta_start+6):
latitude = row[lat_i]
longitude = row[lon_i]
altitude = row[alt_i]
raw_class_name = row[class_i]
#get data
if start_read_2 == True:
#read dates, times, and vals
date = row[8]
time = row[9]
yymmdd.append(date[:4]+date[5:7] + date[8:])
hhmm.append(time[:2]+time[3:])
quality_code = row[13]
if quality_code == 'V0':
vals = np.append(vals,np.float64(row[12]))
else:
vals = np.append(vals,-99999)
try:
if (row[0] == '*TABLE NAME') & (row[1] == 'OZONE_HOURLY'):
start_read_1 = True
except:
pass
if (start_read_1 == True) & (row[0] == '*TABLE COLUMN UNITS'):
unit = row[12]
if (start_read_1 == True) & (row[0] == '*TABLE BEGINS'):
start_read_2 = True
count+=1
#convert all invalids to -99999
test_inv = vals < 0
vals[test_inv] = -99999
#put o3 vals into full grid
date_con = np.array(yymmdd).astype(int)
time_con = np.array(hhmm).astype(int)
#find matching times between actual times and grid of times, return big array of indices of matched indices in grid
converted_time = modules.date_process(date_con,time_con,start_year)
converted_time = np.round(converted_time,decimals=5)
syn_grid_time = np.arange(0,n_days,1./24)
syn_grid_time = np.round(syn_grid_time,decimals=5)
#find matching times between actual times and grid of times, return big array of indices of matched indices in grid
indices = np.searchsorted(syn_grid_time, converted_time, side='left')
vals = np.array(vals)
#make sure no data is past end year
index_test = indices < len(full_data)
indices = indices[index_test]
vals = vals[index_test]
full_data[indices] = vals
#get metadata
lat = np.float64(latitude)
lon = np.float64(longitude)
alt = np.float64(altitude)
#do data quality checks
full_data,data_valid = modules.quality_check_nr(full_data,data_valid,data_resolution,np.float64(altitude),grid_dates,start_year,end_year)
#set measurement method
mm = 'ultraviolet photometry'
#set site file resolution
file_res = 'H'
#set sampling as average
st = 'average'
anthrome_class_name = 'na'
return c,full_data,data_valid,lat,lon,alt,raw_class_name,anthrome_class_name,mm,st,file_res
def site_iter_process(valid_refs,c):
#read files site at a time
#for ref_i in range(len(valid_refs)):
site_ref = valid_refs[c]
all_latitudes = []
all_longitudes = []
all_altitudes = []
all_mm = []
print 'Current Ref is = ', site_ref
#find if sites have full valid range from start year and finishing in end year
s_files = glob.glob('/work/home/db876/observations/surface/%s/EMEP/%s*'%(species,site_ref))
year_files = [file.replace("/work/home/db876/observations/surface/%s/EMEP/"%(species), "") for file in s_files]
cut_year_files = [file[8:12] for file in year_files]
site_files = []
for y in year_array:
for i in range(len(s_files)):
if str(y) in cut_year_files[i]:
site_files.append(s_files[i])
site_files = modules.natsorted(site_files)
year_files = modules.natsorted(year_files)
file_startdate = []
file_height = []
instr_names = []
file_lasttime = []
data_valid = True
yyyymmdd = []
hhmm = []
vals = []
flags = []
#create max possible o3 grid
full_data = np.empty(n_hours)
full_data[:] = -99999
if site_files == []:
print 'No valid files for site\n'
return
for y in year_array:
print 'Processing Year %s'%y
got_year = False
for file in site_files:
last_file_split = file.split('/')[-1]
if str(y) in last_file_split[8:12]:
got_year = True
break
