def calcPET(lat, time, tmin, tmax, tmean):
'''
Calculates Potential Evapotranspiration using
Hargreaves equation (Hargreaves and Samani, 1985)
'''
latrad = pt.deg2rad(lat) #Latitude to radians
dayofyear = pd.Series(time).dt.day.values
etrad = []
pet = []
# Calculate ET radiation
for x in np.nditer(dayofyear):
soldec = pt.sol_dec(x) #Solar declination
sha = pt.sunset_hour_angle(latrad, soldec) #Sunset hour aingle
ird = pt.inv_rel_dist_earth_sun(
x) #Inverse relative distance Earth-Sun
etrad.append(pt.et_rad(latrad, soldec, sha,
ird)) #Extraterrestrial radiation
# Calculate PET using hargreaves
for x in range(0, len(etrad)):
pet.append(pt.hargreaves(tmin[x], tmax[x], tmean[x], etrad[x]))
pet = np.array(pet)
return (pet)
def hargreaves(self, day, tmax, tmin, tmean, latitude, precipitation, Kc, Ef):
def calculate_extraterrestial_radiation(latitude, day):
# Calculate solar declination
solar_declination = pyeto.sol_dec(day)
# Calculate sunset hour angle
sunset_hour_angle = pyeto.sunset_hour_angle(latitude, solar_declination)
# Calculate inverse relative distance earth-sun
inverse_relative_distance = pyeto.inv_rel_dist_earth_sun(day)
# Calculate extraterrestial radiation
extraterrestial_radiation = pyeto.et_rad(latitude, solar_declination,
sunset_hour_angle, inverse_relative_distance)
return extraterrestial_radiation
# Calculate extraterrestial radiation
extraterrestial_radiation = calculate_extraterrestial_radiation(latitude, day)
# Calculate evapotranspiration
ETo = pyeto.hargreaves(tmin, tmax, tmean, extraterrestial_radiation)
# Predict irrigation
prediction = (ETo * Kc - precipitation) / Ef
return prediction > 0
f_tmax = os.path.join(inpath,'wc2.0_30s_tmax','wc2.0_30s_tmax_'+str(month).zfill(2)+'.tif')
#print('Loading tmax:',f_tmax)
f=gdal.Open(f_tmax)
rasterband = f.GetRasterBand(1)
tmax = rasterband.ReadAsArray(yoff=i,win_ysize=1)
#f.close()
f_tmin = os.path.join(inpath,'wc2.0_30s_tmin','wc2.0_30s_tmin_'+str(month).zfill(2)+'.tif')
#print('Loading tmin:',f_tmax)
f=gdal.Open(f_tmin)
rasterband = f.GetRasterBand(1)
tmin = rasterband.ReadAsArray(yoff=i,win_ysize=1)
#f.close()
# pyeto gives mm/day, multiply by monthdays to get mm/month
pet_timeslice[month-1,:] = pyeto.hargreaves(tmin,tmax,tavg, et_rad)*monthdays[month-1]
# Sum over months to get mm/year, and write out
pet = pet_timeslice.sum(0,keepdims=True)
f_out.variables['pet'][i,:] = pet[0,:]
#print(rasterband)
# Also write data out to tiff format
#rasterband.WriteArray(pet,yoff=i)
# Close tif datasets
#f1 = None
#f2 = None
print('done')
def exec(self):
log.info('[START] {}'.format("exec"))
try:
if (platform.system() == 'Windows'):
# 옵션 설정
sysOpt = {
# 시작/종료 시간
'srtDate': '2020-09-01',
'endDate': '2020-09-03'
# 경도 최소/최대/간격
,
'lonMin': 0,
'lonMax': 360,
'lonInv': 0.5
# 위도 최소/최대/간격
,
'latMin': -90,
'latMax': 90,
'latInv': 0.5
}
else:
# 옵션 설정
sysOpt = {
# 시작/종료 시간
# 'srtDate': globalVar['srtDate']
# , 'endDate': globalVar['endDate']
}
