def setup_station():
df_hourly = pd.read_csv('examples/testdata.csv.gz',
index_col=0,
parse_dates=True)
df_hourly = df_hourly.loc['2016-01-01':'2016-12-31']
df_hourly.temp += 273.15
df_daily = melodist.util.daily_from_hourly(df_hourly)
station = melodist.Station(lon=8.86,
lat=51.00,
timezone=1,
data_daily=df_daily)
station.statistics = melodist.StationStatistics(data=df_hourly)
return station
data_obs_hourly = melodist.util.drop_incomplete_days(data_obs_hourly) # optional: select one year (a subselection can be processed much faster) data_obs_hourly = data_obs_hourly.loc['2009-01-01':'2009-12-31'] # for testing purposes in this file: create daily out of hourly values for # disaggregation data_obs_daily = melodist.util.daily_from_hourly(data_obs_hourly) #%% # Create station object longitude = 5.18 latitude = 52.10 timezone = 1 debilt = melodist.Station(lon=longitude, lat=latitude, timezone=timezone) debilt.data_daily = data_obs_daily # initialize station statistics object debilt.statistics = melodist.StationStatistics(data_obs_hourly) # Step 2: Compute statistics for various methods stats = debilt.statistics stats.calc_wind_stats() stats.calc_humidity_stats() stats.calc_temperature_stats() # the computations of statistics is similar: stats.calc_precipitation_stats() # as an alternative you can also define 2 seasons for precipitation statistics
longitude = 8.86
latitude = 51.00
timezone = 1
daily_data=pd.read_csv(path_inp_daily, index_col=0, parse_dates=True)
calibration_period = slice('2003-01-01', '2017-12-31')
validation_period = slice('2018-01-01', '2018-12-31')
plot_period = slice('2016-07-03', '2016-08-14')
data_obs_hourly = pd.read_csv(path_inp, index_col=0, parse_dates=True)
data_obs_hourly = melodist.util.drop_incomplete_days(data_obs_hourly)
station = melodist.Station(lon=longitude, lat=latitude,
timezone=timezone,
data_daily=daily_data)
station.statistics = melodist.StationStatistics(data_obs_hourly.loc[calibration_period])
stats = station.statistics
stats.calc_precipitation_stats()
precipdf = pd.DataFrame()
for method in ('equal', 'cascade'):
station.disaggregate_precipitation(method=method)
precipdf[method] = station.data_disagg.precip
precipdf['cascade'].to_csv (r'C:\Users\rober\OneDrive\Internship\SWAT model\inputdataNOTDELETE\climatedata\robert.keppler7jAESi\output_cascade_hourly.csv', index=False)