if got_year == False:
#fill in data for missing year
timedelta_diff = datetime.date(y+1, 1, 1) - datetime.date(y, 1, 1)
ndays_missing = timedelta_diff.days
print 'ndays missing = ', ndays_missing
continue
if data_valid == True:
data_start = 9999999
count = 0
start_read = False
with open(file, 'rb') as f:
read_count = 0
reader = csv.reader(f,delimiter=' ')
print file
for row in reader:
try:
row = filter(lambda a: a != '', row)
except:
pass
try:
row = filter(lambda a: a != ',', row)
except:
pass
#get start date of file
if row[0] == 'Startdate:':
data = row[1]
s_yyyy = data[:4]
s_mm = data[4:6]
s_dd = data[6:8]
s_hh = data[8:10]
s_min = data[10:12]
start_datetime = datetime.datetime(int(s_yyyy),1,1,0,0)
#get unit
if row[0] == 'Unit:':
try:
unit_part1 = row[1]
unit_part2 = row[2]
unit = unit_part1+'_'+unit_part2
except:
unit = row[1]
#get resolution
if row[0] == 'Resolution':
if row[1] == 'code:':
file_res = row[2]
print 'Resolution = %s'%file_res
if (output_res == 'H'):
if (file_res == '1d') or (file_res == '1mo'):
print 'File resolution has to be Minimum Hourly. Skipping'
data_valid = False
return c,full_data,data_valid,-999,-999,-999,'na','na','na','na','na'
elif (output_res == 'D'):
if (file_res == '1mo'):
print 'File resolution has to be Minimum Daily. Skipping'
data_valid = False
return c,full_data,data_valid,-999,-999,-999,'na','na','na','na','na'
#get latitude
if row[0] == 'Station':
if row[1] == 'latitude:':
latitude = row[2]
all_latitudes.append(latitude)
#get longitude
if row[0] == 'Station':
if row[1] == 'longitude:':
longitude = row[2]
all_longitudes.append(longitude)
#get altitude
if row[0] == 'Station':
if row[1] == 'altitude:':
altitude = row[2][:-1]
all_altitudes.append(altitude)
#get period
if row[0] == 'Period':
period_code = row[2]
#get stats method
if row[0] == 'Statistics:':
try:
st = row[1] + row[2]
if st != 'arithmeticmean':
print 'Not Arithmetic Mean!'
print row[1]
print 1+'a'
except:
print 'Not Arithmetic Mean!'
print row[1]
print 1+'a'
#get instrument method
if row[0] == 'Instrument':
if row[1] == 'type:':
mm_list = row[2:]
if len(mm_list) > 1:
site_mm = ''
for x in range(len(mm_list)):
site_mm = site_mm+mm_list[x]+' '
site_mm = site_mm.strip()
else:
site_mm = mm_list[0]
all_mm.append(site_mm)
#get data
if start_read == True:
#calc dates, times, and take o3 vals
time_since_start = np.float64(row[0])
days_since_start = math.trunc(time_since_start)
remainder = time_since_start - days_since_start
unrounded_hour = remainder*24
hour = np.round(unrounded_hour)
time_delta = datetime.timedelta(days = days_since_start,hours = hour)
calc_datetime = start_datetime + time_delta
calc_yyyymmdd = calc_datetime.strftime("%Y%m%d")
calc_hhmm = calc_datetime.strftime("%H%M")
line_val = np.float64(row[2])
#convert units by line (only if value is >= than 0
if line_val >= 0:
if (unit.lower() != 'ppb') & (unit.lower() != 'ppbv'):
if unit == 'ug/m3':
#print 'converting units, temp = 20degC'
#calculate conversion factor from mg/m3 assuming 20 degC and 1 atm - default for O3 instruments
#R/MW*(TEMP0C(K)*TEMP(degC)/P(hPa)/10
conv_fact = 8.3144/mol_mass*(273.15+20)/(1013.25/10)
line_val = conv_fact*line_val
#print 'Converting Units from ug/m3 20degC to ppbv'
elif unit == 'ug_N/m3':
conv_fact = 8.3144/mol_mass*(273.15+20)/(1013.25/10)
line_val = conv_fact*line_val