# globalVar['outPath'] = 'F:/Global Temp/aski'
modelList = ['MRI-ESM2-0']
for i, modelInfo in enumerate(modelList):
log.info("[CHECK] modelInfo : {}".format(modelInfo))
inpFile = '{}/{}/{} ssp585 2015-2100_*.nc'.format(
globalVar['inpPath'], serviceName, modelInfo)
fileList = sorted(glob.glob(inpFile))
dsData = xr.open_mfdataset(fileList)
# dsData = dsData.sel(lon=slice(120, 150), time=slice('2015-01', '2016-12'))
# dsData = dsData.sel(lat=slice(50, 50), lon=slice(120, 120), time=slice('2015-01', '2015-12'))
# 월별 시간 변환
dsData['time'] = pd.to_datetime(pd.to_datetime(
dsData['time'].values).strftime("%Y-%m"),
format='%Y-%m')
# 단위 설정
dsData['tasmin'].attrs['units'] = 'degC'
dsData['tasmax'].attrs['units'] = 'degC'
dsData['tas'].attrs['units'] = 'degC'
# 단위 환산을 위한 매월 마지막 날 계산
lon1D = dsData['lon'].values
lat1D = dsData['lat'].values
time1D = dsData['time'].values
timeEndMonth = []
timeYear = dsData['time.year'].values
timeMonth = dsData['time.month'].values
for i in range(0, len(timeYear)):
timeEndMonth.append(
calendar.monthrange(timeYear[i], timeMonth[i])[1])
latRad1D = pyeto.deg2rad(dsData['lat'])
# 2022.08.13 doy 순서 변경
# 시작 순서 1, 32, ...
dayOfYear1D = dsData['time.dayofyear']
latRad3D = np.tile(
np.transpose(np.tile(latRad1D, (len(lon1D), 1))),
(len(time1D), 1, 1))
dayOfYear3D = np.transpose(
np.tile(dayOfYear1D, (len(lon1D), len(lat1D), 1)))
timeEndMonth3D = np.transpose(
np.tile(timeEndMonth, (len(lon1D), len(lat1D), 1)))
tmpData = xr.Dataset(
{
'timeEndMonth':
(('time', 'lat', 'lon'), (timeEndMonth3D).reshape(
len(time1D), len(lat1D), len(lon1D))),
'latRad': (('time', 'lat', 'lon'), (latRad3D).reshape(
len(time1D), len(lat1D), len(lon1D)))
# , 'dayOfYear': (('time', 'lat', 'lon'), (dayOfYear3D).reshape(len(time1D), len(lat1D), len(lon1D)))
,
'doy': (('time', 'lat', 'lon'), (dayOfYear3D).reshape(
len(time1D), len(lat1D), len(lon1D)))
},
coords={
'lat': lat1D,
'lon': lon1D,
'time': time1D
})
# 마지막 순서 31, 59, ...
# tmpData['dayOfYear'] = tmpData['doy']
tmpData['dayOfYear'] = tmpData['doy'] + timeEndMonth3D
# ********************************************************************************************
# FAO-56 Penman-Monteith 방법
# ********************************************************************************************
# https://pyeto.readthedocs.io/en/latest/fao56_penman_monteith.html 매뉴얼 참조
# 1 W/m2 = 1 J/m2를 기준으로 MJ/day 변환
# dsData['rsdsMJ'] = dsData['rsds'] * 86400 / (10 ** 6)
dsData['rsdsMJ'] = dsData['rsds'] * 2592000 / (10**6)
# 섭씨 to 켈빈
dsData['tasKel'] = dsData['tas'] + 273.15
dsData['tasminKel'] = dsData['tasmin'] + 273.15
dsData['tasmaxKel'] = dsData['tasmax'] + 273.15
dsData['svp'] = svp_from_t(dsData['tas'])
dsData['svpMax'] = svp_from_t(dsData['tasmax'])
dsData['svpMin'] = svp_from_t(dsData['tasmin'])
tmpData['solDec'] = sol_dec(tmpData['dayOfYear'])
tmpData['sha'] = sunset_hour_angle(tmpData['latRad'],
tmpData['solDec'])
tmpData['dayLightHour'] = daylight_hours(tmpData['sha'])
tmpData['ird'] = inv_rel_dist_earth_sun(tmpData['dayOfYear'])