def main(raw_dir, mylon, mylat, tz, nc_standard_clim, nc_standard_hist,
cal_period, val_period, plot_period, path_inpt,
outdir): # create 1h wfj era5 obs data by
#===============================================================================
#
# MODULE: Subdaily dissagregaration
#
#===============================================================================
#===============================================================================
# Settings
#===============================================================================
# Do we want to run all methods and plot to look at performance?
test_methods = False
geotop_out = False
fsm_out = True
# standard calenders to interp non-standard cals to in hist and clim period
# nc_standard_clim='/home/joel/sim/qmap/test/pyout/aresult/ICHEC-EC-EARTH_rcp85_r12i1p1_CLMcom-CCLM5-0-6_v1__TS.nc_TS_ALL_ll.nc'
# nc_standard_hist='/home/joel/sim/qmap/test/pyout/aresult/ICHEC-EC-EARTH_historical_r1i1p1_KNMI-RACMO22E_v1__TS.nc_TS_ALL_ll.nc'
# # station attributes
# longitude = 9.80988
# latitude = 46.82786
# timezone = 1
# calibration_period = slice('2000-01-01', '2015-12-31')
# validation_period = slice('2016-01-01', '2016-12-31')
# plot_period = slice('2016-09-03', '2030-10-13')
# path_inpt = "/home/joel/sim/qmap/wfj_long.csv"
# station attributes
longitude = mylon
latitude = mylat
timezone = tz
calibration_period = cal_period
validation_period = val_period
plot_period = plot_period
path_inpt = path_inpt
#===============================================================================
# model downloads accounting
#===============================================================================
nchistfiles = glob.glob(raw_dir + "*historical*")
ncrcp26files = glob.glob(raw_dir + "*rcp26*")
ncrcp85files = glob.glob(raw_dir + "*rcp85*")
# file names
ncfiles_vec = nchistfiles, ncrcp26files, ncrcp85files
for ncfiles in ncfiles_vec:
allfiles = [i.split('raw_cordex/', 1)[1]
for i in ncfiles] # c. 9000 models, pars, times
rootfiles = [i.split('day', 1)[0] for i in allfiles]
modelPars = list(set(rootfiles)) # c.5000 models, par
models2 = list(set([i.split('_', 1)[1]
for i in rootfiles])) # c. 60 models
print(len(models2))
#26
#10
#25
#total = 61 models downloaded
# 44 now available complete eg len(nc_complete) not all models have all parameters
#===============================================================================
# Observations (toposacle output)
# Units:
# DW degrees
# ILWR: wm-2
# ISWR: wm-2
# PINT: mm/hr
# PSUM: mm/timestep
# Rf: mm/hr
# Sf: mm/hr
# TA: K
# VW: m s-1
# RH: 0-1
# P: Pa
#===============================================================================
# lesson of psum mess is dont use psum!!
df_obs = pd.read_csv(path_inpt, index_col=0, parse_dates=True)
# coorect for toposcale error where PINT = PSUM at 3hr timestep. now it is mm/h 19/8/20
#psum is actually wrong! 20/8/20 eg this is too little:
# correction!!!!!!!!
# df_obs.PSUM *= 3.
# # resample 3h OBS to 1h data (psum no longer valid) - can use 1h data directly here in future
#df_interpol = df_obs.resample('H').mean()
#df_interpol = df_interpol.interpolate()
#df_interpol.head(4)
#df_interpol.PSUM *= 1/3. # devide by 3 as is psum over 3h
# now interpolation done in preprocessing
df_interpol = df_obs
data_obs_hourly = df_interpol.loc[:, ('TA', 'PINT', 'RH', 'ISWR', 'ILWR',
'VW')]
data_obs_hourly.rename(columns={
'TA': 'temp',
'PINT': 'precip',
'RH': 'hum',
'ISWR': 'glob',
'ILWR': 'ilwr',
'VW': 'wind'
},
inplace=True)
# unit conversions
data_obs_hourly.precip *= 1. #(1/3600.)#mm/hr to to kgm-2s-1
data_obs_hourly.hum *= 100. # 0-1 to 0-100
#data_obs_hourly = df2
#data_obs_hourly = melodist.util.drop_incomplete_days(data_obs_hourly)
"""Aggregates data (hourly to daily values) according to the characteristics