#print 'Converting Units from ug/Nm3 20degC to ppbv'
elif (unit == 'ppm') or (unit == 'ppmv'):
line_val = line_val*1e3
#print 'Converting Units from ppmv to ppbv'
elif (unit == 'ppt') or (unit == 'pptv'):
line_val = line_val/1e3
#print 'Converting Units from pptv to ppbv'
else:
print 'Unknown Unit'
data_valid = False
1+'a'
if file_res == '1h':
yyyymmdd=np.append(yyyymmdd,calc_yyyymmdd)
hhmm=np.append(hhmm,calc_hhmm)
vals = np.append(vals,line_val)
flags = np.append(flags,np.float64(row[3]))
elif file_res == '1d':
yyyymmdd=np.append(yyyymmdd,calc_yyyymmdd)
hhmm=np.append(hhmm,'0000')
vals = np.append(vals,line_val)
flags = np.append(flags,np.float64(row[3]))
for j in range(1,24):
time_delta = datetime.timedelta(days = days_since_start,hours = j)
calc_datetime = start_datetime + time_delta
vals = np.append(vals,vals[-1])
flags = np.append(flags,flags[-1])
yyyymmdd = np.append(yyyymmdd,calc_datetime.strftime("%Y%m%d"))
hhmm = np.append(hhmm,calc_datetime.strftime("%H%M"))
elif file_res == '1mo':
yyyymmdd=np.append(yyyymmdd,calc_yyyymmdd)
hhmm=np.append(hhmm,'0000')
vals = np.append(vals,line_val)
flags = np.append(flags,np.float64(row[3]))
month_days = monthrange(int(yyyymmdd[-1][:4]), int(yyyymmdd[-1][4:6]))[1]
for j in range(1,24*month_days):
time_delta = datetime.timedelta(days = days_since_start,hours = j)
calc_datetime = start_datetime + time_delta
vals = np.append(vals,vals[-1])
flags = np.append(flags,flags[-1])
yyyymmdd = np.append(yyyymmdd,calc_datetime.strftime("%Y%m%d"))
hhmm = np.append(hhmm,calc_datetime.strftime("%H%M"))
if row[0] == 'starttime':
start_read = True
count+=1
if (y == year_array[-1]):
#convert all invalids by flags to -99999
test_inv = flags != 0
if len(test_inv) != 0:
vals[test_inv] = -99999
#any values less than zero are -99999
test_inv = vals < 0
if len(test_inv) != 0:
vals[test_inv] = -99999
#do additional invalid test, as flags not always correct
#test_inv_2 = vals > 300
#vals[test_inv_2] = -99999
#put o3 vals into full grid
date_con = np.array(yyyymmdd).astype(int)
time_con = np.array(hhmm).astype(int)
#find matching times between actual times and grid of times, return big array of indices of matched indices in grid
converted_time = date_process(date_con,time_con,start_year)
converted_time = np.round(converted_time,decimals=5)
syn_grid_time = np.arange(0,n_days,1./24)
syn_grid_time = np.round(syn_grid_time,decimals=5)
#find matching times between actual times and grid of times, return big array of indices of matched indices in grid
indices = np.searchsorted(syn_grid_time, converted_time, side='left')
vals = np.array(vals)
#make sure no data is past end year
index_test = indices < len(full_data)
indices = indices[index_test]
vals = vals[index_test]
full_data[indices] = vals
#get mode of metadata
lat = np.float64(stats.mode(all_latitudes)[0][0])
lon = np.float64(stats.mode(all_longitudes)[0][0])
alt = np.float64(stats.mode(all_altitudes)[0][0])
mm = stats.mode(all_mm)[0][0]
#check site is not urban using anthrome map from 2000
anthfile = '/work/home/db876/plotting_tools/core_tools/anthro2_a2000.nc'
anthload = Dataset(anthfile)
class_valid,anthrome_class_name = modules.anthrome_classify(anthload,[lat],[lon])
if class_valid == 'invalid':
data_valid = False
print 'Site Invalid, site classed as urban by anthrome map.'