# tmpData['etRad'] = et_rad(tmpData['latRad'], tmpData['solDec'], tmpData['sha'], tmpData['ird'])
tmpData['etRad'] = dsData['rsdsMJ']
tmpData['csRad'] = fao.cs_rad(altitude=1.5,
et_rad=tmpData['etRad'])
dsData['deltaSvp'] = delta_svp(dsData['tas'])
# 대기 온도 1.5 m 가정
psy = fao.psy_const(atmos_pres=fao.atm_pressure(altitude=15))
dsData['avp'] = fao.avp_from_rhmin_rhmax(
dsData['svpMax'], dsData['svpMin'], dsData['hurs'].min(),
dsData['hurs'].max())
niSwRad = pyeto.net_in_sol_rad(dsData['rsdsMJ'], albedo=0.23)
niLwRad = net_out_lw_rad(dsData['tasminKel'],
dsData['tasmaxKel'], dsData['rsdsMJ'],
tmpData['csRad'], dsData['avp'])
dsData['net_rad'] = fao.net_rad(ni_sw_rad=niSwRad,
no_lw_rad=niLwRad)
# 2022.08.13
# 상수
# shfData = 0.336
# 이전, 현재 대기온도를 통해 계산
timeList = dsData['time'].values
shfDataL1 = xr.Dataset()
for i, timeInfo in enumerate(timeList):
prevYmd = (pd.to_datetime(timeInfo) +
pd.DateOffset(months=-1)).strftime('%Y-%m-%d')
nowYmd = pd.to_datetime(timeInfo).strftime('%Y-%m-%d')
prevData = dsData['tas'].interp(
time=prevYmd,
method='nearest',
kwargs={'fill_value': 'extrapolate'})
nowData = dsData['tas'].interp(
time=nowYmd,
method='nearest',
kwargs={'fill_value': 'extrapolate'})
shfData = pyeto.monthly_soil_heat_flux2(
prevData, nowData).rename('shf')
shfData['time'] = pd.to_datetime(timeInfo)
if (i == 0):
shfDataL1 = shfData
else:
shfDataL1 = xr.concat([shfDataL1, shfData], dim='time')
dsData = xr.merge([dsData, shfDataL1])
# Daily eto
# faoRes = fao.fao56_penman_monteith(dsData['net_rad'], dsData['tasKel'], dsData['sfcWind'], dsData['svp'], dsData['avp'], dsData['deltaSvp'], psy, shf = 0)
# Monthly eto
faoRes = fao.fao56_penman_monteith(dsData['net_rad'],
dsData['tasKel'],
dsData['sfcWind'],
dsData['svp'],
dsData['avp'],
dsData['deltaSvp'],
psy,
shf=dsData['shf'])
# ********************************************************************************************
# Hargreaves 방법
# ********************************************************************************************
# https://xclim.readthedocs.io/en/stable/indicators_api.html 매뉴얼 참조
# harRes = xclim.indices.potential_evapotranspiration( dsData['tasmin'], dsData['tasmax'], dsData['tas'], dsData['lat'], method='hargreaves85')
# 1 kg/m2/s = 86400 mm/day를 기준으로 mm/month 변환
# harResL1 = harRes * 86400.0 * tmpData['timeEndMonth']
# https://pyeto.readthedocs.io/en/latest/thornthwaite.html 매뉴얼 참조
harRes = pyeto.hargreaves(dsData['tasmin'], dsData['tasmax'],
dsData['tas'], tmpData['etRad'])
harResL1 = harRes
# ********************************************************************************************
# Thornthwaite 방법
# ********************************************************************************************
# https://xclim.readthedocs.io/en/stable/indicators_api.html 매뉴얼 참조
# thwRes = xclim.indices.potential_evapotranspiration(dsData['tasmin'], dsData['tasmax'], dsData['tas'], dsData['lat'], method ='thornthwaite48')
# 1 kg/m2/s = 86400 mm/day를 기준으로 mm/month 변환
# thwResL1 = thwRes * 86400.0 * tmpData['timeEndMonth']
dsData['tasAdj'] = xr.where((dsData['tas'] >= 0),