of each variable (e.g., average for temperature, sum for precipitation)"""
data_obs_daily = melodist.util.daily_from_hourly(data_obs_hourly) #
#======================================================================================
# Methods from melodist
#======================================================================================
def plot(obs, sim):
plt.figure()
ax = plt.gca()
obs.loc[plot_period].plot(ax=ax, color='black', label='obs', lw=2)
sim.loc[plot_period].plot(ax=ax)
plt.legend()
plt.show()
def calc_stats(obs, sim):
df = pd.DataFrame(columns=['mean', 'std', 'r', 'rmse', 'nse'])
obs = obs.loc[validation_period]
sim = sim.loc[validation_period]
df.loc['obs'] = obs.mean(), obs.std(), 1, 0, 1
for c in sim.columns:
osdf = pd.DataFrame(data=dict(obs=obs, sim=sim[c])).dropna(
how='any')
o = osdf.obs
s = osdf.sim
r = scipy.stats.pearsonr(o, s)[0]
rmse = np.mean((o - s)**2)
nse = 1 - np.sum((o - s)**2) / np.sum((o - o.mean())**2)
df.loc[c] = s.mean(), s.std(), r, rmse, nse
return df
def print_stats(obs, sim):
df = calc_stats(obs, sim)
html = df.round(2).style
return html
#======================================================================================
# Climate data to dissagg
# 'tas' K
# 'tasmin' K
# 'tasmax' K
# 'pr' kg m-2 s-1
# 'hurs' % 0-100
# 'rsds' W m-2
# 'rlds' W m-2
# 'uas' m s-1
# 'vas' m s-1
# 'ps' pa
#===============================================================================
# check for complete nc files
ncfiles = glob.glob(raw_dir + '/aresult/' + '*_TS_ALL_ll.nc')
nc_complete = []
for nc in ncfiles:
ds = xr.open_dataset(nc, decode_times=True)
# this bit depend on the model and calender i think
#either this:
datetimeindex = ds.indexes["time"] #.to_datetimeindex()
datetimeindex.is_all_dates
#or this
#datetimeindex = ds.indexes["time"].to_datetimeindex()
#datetimeindex.is_all_dates
#ds['dtime'] = datetimeindex
# myx =np.argmin(abs(ds['lon']-longitude))
# myy = np.argmin(abs(ds['lat']-latitude))
df = ds.sel(lon=longitude, lat=latitude,
method='nearest').to_dataframe()
if df.shape[1] < 14:
#print "not enouigh vars"
next
try:
#df = ds.to_dataframe()
df2 = df[[
'tas', 'tasmin', 'tasmax', 'pr', 'hurs', 'rsds', 'rlds', 'ps'
]] #, 'vas']]
#print 'ok'
print(nc)
print(df.columns)
nc_complete.append(nc)
#break
except:
#print "no vars"
#print df.columns
next
for nc in nc_complete:
print(nc)
#nc=nc_complete[9] no leapS
#nc=nc_complete[10] # 360 day
# sthis file has a standard calender and used as target in interpolation (hist or clim periods)
if 'historical' in nc:
ds1 = xr.open_dataset(nc_standard_hist, decode_times=True)
if 'rcp' in nc:
ds1 = xr.open_dataset(nc_standard_clim, decode_times=True)
ds = xr.open_dataset(nc, decode_times=True)
datetimeindex = ds.indexes["time"]
if datetimeindex.is_all_dates is False:
dateob = datetimeindex[0]
cal = dateob.calendar
print(cal)
# print(dateob.datetime_compatible)
#ds_conv = convert_calendar(ds, ds1,'year', 'time')
ds_interp = cal3.interp_calendar(ds, ds1, 'time')
# now we have to fill na in first and last timestep which results from interpolation
vars = list(ds_interp.data_vars.keys())
for var in vars:
# backfill step 0
ds_interp[var][0, :, :] = ds_interp[var][1, :, :]
# foward fill step -1
ds_interp[var][-1, :, :] = ds_interp[var][-2, :, :]
# this just checks it worked
datetimeindex = ds_interp.indexes["time"]
if datetimeindex.is_all_dates is True:
#dateob = datetimeindex[0]
#cal = dateob.calendar
print("Succesfully converted " + cal + "to standard calender")
#print(dateob.datetime_compatible)
# rename object for next step
ds = ds_interp
else:
"Something went wrong cal3 conversionfailed"
else:
print("Standard calender exists")