#get measurement method
if (mm == 'uv_abs') or (mm == 'chemiluminesc') or (mm == 'uv_fluoresc'):
if species == 'O3':
mm = 'ultraviolet photometry'
if (species == 'NO') or (species == 'NO2') or (species == 'CO'):
mm = 'chemiluminescence'
elif (mm == 'ndir') or (mm == 'infrared_absorption'):
mm = 'non-dispersive infrared spectroscopy'
elif (mm == 'GC-HgO'):
mm = 'gas chromatography reduction gas detection'
elif (mm == 'tracegas_monitor'):
mm = 'cavity attenuated phase shift spectroscopy'
elif (mm == 'filter_1pack') or (mm == 'filter_2pack') or (mm == 'filter_3pack'):
if species == 'NO2':
mm = 'griess saltzman colorimetric'
elif species == 'CO':
mm = 'ion chromatography'
elif (mm == 'steel_canister'):
mm = 'gas chromatography flame ionisation detection'
elif (mm == 'online_gc'):
mm = 'online gas chromatography'
elif (mm == 'glass_sinter') or (mm == 'abs_solution') or (mm == 'filter_abs_solution') or (mm == 'abs_tube') or (mm == 'continuous_colorimetric'):
mm = 'griess saltzman colorimetric'
elif (mm == 'NaJ_solution'):
mm = 'flame ionisation detection'
elif (mm == 'doas'):
mm = 'differential optical absorption spectrosocopy'
elif (mm == 'diffusion_tube'):
mm = 'diffusive sampler'
elif (mm == 'NA') or (mm == ''):
if species == 'O3':
mm = 'ultraviolet photometry'
if species == 'CO':
mm = 'non-dispersive infrared spectroscopy'
if species == 'NO2':
mm = 'chemiluminescence'
if species == 'NO':
mm = 'chemiluminescence'
if species == 'ISOP':
mm = 'gas chromatography flame ionisation detection'
else:
print mm
1+'a'
#do data quality checks
full_data,data_valid = modules.quality_check(full_data,data_valid,data_resolution,alt,grid_dates,start_year,end_year)
#convert file res to standard format
if file_res == '1h':
file_res = 'H'
elif file_res == '1d':
file_res = 'D'
elif file_res == '1mo':
file_res = 'M'
#no raw class so set as na
raw_class_name = 'na'
#set sampling as average
st = 'average'
return c,full_data,data_valid,lat,lon,alt,raw_class_name,anthrome_class_name,mm,st,file_res
def site_iter_process(valid_refs, c):
#set local counts
inv_nometa = 0
inv_anyvaliddata = 0
inv_nokeymeta = 0
inv_resolution = 0
inv_badmeasurementmethod = 0
n_all = 0
n_after_nometa = 0
n_after_flagsandlod = 0
n_after_duplicate = 0
n_after_anyvaliddata = 0
n_after_nokeymeta = 0
n_after_resolution = 0
n_after_badmeasurementmethod = 0
#set local unknown lists
unknown_mm_list = []
unknown_mm_refs_list = []
unknown_local_tz_list = []
ref = valid_refs[c]
print 'ref = ', ref, c
#get site instrument for species
met_i = file_refs.index(ref)
file_name = met_refs[met_i]
site_name = met_sitenames[met_i]
print site_name
site_species = list(met_species[met_i])
print site_species
site_instruments = list(met_instruments[met_i])
m_method = site_instruments[site_species.index(species)]
site_resolutions = []
data_valid = True
s_files = insensitive_glob(
'/work/home/db876/observations/surface/%s/EANET/*%s.csv' %
(fname_species, file_name))
site_files = []
for y in year_array:
for f in s_files:
if str(y)[-2:] in f:
site_files.append(f)
site_files = modules.natsorted(site_files)
years = []
months = []
days = []
hours = []
vals = []
yyyymmdd = []
hhmm = []
n_dup_array = []
last_year_index = len(site_files)
for y in year_array:
got_year = False
for file in site_files:
last_file_split = file.split('/')[-1]
if str(y)[2:] in last_file_split:
got_year = True
break
if got_year == False:
timedelta_diff = datetime.date(y + 1, 1, 1) - datetime.date(
y, 1, 1)
ndays_missing = timedelta_diff.days
continue
print file
valid = True
with open(file, 'rb') as f:
reader = csv.reader(f, delimiter=',')
counter = 0
#get resolution
for row in reader:
if counter == 0:
all_units = row
elif counter == 1:
file_res = 'H'
try:
hour_index = row.index('Hour')
except:
file_res = 'D'
try:
day_index = row.index('Day')
except:
file_res = 'M'
month_index = row.index('Month')
year_index = row.index('Year')
try:
spec_index = row.index(species.upper())
unit = all_units[spec_index]
except:
valid = False
break
#make sure each year units are ppb
if unit != 'ppb':
print 'Units not ppb!'