dsData['tas'], 0)
dsData['heatIdx'] = (dsData['tasAdj'] / 5.0)**1.514
sumHeatIdx = dsData['heatIdx'].groupby('time.year').sum(
skipna=True)
sumHeatIdxData = xr.Dataset()
timeList = dsData['time'].values
for i, timeInfo in enumerate(timeList):
iYear = int(pd.to_datetime(timeInfo).strftime('%Y'))
selHeatIdx = sumHeatIdx.sel(year=iYear).rename(
{'year': 'time'})
selHeatIdx['time'] = timeInfo
if (i == 0):
sumHeatIdxData = selHeatIdx
else:
sumHeatIdxData = xr.concat(
[sumHeatIdxData, selHeatIdx], dim='time')
# 2022.08.15
# dsData['thwConst'] = (6.75e-07 * dsData['heatIdx'] ** 3) - (7.71e-05 * dsData['heatIdx'] ** 2) + (1.792e-02 * dsData['heatIdx']) + 0.49239
# thwRes = 1.6 * (tmpData['dayLightHour'] / 12.0) * (tmpData['timeEndMonth'] / 30.0) * ((10.0 * dsData['tasAdj'] / dsData['heatIdx']) ** dsData['thwConst']) * 10.0
dsData['thwConst'] = (6.75e-07 * sumHeatIdxData**3) - (
7.71e-05 * sumHeatIdxData**2) + (1.792e-02 *
sumHeatIdxData) + 0.49239
thwRes = 1.6 * (tmpData['dayLightHour'] /
12.0) * (tmpData['timeEndMonth'] / 30.0) * (
(10.0 * dsData['tasAdj'] / sumHeatIdxData)
**dsData['thwConst']) * 10.0
# 0보다 작은 경우 0으로 대체
thwRes = xr.where((thwRes >= 0), thwRes, 0)
# ********************************************************************************************
# 데이터 병합
# ********************************************************************************************
etoData = xr.Dataset(
{
'hargreaves':
(('time', 'lat', 'lon'), (harResL1.values).reshape(
len(time1D), len(lat1D), len(lon1D))),
'thornthwaite':
(('time', 'lat', 'lon'), (thwRes.values).reshape(
len(time1D), len(lat1D), len(lon1D))),
'penman-monteith':
(('time', 'lat', 'lon'), (faoRes.values).reshape(
len(time1D), len(lat1D), len(lon1D)))
},
coords={
'lat': lat1D,
'lon': lon1D,
'time': time1D
})
# NetCDF 파일 저장
saveFile = '{}/{}/{}_eto.nc'.format(globalVar['outPath'],
serviceName, modelInfo)
os.makedirs(os.path.dirname(saveFile), exist_ok=True)
etoData.to_netcdf(saveFile)
log.info('[CHECK] saveFile : {}'.format(saveFile))
except Exception as e:
log.error("Exception : {}".format(e))
raise e
finally:
log.info('[END] {}'.format("exec"))
def _compute_ref_eto(day_ds, lat):
return pyeto.hargreaves(
day_ds['TminD'], day_ds['TabsD'], day_ds['TmaxD'],
_compute_solar_radiation(pd.to_datetime(day_ds.time.values), lat))
def test_hargreaves(self):
# Tested against worked example from "Estimating Evapotranspiration
# from weather data" by Vishal K. Mehta, Arghyam/Cornell University,
# Nov 2, 2006.
eto = pyeto.hargreaves(tmin=28, tmax=38, tmean=35, et_rad=38.93715)
self.assertAlmostEqual(eto, 6.1, 1)
def test_hargreaves(self):
# Tested against worked example from "Estimating Evapotranspiration
# from weather data" by Vishal K. Mehta, Arghyam/Cornell University,
# Nov 2, 2006.
eto = pyeto.hargreaves(tmin=28, tmax=38, tmean=35, et_rad=38.93715)
self.assertAlmostEqual(eto, 6.1, 1)