# this bit depend on the model and calender i think
# datetimeindex = ds.indexes["time"]
# if datetimeindex.is_all_dates is False:
# dateob = datetimeindex[0]
# cal = dateob.calendar
# print(cal)
# print(dateob.datetime_compatible)
# if datetimeindex.is_all_dates is False: # this means we have 365 or 360 day calender
# <p> # somehow test for cal typ here
# ds.variables['time']
# # 'noleapcal' strategy is to just interpolate leap days 29 Feb
# if cal == 'noleap':
# print 'yes'
# start = ds.variables['time'][0].values
# end = ds.variables['time'][-1].values
# # convert cftime index to datetime object
# # https://stackoverflow.com/questions/55786995/converting-cftime-datetimejulian-to-datetime
# ds_ccaldatetime = ds.indexes['time'].to_datetimeindex()
# ds_realdatetime =pd.date_range(ds_ccaldatetime[0],ds_ccaldatetime[-1], freq='D')
# ds.reindex(ds_realdatetime, fill_value=0)
# # here cal is 30 days in each year
# if cal == '360_day':
# print 'yes'
# print 'yes'
# start = ds.variables['time'][0].values
# end = ds.variables['time'][-1].values
# # convert cftime index to datetime object
# # https://stackoverflow.com/questions/55786995/converting-cftime-datetimejulian-to-datetime
# ds_ccaldatetime = ds.indexes['time'].to_datetimeindex()
# ds_realdatetime =pd.date_range(ds_ccaldatetime[0],ds_ccaldatetime[-1], freq='D')
# ds.reindex(ds_realdatetime, fill_value=0)
# if datetimeindex.is_all_dates is False:
# try:
# #http://xarray.pydata.org/en/stable/weather-climate.html#non-standard-calendars-and-dates-outside-the-timestamp-valid-range
# datetimeindex = ds.indexes["time"].to_datetimeindex()
# except:
# print("Non standard time index 360 day cal, cant create datetime index, skipping")
# continue # cant handle 360day claender
df = ds.sel(lon=longitude, lat=latitude,
method='nearest').to_dataframe()
df2 = df.loc[:, ('tas', 'tasmin', 'tasmax', 'pr', 'hurs', 'rsds',
'rlds', 'ps')]
#compute wind speed and wind direction
df2.loc[:, 'ws'] = np.sqrt(df.uas**2 + df.vas**2)
df2.loc[:,
'wd'] = (180 / np.pi) * np.arctan(df.uas / df.vas) + np.where(
df.vas > 0, 180, np.where(df.uas > 0, 360, 0))
# unit conversions
df2.loc[:, 'pr'] *= 3600 * 24 # kgm-2s-1 (same as mm /s)-> mm/day
df2.rename(columns={
'tas': 'temp',
'tasmax': 'tmax',
'tasmin': 'tmin',
'pr': 'precip',
'hurs': 'hum',
'rsds': 'glob',
'ws': 'wind'
},
inplace=True)
#df2["datetime"] = datetimeindex
# set index
df3 = df2.set_index(datetimeindex)
df4 = df3.resample('D').mean()
# filter tmin /tmax error in #'/home/joel/sim/qmap/test/pyout/aresult/MOHC-HadGEM2-ES_historical_r1i1p1_CLMcom-CCLM5-0-6_v1__TS.nc_TS_ALL_ll.nc'
# vals of 500 / 20
df4.tmin[df4.tmin > 330] = np.nan
df4.tmax[df4.tmax < 220] = np.nan
# for case of 360 calender that ends on dec 26 (rare) for example:
#'/home/joel/sim/qmap/test/pyout/aresult/MOHC-HadGEM2-ES_historical_r1i1p1_CLMcom-CCLM5-0-6_v1__TS.nc_TS_ALL_ll.nc'
# we run ffill() for last 5 days
# has no effect if contains data
df4 = df4.ffill()
station = melodist.Station(lon=longitude,
lat=latitude,
timezone=timezone,
data_daily=df4)
# aggregated hourly obs with min and max for calculating stats
#station2 = melodist.Station(lon=longitude, lat=latitude, timezone=timezone, data_daily=data_obs_daily)
station.statistics = melodist.StationStatistics(
data_obs_hourly.loc[calibration_period])
stats = station.statistics
stats.calc_wind_stats()
stats.calc_humidity_stats()
stats.calc_temperature_stats()
stats.calc_radiation_stats()
#stats.calc_precipitation_stats()
if test_methods is True:
tempdf = pd.DataFrame()
for method in ('sine_min_max', 'sine_mean', 'mean_course_min_max',
'mean_course_mean'):
station.disaggregate_temperature(method=method,
min_max_time='sun_loc_shift')
tempdf[method] = station.data_disagg.temp
plot(data_obs_hourly.temp, tempdf)
print_stats(data_obs_hourly.temp, tempdf)