1 + 'a'
if counter == 2:
if file_res == 'H':
yyyy = row[year_index]
mm = row[month_index]
dd = row[day_index]
hh = row[hour_index]
elif file_res == 'D':
yyyy = row[year_index]
mm = row[month_index]
dd = row[day_index]
hh = 1
elif file_res == 'M':
yyyy = row[year_index]
mm = row[month_index]
dd = 1
hh = 1
start_datetime = datetime.datetime(int(yyyy), int(mm),
int(dd), int(hh))
if counter == 3:
if file_res == 'H':
yyyy = row[year_index]
mm = row[month_index]
dd = row[day_index]
hh = row[hour_index]
elif file_res == 'D':
yyyy = row[year_index]
mm = row[month_index]
dd = row[day_index]
hh = 1
elif file_res == 'M':
yyyy = row[year_index]
mm = row[month_index]
dd = 1
hh = 1
present_datetime = datetime.datetime(
int(yyyy), int(mm), int(dd), int(hh))
time_delt = present_datetime - start_datetime
hour_delt = datetime.timedelta(hours=1)
day_delt = datetime.timedelta(hours=24)
week_delt = datetime.timedelta(hours=24 * 7)
month_delt = datetime.timedelta(hours=24 * 28)
print time_delt
if (time_delt < day_delt):
print 'Hourly Data'
file_res = 'H'
site_resolutions.append(file_res)
elif (time_delt > hour_delt) & (time_delt < week_delt):
print 'Daily Data'
file_res = 'D'
site_resolutions.append(file_res)
elif (time_delt > week_delt):
print 'Monthly Data'
file_res = 'M'
site_resolutions.append(file_res)
counter += 1
#READ IN DATA
if valid == True:
#limit to sites with hourly date files for, if required
if output_res == 'H':
if file_res != 'H':
print 'Not processing as only want hourly files'
continue
if output_res == 'HD':
if file_res == 'M':
print 'Not processing as only want hourly and daily files'
continue
with open(file, 'rb') as f:
reader = csv.reader(f, delimiter=',')
counter = 0
val_count = 0
for row in reader:
if counter >= 2:
yyyy = row[year_index]
mm = row[month_index]
#add to n_obs_all
n_all += 1
n_after_nometa += 1
if file_res == 'H':
try:
vals = np.append(vals,
np.float64(row[spec_index]))
except:
vals = np.append(vals, -99999)
current_datetime = present_datetime + relativedelta(
hours=val_count)
yyyymmdd.append(
current_datetime.strftime("%Y%m%d"))
hhmm.append(current_datetime.strftime("%H%M"))
n_dup_array = np.append(n_dup_array, 0)
elif file_res == 'D':
try:
vals = np.append(
vals, [np.float64(row[spec_index])] * 24)
except:
vals = np.append(vals, [-99999] * 24)
current_datetime = present_datetime + relativedelta(
days=val_count)
next_datetime = present_datetime + relativedelta(
days=val_count + 1)
all_datetimes = pd.date_range(current_datetime,
next_datetime,
freq='H')[:-1]
for d in all_datetimes:
yyyymmdd.append(d.strftime("%Y%m%d"))
hhmm.append(d.strftime("%H%M"))
n_dup_array = np.append(n_dup_array, 0)
n_dup_array = np.append(n_dup_array, [1] * 23)
elif file_res == 'M':
month_days = monthrange(int(yyyy), int(mm))[1]
try:
vals = np.append(
vals, [np.float64(row[spec_index])] *
(month_days * 24))
except:
vals = np.append(vals,
[-99999] * (month_days * 24))
current_datetime = present_datetime + relativedelta(
months=int(mm) - 1)
next_datetime = present_datetime + relativedelta(
months=int(mm))
all_datetimes = pd.date_range(current_datetime,
next_datetime,
freq='H')[:-1]
for d in all_datetimes:
yyyymmdd.append(d.strftime("%Y%m%d"))
hhmm.append(d.strftime("%H%M"))
n_dup_array = np.append(n_dup_array, 0)
n_dup_array = np.append(n_dup_array, [1] *
((month_days * 24) - 1))
val_count += 1
counter += 1
else:
print 'Species is not in file header. Skipping Year'
timedelta_diff = datetime.date(y + 1, 1, 1) - datetime.date(
y, 1, 1)
ndays_missing = timedelta_diff.days
print 'ndays missing = ', ndays_missing
#test if have no data due to not required time resolution, if so exit
if len(vals) == 0:
