Skip to content

Instantly share code, notes, and snippets.

@practice

practice/aerosol_classify.py

Last active January 2, 2024 07:23
Show Gist options
  • Save practice/46c20582214d0ed2581fd5e446e4261a to your computer and use it in GitHub Desktop.
Save practice/46c20582214d0ed2581fd5e446e4261a to your computer and use it in GitHub Desktop.
Download ZIP
Raw
aerosol_classify.py
'''Aerosol Classification Code'''
#%% Packages#######################################################################################################
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import datetime
from datetime import timedelta
import glob
import os
os.getcwd()
def load_df(file):
print(f"loading file:{file}")
###############bring data######################################
df = pd.read_table(file, sep=",", skiprows=6)
###############change column name of fmf#######################
if file.endswith('.ONEILL_lev15'):
df.rename(columns={'Date_(dd:mm:yyyy)':'Date(dd:mm:yyyy)'}, inplace=True)
df.rename(columns={'Time_(hh:mm:ss)':'Time(hh:mm:ss)'}, inplace=True)
###############remove -999#####################################
df = df.replace(to_replace = -999, value = np.nan)
###############converge date and time##########################
# print(df.columns)
df['times'] = df[['Date(dd:mm:yyyy)','Time(hh:mm:ss)']].apply(lambda x:' '.join(x),axis=1)
df.drop(columns=['Date(dd:mm:yyyy)', 'Time(hh:mm:ss)'], inplace=True)
##############make datetime index##############################
df['times'] = pd.to_datetime(df['times'], format = "%d:%m:%Y %H:%M:%S")
df['times'] = pd.DatetimeIndex(df['times']) + timedelta(hours = 9)
df = df.set_index('times')
return df
#%% File Path #######################################################################################################
path = '/work/nb/aeronet/20190101_20231231_Yonsei_University' #file path + period + site name (all point data)
output_dir = '/work/nb/aeronet/'
#### Bring every products (all point) ####
# AEROSOL OPTICAL DEPTH Category
aod_file = path+'.lev15' # 1.5 : Aerosol Optical Depth (AOD) with Precipitable Water and Angstrom Parameter
fmf_file = path+'.ONEILL_lev15' # 1.5 : Spectral Deconvolution Algorithm (SDA) Retrievals --Fine Mode AOD, Coarse Mode AOD, and Fine Mode Fraction
# AEROSOL INVERSION Category
aae_file = path+'.tab' # 1.5 Almucantar : Absorption AOD
eae_file = path+'.aod' # 1.5 Almucantar : Extinction AOD
data_aod = load_df(aod_file)
data_fmf = load_df(fmf_file)
data_aae = load_df(aae_file)
data_eae = load_df(eae_file)
#%% ARRANGE #######################################################################################################
data_AOD = data_aod[['AOD_440nm','AOD_500nm']]
data_FMF = data_fmf[['FineModeFraction_500nm[eta]']]
data_AAE = data_aae[['Absorption_AOD[440nm]','Absorption_AOD[870nm]','Absorption_Angstrom_Exponent_440-870nm']]
data_EAE = data_eae[['AOD_Extinction-Total[440nm]','AOD_Extinction-Total[870nm]']]
#site
site_name = data_aae.Site[0]
#merge AOD-FMF, AAE-EAE (depend on time index)
data_AOD_FMF = data_AOD.join(data_FMF, how='left')
data_AAE_EAE = data_AAE.join(data_EAE, how='left')
#latest date
date_latest = data_AAE_EAE.index[-1]
#%% Calculte SAE #######################################################################################################
data_AAE_EAE = data_AAE_EAE.rename(columns={'Absorption_Angstrom_Exponent_440-870nm':'AAE'})
data_AAE_EAE['scattering440'] = data_AAE_EAE['AOD_Extinction-Total[440nm]']-data_AAE_EAE['Absorption_AOD[440nm]']
data_AAE_EAE['scattering870'] = data_AAE_EAE['AOD_Extinction-Total[870nm]']-data_AAE_EAE['Absorption_AOD[870nm]']
data_AAE_EAE['SAE'] = -np.log(data_AAE_EAE['scattering440']/data_AAE_EAE['scattering870'])/np.log(440/870)