# ok, let's say we have decided to use sine_min_max. next we want to disaggregate humidity.
# as some of hum disagg functions rely on disagg'd temperature values we disagg temp again
# with our chosen method
station2.disaggregate_temperature(method='mean_course_mean',
min_max_time='sun_loc_shift')
humdf = pd.DataFrame()
for method in (
'equal',
'minimal',
'dewpoint_regression',
'linear_dewpoint_variation',
#'min_max', not poss as no min max in obs
'month_hour_precip_mean'):
station.disaggregate_humidity(method=method)
humdf[method] = station.data_disagg.hum
plot(data_obs_hourly.hum, humdf)
print_stats(data_obs_hourly.hum, humdf)
globdf = pd.DataFrame()
for method in (
'pot_rad',
# 'pot_rad_via_ssd', # not possible here as we do not have sunshine duration data
'pot_rad_via_bc',
'mean_course'):
station.disaggregate_radiation(method=method)
globdf[method] = station.data_disagg.glob
plot(data_obs_hourly.glob, globdf)
print_stats(data_obs_hourly.glob, globdf)
winddf = pd.DataFrame()
for method in ('equal', 'cosine', 'random'):
station.disaggregate_wind(method=method)
winddf[method] = station.data_disagg.wind
plot(data_obs_hourly.wind, winddf)
print_stats(data_obs_hourly.wind, winddf)
precipdf = pd.DataFrame()
for method in ('equal'):
station.disaggregate_precipitation(method=method)
precipdf[method] = station.data_disagg.precip
plot(data_obs_hourly.precip, precipdf)
#final dissagregation with selected methods
station.disaggregate_temperature(method='sine_mean')
station.disaggregate_humidity(method='equal')
station.disaggregate_radiation(method='pot_rad_via_bc')
station.disaggregate_wind(method='random')
station.disaggregate_precipitation(method='equal')
# make wind direction equal each hour
station.data_disagg['DW'] = df4.wd.resample('H').mean().interpolate(
method='pad')
# FIll last values that go to NA in interp
# foward fill step -1
station.data_disagg['DW'][-1] = station.data_disagg['DW'][-2]
# make air pressure equal each hour
station.data_disagg['P'] = df4.ps.resample('H').mean().interpolate(
method='pad')
station.data_disagg['P'][-1] = station.data_disagg['P'][-2]
# disagg lwin with T
# compute a daily scaling factor and apply to each hour
scalingFact = df4.rlds / df4.temp
scalingFact_day = scalingFact.resample('H').mean()
scale_fact_day_interp = scalingFact_day.interpolate(method='pad')
station.data_disagg[
'ILWR'] = station.data_disagg.temp * scale_fact_day_interp
station.data_disagg['ILWR'][-1] = station.data_disagg['ILWR'][-2]
#plot(data_obs_hourly.ilwr,station.data_disagg.ILWR)
# write out netcdf here
if geotop_out is True:
outname = nc.split('/')[-1]
df_gtop = pd.DataFrame({
"Prec": station.data_disagg.precip,
"Ws": station.data_disagg.wind,
"Wd": station.data_disagg.DW,
"RH": station.data_disagg.hum, #*0.01, #meteoio 0-1
"Tair": station.data_disagg.temp - 273.15,
"SW": station.data_disagg.glob,
"LW": station.data_disagg.ILWR,
})
df_gtop.index = df_gtop.index.strftime("%d/%m/%Y %H:00")
df_gtop.index.name = "Date"
# fill outstanding nan in SW routine with 0 (night)
# fileout=outDir+"/meteo"+"c"+str(i+1)+".csv"
df_gtop.to_csv(path_or_buf="/home/joel/Desktop/" + outname +
".txt",
na_rep=-999,
float_format='%.3f')
if fsm_out is True:
outname = nc.split('/')[-1]
dates = station.data_disagg.temp.index
df_fsm = pd.DataFrame({
# "Sf":snowTot/(60.*60.), # prate in mm/hr to kgm2/s
# "Rf":rainTot/(60.*60.), # prate in mm/hr to kgm2/s
"TA": station.data_disagg.temp,