n_all, inv_nometa, n_after_nometa, n_after_flagsandlod, n_after_duplicate, inv_anyvaliddata, n_obs_after_anyvaliddata, inv_nokeymeta, n_obs_after_nokeymeta, inv_resolution, n_obs_after_resolution, inv_badmeasurementmethod, n_obs_after_badmeasurementmethod = 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
exit_c_list = np.array([
inv_nometa, inv_anyvaliddata, inv_nokeymeta, inv_resolution,
inv_badmeasurementmethod
])
n_c_list = np.array([
n_all, n_after_nometa, n_after_flagsandlod, n_after_duplicate,
n_after_anyvaliddata, n_after_nokeymeta, n_after_resolution,
n_after_badmeasurementmethod
])
unknown_list = [
unknown_mm_list, unknown_mm_refs_list, unknown_local_tz_list
]
meta = [
'na', 'na', 'na', 'na', 'na', 'na', 'na', 'na', 'na', 'na', 'na',
'na'
]
return c, ['na'], ['na'], [
'na'
], False, meta, exit_c_list, n_c_list, unknown_list, 'nothourly', np.zeros(
0)
#create max possible grid
full_data = np.empty(n_hours)
full_data_after_flagsandlod = np.empty(n_hours)
big_n_dup_array = np.zeros(n_hours)
full_data[:] = -99999
full_data_after_flagsandlod[:] = -99999
#convert blank values to -99999
test_inv = vals == ''
vals[test_inv] = -99999
#convert number invalids to -99999
test_inv = vals < 0
vals[test_inv] = -99999
#if all site resolutions are same continue then take first file_res
all_same = all(x == site_resolutions[0] for x in site_resolutions)
if all_same == True:
file_res = site_resolutions[0]
else:
#otherwise take lowest frequency res as file_res
if 'M' in site_resolutions:
file_res = 'M'
elif 'D' in site_resolutions:
file_res = 'D'
else:
file_res = 'H'
#get meta
i_ref = file_refs.index(ref)
site_ref = ref
data_tz = np.float32(met_tz[i_ref])
all_tz = [data_tz]
lat = np.float32(met_lats[i_ref])
lon = np.float32(met_lons[i_ref])
alt = np.float32(met_alts[i_ref])
raw_class_name = met_class[i_ref]
country = met_country[i_ref]
unit = str(unit)
contact = 'Ayako Aoyagi, Asia Center for Air Pollution Research, [email protected]'
#adjust dates and times if tz is not equal to 0
tz = int(data_tz)
if tz != 0:
for i in range(len(yyyymmdd)):
#create datetime
dt = datetime.datetime(int(yyyymmdd[i][:4]), int(yyyymmdd[i][4:6]),
int(yyyymmdd[i][6:]), int(hhmm[i][:2]),
int(hhmm[i][2:]))
if tz > 0:
dt = dt - datetime.timedelta(hours=int(tz))
elif tz < 0:
dt = dt + datetime.timedelta(hours=np.abs(int(tz)))
yyyymmdd[i] = dt.strftime("%Y%m%d")
hhmm[i] = dt.strftime("%H%M")
#put vals into full grid
date_con = np.array(yyyymmdd).astype(int)
time_con = np.array(hhmm).astype(int)
#remove data < 1970 and >= 2015
test_inds = (date_con >= 19700101) & (date_con < 20150101)
date_con = date_con[test_inds]
time_con = time_con[test_inds]
vals = vals[test_inds]
n_dup_array = n_dup_array[test_inds]
#set st_big and mm_big
st_big = ['continuous'] * len(vals)
mm_big = [m_method] * len(vals)
#get obs valid
test = vals >= 0
valid_hours_dup = np.sum(n_dup_array[test])
n_obs_valid = int(len(vals[test]) - valid_hours_dup)
n_after_flagsandlod += n_obs_valid
#find matching times between actual times and grid of times, return big array of indices of matched indices in grid
converted_time = modules.date_process(date_con, time_con, start_year)
converted_time = np.round(converted_time, decimals=5)
syn_grid_time = np.arange(0, n_days, 1. / 24)
syn_grid_time = np.round(syn_grid_time, decimals=5)
raw_indices = np.searchsorted(syn_grid_time, converted_time, side='left')
vals = np.array(vals)
full_data_after_flagsandlod[raw_indices] = vals
raw_st = np.copy(st_big)
raw_mm = np.copy(mm_big)
# test and remove duplicate and overlap points
converted_time, vals, mm_big, st_big, n_dup_array = modules.remove_duplicate_points(
site_ref, converted_time, vals, mm_big, st_big, n_dup_array,
output_res)
test = vals >= 0