#%% Make Daily data ######################################################################################################
data_AOD_FMF = data_AOD_FMF.resample('1d').mean()
#std of AAE and SAE
data_AAE_std = data_AAE_EAE['AAE'].resample('1d').std()
data_SAE_std = data_AAE_EAE['SAE'].resample('1d').std()
data_AAE_SAE = data_AAE_EAE.resample('1d').mean()
data_AAE_SAE['SAE_std'] = data_SAE_std
data_AAE_SAE['AAE_std'] = data_AAE_std
data_AAE_SAE = data_AAE_SAE[['AAE','SAE','AAE_std','SAE_std']]
### total 7 products ###
data = data_AOD_FMF.join(data_AAE_SAE, how='left')
#%% Aerosol Classification ####################################################################################################
#make columns
data['type'] = None
data['color'] = None
data['bc'] = np.nan
data.astype({'type':'object'})
data.astype({'color':'object'})
#AAE-SAE criteria
for index, row in data.iterrows():
if row['AOD_440nm'] <= 0.4: ##classification AOD minimum criteria
data.at[index, 'type'] = 'Low'
data.at[index, 'color'] = 'white'
if row['AOD_440nm'] > 0.4:
if row['SAE'] <= 0 and row['AAE'] >= 2:
data.at[index, 'type'] = 'Dust'
data.at[index, 'color'] = 'gold'
if row['SAE'] <= 1.5 and row['AAE'] >= 1.5 \
and row['FineModeFraction_500nm[eta]'] < 0.4:
data.at[index, 'type'] = 'Dust'
data.at[index, 'color'] = 'gold'
if row['SAE']<=1.5 and row['AAE']>=1.5 \
and row['FineModeFraction_500nm[eta]']>=0.4 :
data.at[index, 'type'] = 'BC+BrC'
data.at[index, 'color'] = 'tan'
if row['SAE']<1 and row['AAE']<1:
data.at[index, 'type'] = 'Uncertain'
data.at[index, 'color'] = 'skyblue'
if row['SAE']>=1 and row['AAE']<1:
data.at[index, 'type'] = 'NA'
data.at[index, 'color'] = 'hotpink'
if row['SAE']<1.5 and row['AAE']>=1.0 \
and row['AAE']<1.5 and \
row['AAE']>row['SAE'] \
and row['FineModeFraction_500nm[eta]']>0.6:
data.at[index, 'bc'] = 1
if row['SAE']<1.5 and row['AAE']>=1.0 \
and row['AAE']<1.5 and \
row['AAE'] > row['SAE'] \
and row['FineModeFraction_500nm[eta]']<=0.6:
data.at[index, 'type'] = 'Uncertain'
data.at[index, 'color'] = 'skyblue'
if row['SAE']>=1 and row['AAE']>=1.0 \
and row['AAE']<1.5 and \
row['AAE']<=row['SAE']:
data.at[index, 'bc'] = 1
if row['SAE']>=1.5 and row['AAE']>1.5:
data.at[index, 'bc'] = 1
if row['bc'] == 1 and row['AAE']>=2:
data.at[index, 'type'] = 'BrC'
data.at[index, 'color'] = 'brown'
if row['bc'] == 1 and 1.5<=row['AAE']<2:
data.at[index, 'type'] = 'BC+BrC'
data.at[index, 'color'] = 'tan'
if row['bc'] == 1 and row['AAE']<1.5:
data.at[index, 'type'] = 'BC'
data.at[index, 'color'] = 'dimgray'
#%% ##############################################################################
#%% Date #######################################################################################################
# date_today = pd.Timestamp(datetime.date.today())
date_today = pd.Timestamp(datetime.datetime(2019,1,17))
data_recent = data[date_today-timedelta(days=10):date_today]
data_recent = data_recent.dropna(axis=0, subset=['AAE']) #if AAE is none, delete row
#%%1. Make Grid ##############################################################################
fig, ax = plt.subplots(figsize = (12,12))
#Data Plotting
plt.scatter(data_recent['SAE'], data_recent['AAE'], c=data_recent['color'], s=300, edgecolors='red', marker='o', zorder=2)
plt.errorbar(data_recent['SAE'], data_recent['AAE'], xerr=data_recent['SAE_std'], yerr=data_recent['AAE_std'], color='None', \
ecolor='red',elinewidth=1.5, capsize=8, zorder=1)
#Write down Dates
for t in range (0, np.size(data_recent.index)):
if np.isnan(data_recent['SAE'][t]) == False:
plt.text(data_recent['SAE'][t]+0.03, data_recent['AAE'][t]+0.03, data_recent.index[t].strftime("%m%d"), fontsize=15, fontweight='bold')
plt.xlim(-0.5,2.51)