"RH": station.data_disagg.hum, #*0.01, #meteoio 0-1
"VW": station.data_disagg.wind,
"DW": station.data_disagg.DW,
"P": station.data_disagg.P,
"ISWR": station.data_disagg.glob,
"ILWR": station.data_disagg.ILWR,
"PINT": station.data_disagg.precip # prate mm/hr
})
df_fsm.index.name = "Date"
dffsm = df_fsm.ffill(
) # ensure last day of nans (DW,ILWR,P) is filled with last known values (same as other variables, they are just done at the daily level already)
dffsm.to_csv(
path_or_buf=outdir + '/' + outname + '_HOURLY.txt',
na_rep=-999,
float_format='%.8f',
header=True,
sep=',',
columns=['TA', 'RH', 'VW', 'DW', 'P', 'ISWR', 'ILWR', 'PINT'])
print("WRITTEN: " + outname)
def main(daily_cordex, hourly_obs, mylon, mylat, mytz, slope):
print(mytz)
slope = float(slope)
df_obs = pd.read_csv(hourly_obs, index_col=0, parse_dates=True)
data_obs_hourly = df_obs.loc[:, ('TA', 'RH', 'ISWR', 'ILWR', 'VW')]
data_obs_hourly.rename(columns={
'TA': 'temp',
'RH': 'hum',
'ISWR': 'glob',
'ILWR': 'ilwr',
'VW': 'wind'
},
inplace=True)
# unit conversions
data_obs_hourly['precip'] = df_obs.Rf + df_obs.Sf #mm/hr to to kgm-2s-1
#data_obs_hourly.hum *=100. # 0-1 to 0-100
"""Aggregates data (hourly to daily values) according to the characteristics
of each variable (e.g., average for temperature, sum for precipitation)"""
data_obs_daily = melodist.util.daily_from_hourly(data_obs_hourly) #
# cordex qmap file
df_cordex = pd.read_csv(daily_cordex, index_col=0, parse_dates=True)
#compute wind speed and wind direction
# df2.loc[:,'ws'] = np.sqrt(df.uas**2+df.vas**2)
# df2.loc[:,'wd'] = (180 / np.pi) * np.arctan(df.uas/df.vas) + np.where(df.vas>0,180,np.where(df.uas>0,360,0))
# unit conversions
df_cordex.loc[:, 'PINT'] *= 3600 * 24 # kgm-2s-1 (same as mm /s)-> mm/day
df_cordex.rename(columns={
'TA': 'temp',
'TAMAX': 'tmax',
'TAMIN': 'tmin',
'PINT': 'precip',
'RH': 'hum',
'ISWR': 'glob',
'VW': 'wind'
},
inplace=True)
#df3 =df2.set_index(datetimeindex)
#df_cordex = df3.resample('D').mean()
# filter tmin /tmax error in #'/home/joel/sim/qmap/test/pyout/aresult/MOHC-HadGEM2-ES_historical_r1i1p1_CLMcom-CCLM5-0-6_v1__TS.nc_TS_ALL_ll.nc'
# vals of 500 / 20
df_cordex.tmin[df_cordex.tmin < 220] = np.nan
df_cordex.tmax[df_cordex.tmax < 220] = np.nan
df_cordex.temp[df_cordex.temp < 220] = np.nan
df_cordex.temp[df_cordex.temp > 330] = np.nan
# for case of 360 calender that ends on dec 26 (rare) for example:
#'/home/joel/sim/qmap/test/pyout/aresult/MOHC-HadGEM2-ES_historical_r1i1p1_CLMcom-CCLM5-0-6_v1__TS.nc_TS_ALL_ll.nc'
# we run ffill() for last 5 days
# has no effect if contains data
df_cordex = df_cordex.ffill()
station = melodist.Station(lon=float(mylon),
lat=float(mylat),
timezone=int(mytz),
data_daily=df_cordex)
station.statistics = melodist.StationStatistics(
data_obs_hourly.loc[cal_period])
stats = station.statistics
stats.calc_wind_stats()
stats.calc_humidity_stats()
stats.calc_temperature_stats()
stats.calc_radiation_stats()
#final dissagregation with selected methods
station.disaggregate_temperature(method='sine_mean')
station.disaggregate_humidity(method='equal')
station.disaggregate_radiation(method='pot_rad')
station.disaggregate_wind(method='random')
station.disaggregate_precipitation(method='equal')
# make wind direction equal each hour
#station.data_disagg['DW'] = df_cordex.DW.resample('H').mean().interpolate(method='pad')