valid_hours_dup = np.sum(n_dup_array[test])
n_obs_valid = int(len(vals[test]) - valid_hours_dup)
n_after_duplicate += n_obs_valid
#find matching times between actual times and grid of times, return big array of indices of matched indices in grid
indices = np.searchsorted(syn_grid_time, converted_time, side='left')
full_data[indices] = vals
big_n_dup_array[indices] = n_dup_array
key_meta = [lat, lon, alt]
#get sampling/instrument grids
raw_st_grid, p_st_grid, p_st_grid_after_flagsandlod, raw_mm_grid, p_mm_grid, p_mm_grid_after_flagsandlod, unknown_mm_list, unknown_mm_refs_list = modules.sampling_and_instruments_by_processgroup(
site_ref, process_group, species, raw_st, raw_mm,
full_data_after_flagsandlod, full_data, raw_indices, unknown_mm_list,
unknown_mm_refs_list, no2_type)
#do quality checks
data_valid, full_data, valid_hours_dup, p_st_grid, p_mm_grid, n_all, inv_nometa, n_after_nometa, n_after_flagsandlod, n_after_duplicate, inv_anyvaliddata, n_after_anyvaliddata, inv_nokeymeta, n_after_nokeymeta, inv_resolution, n_after_resolution, inv_badmeasurementmethod, n_after_badmeasurementmethod, exit_r = modules.primary_quality_control(
site_ref, species, file_res, no2_type, grid_dates, full_data,
big_n_dup_array, valid_hours_dup, raw_st_grid, p_st_grid,
p_st_grid_after_flagsandlod, raw_mm_grid, p_mm_grid,
p_mm_grid_after_flagsandlod, data_resolution, n_obs_valid, key_meta,
n_all, inv_nometa, n_after_nometa, n_after_flagsandlod,
n_after_duplicate, inv_anyvaliddata, n_after_anyvaliddata,
inv_nokeymeta, n_after_nokeymeta, inv_resolution, n_after_resolution,
inv_badmeasurementmethod, n_after_badmeasurementmethod)
if data_valid == False:
exit_c_list = np.array([
inv_nometa, inv_anyvaliddata, inv_nokeymeta, inv_resolution,
inv_badmeasurementmethod
])
n_c_list = np.array([
n_all, n_after_nometa, n_after_flagsandlod, n_after_duplicate,
n_after_anyvaliddata, n_after_nokeymeta, n_after_resolution,
n_after_badmeasurementmethod
])
unknown_list = [
unknown_mm_list, unknown_mm_refs_list, unknown_local_tz_list
]
meta = [
lat, lon, alt, 'na', 'na', 'na', 'na', 'na', 'na', 'na', 'na', 'na'
]
return c, ['na'], ['na'], [
'na'
], False, meta, exit_c_list, n_c_list, unknown_list, exit_r, np.zeros(
1)
#make tz int after checks
data_tz = np.float32(data_tz)
#set processed unit
p_unit = 'pbbv'
#get local timezone
try:
local_tz_name = tz_root.tzNameAt(lat, lon, forceTZ=True)
pytz_obj = pytz.timezone(local_tz_name)
datetime_offset = pytz_obj.utcoffset(datetime.datetime(2000, 1, 1))
if datetime_offset < datetime.timedelta(0):
local_tz = -(24 - int(datetime_offset.seconds / 60 / 60))
else:
local_tz = int(datetime_offset.seconds / 60 / 60)
except:
local_tz = 'na'
print 'TIMEZONE NOT KNOWN, SITE IS %s' % (site_ref)
unknown_local_tz_list.append(site_ref)
#pack meta
meta = [
lat, lon, alt, raw_class_name, file_res, unit, p_unit, data_tz,
local_tz, site_name, country, contact
]
#if blank strings in meta then convert to 'na'
for i in range(len(meta)):
try:
if meta[i].strip() == '':
meta[i] = 'na'
except:
pass
print set(raw_st_grid)
print set(raw_mm_grid)
print set(p_st_grid)
print set(p_mm_grid)
print meta
exit_c_list = np.array([
inv_nometa, inv_anyvaliddata, inv_nokeymeta, inv_resolution,
inv_badmeasurementmethod
])
n_c_list = np.array([
n_all, n_after_nometa, n_after_flagsandlod, n_after_duplicate,
n_after_anyvaliddata, n_after_nokeymeta, n_after_resolution,
n_after_badmeasurementmethod
])
print 'exit counts = ', exit_c_list
print 'n obs counts = ', n_c_list
unknown_list = [
unknown_mm_list, unknown_mm_refs_list, unknown_local_tz_list
]
return c, full_data, p_st_grid, p_mm_grid, data_valid, meta, exit_c_list, n_c_list, unknown_list, 'na', big_n_dup_array