# FIll last values that go to NA in interp
# foward fill step -1
#station.data_disagg['DW'][-1] =station.data_disagg['DW'][-2]
# make air pressure equal each hour
station.data_disagg['P'] = df_cordex.P.resample('H').mean().interpolate(
method='pad')
#station.data_disagg['P'][-1] =station.data_disagg['P'][-2]
# disagg lwin with T
# compute a daily scaling factor and apply to each hour
scalingFact = df_cordex.ILWR / df_cordex.temp
scalingFact_day = scalingFact.resample('H').mean()
scale_fact_day_interp = scalingFact_day.interpolate(method='pad')
station.data_disagg[
'ILWR'] = station.data_disagg.temp * scale_fact_day_interp
# station.data_disagg['ILWR'][-1] =station.data_disagg['ILWR'][-2]
#plot(data_obs_hourly.ilwr,station.data_disagg.ILWR)
# partition prate to rain snow (mm/hr)
lowthresh = 272.15
highthresh = 274.15
d = {'prate': station.data_disagg.precip, 'ta': station.data_disagg.temp}
df = pd.DataFrame(data=d)
snow = df.prate.where(df.ta < lowthresh)
rain = df.prate.where(df.ta >= highthresh)
mix1S = df.prate.where((df.ta >= lowthresh) & (df.ta <= highthresh),
inplace=False)
mix1T = df.ta.where((df.ta >= lowthresh) & (df.ta <= highthresh),
inplace=False)
mixSno = (highthresh - mix1T) / (highthresh - lowthresh)
mixRain = 1 - mixSno
addSnow = mix1S * mixSno
addRain = mix1S * mixRain
# nas to 0
snow[np.isnan(snow)] = 0
rain[np.isnan(rain)] = 0
addRain[np.isnan(addRain)] = 0
addSnow[np.isnan(addSnow)] = 0
# # linearly reduce snow to zero in steep slopes
#if steepSnowReduce=="TRUE": # make this an option if need that in future
snowSMIN = 30.
snowSMAX = 80.
slope = slope
k = (snowSMAX - slope) / (snowSMAX - snowSMIN)
if slope < snowSMIN:
k = 1
if slope > snowSMAX:
k = 0
snowTot = (snow + addSnow) * k
rainTot = rain + addRain
#Filter TA for absolute 0 vals - still need this?
station.data_disagg.temp[station.data_disagg.temp < 220] = np.nan
station.data_disagg.temp = station.data_disagg.temp.ffill()
station.data_disagg.P[station.data_disagg.P < 10000] = np.nan
station.data_disagg.P = station.data_disagg.P.ffill()
station.data_disagg.ILWR[station.data_disagg.ILWR < 100] = np.nan
station.data_disagg.ILWR = station.data_disagg.ILWR.ffill()
station.data_disagg.hum[station.data_disagg.hum < 5] = np.nan
station.data_disagg.hum = station.data_disagg.hum.ffill()
outname = daily_cordex.split('.txt')[0] + '_F.txt'
dates = station.data_disagg.temp.index
df_fsm = pd.DataFrame({
"year": dates.year,
"month": dates.month,
"day": dates.day,
"hour": dates.hour,
"ISWR": station.data_disagg.glob,
"ILWR": station.data_disagg.ILWR,
"Sf": snowTot / 3600, # prate in mm/hr to kgm2/s
"Rf": rainTot / 3600, # prate in mm/hr to kgm2/s
"TA": station.data_disagg.temp,
"RH": station.data_disagg.hum, #*0.01, #meteoio 0-1
"VW": station.data_disagg.wind,
"P": station.data_disagg.P,
})
df_fsm.index.name = "Date"
dffsm = df_fsm.ffill(
) # ensure last day of nans (DW,ILWR,P) is filled with last known values (same as other variables, they are just done at the daily level already)
#dffsm_3h = dffsm.resample("3H").mean() #dont need know we do inline processing
dffsm.to_csv(path_or_buf=outname,
na_rep=-999,
float_format='%.4f',
header=False,
sep='\t',
index=False,
columns=[
'year', 'month', 'day', 'hour', 'ISWR', 'ILWR', 'Sf',
'Rf', 'TA', 'RH', 'VW', 'P'
])
print("WRITTEN: " + outname)