plt.xticks(np.arange(-0.5,2.501,0.5),fontsize=15)
plt.ylim(0,3)
plt.yticks(np.arange(0,3.01,1),fontsize=15)
plt.xlabel('SAE 440-870nm', fontsize=25)
plt.ylabel('AAE 440-870nm', fontsize=25)
plt.title('< Aerosol Type Classification [ '+site_name+' ] >', fontsize=20, fontweight='bold')
#draw lines
plt.plot([1,1],[0,1],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([-0.5,2.5],[1,1],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([-0.5,2.5],[1.5,1.5],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([1.5,1.5],[1.5,3],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([1.5,2.5],[1.5,1.5],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([1.5,2.5],[2,2],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([-0.5,0],[2,2],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([0,0],[2,3],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([1,1.5],[1,1.5],'-',color='grey',linewidth=1,alpha = 0.8)
#type names
plt.text(-0.1,0.45,'Uncertain', fontsize=30)
plt.text(-0.28,0.33,'large and low absorbing', fontsize=18, color='grey')
plt.text(-0.4,2.43,'Dust', fontsize=30)
plt.text(1.63,0.45,' NA', fontsize=30)
plt.text(1.13,0.33,' small and low (non) absorbing', fontsize=18, color='grey')
plt.text(1.88,1.18,'BC', fontsize=30)
plt.text(1.7,1.06,'black carbon', fontsize=18, color='grey')
plt.text(1.7,1.68,'BC + BrC', fontsize=30)
plt.text(1.88,2.43,'BrC', fontsize=30)
plt.text(1.72,2.31,'brown carbon', fontsize=18, color='grey')
plt.text(-0.1,1.2,'Uncertain / BC', fontsize=30)
plt.text(0.3,2.2,'Dust / BC + BrC', fontsize=30)
#fmf criteria
plt.text(-0.1,1.06,'FMF<0.6', fontsize=12, color='blue')
plt.text(0.68,1.06,'FMF>0.6', fontsize=12, color='blue')
plt.text(-0.1,1.15,'-------------------------- --------', fontsize=16, color='blue')
plt.text(0.3,2.06,'FMF<0.4', fontsize=12, color='blue')
plt.text(0.75,2.06,'FMF>0.4', fontsize=12, color='blue')
plt.text(0.3,2.15,'------------ -------------------------', fontsize=16, color='blue')
fig.savefig(output_dir + 'criteria_'+date_today.strftime("%y%m%d")+'.png', dpi=300)
#%% 2. Make Legend ##############################################################################
fig2, ax2 = plt.subplots(figsize = (6,12))
plt.xlim(0,2.8)
plt.ylim(0,4)
plt.text(1.2,2.8,'< Aerosol Types >', fontsize=20, fontweight='bold')
plt.scatter(0.8,2.5, c='gold',s=300, edgecolor='red')
plt.text(1.1,2.47,'Dust', fontsize=20)
plt.scatter(0.8,2.3, c='dimgray',s=300, edgecolor='red')
plt.text(1.1,2.27,'Black Carbon', fontsize=20)
plt.scatter(0.8,2.1, c='tan',s=300, edgecolor='red')
plt.text(1.1,2.07,'Black + Brown Carbon', fontsize=20)
plt.scatter(0.8,1.9, c='brown',s=300, edgecolor='red')
plt.text(1.1,1.87,'Brown Carbon', fontsize=20)
plt.scatter(0.8,1.7, c='hotpink',s=300, edgecolor='red')
plt.text(1.1,1.67,'Non-Absorbing', fontsize=20)
plt.scatter(0.8,1.5, c='skyblue',s=300, edgecolor='red')
plt.text(1.1,1.47,'Uncertain', fontsize=20)
plt.scatter(0.8,1.2, c='white',s=300, edgecolor='red')
plt.text(1.1,1.17,'Low (AOD440 < 0.4)', fontsize=20)
plt.scatter(0.73,1.0, c='red',s=300, marker=1)
plt.text(1.1,0.97,'Daily Std', fontsize=20)
plt.axis(False)
fig2.savefig(output_dir + 'legend_'+date_today.strftime("%y%m%d")+'.png', dpi=300)
#%%3. Make Pie Chart ########################################################################################
if data_recent.count()[0] == 0 : #if data is absent, show 'no data available'
fig3, ax3 = plt.subplots(figsize = (12,12))
plt.xlim(0,2)
plt.ylim(0,2)
plt.axis(False)
plt.text(0.5,1,'No Data Available', fontsize=30)
fig3.savefig(output_dir + 'pie_chart.png', dpi=300)
else:
fig3, ax3 = plt.subplots(figsize = (12,12))
labels = ['Dust','BC','BC+BrC','Low','NA','Uncertain']
colors = ['gold','dimgray','tan','white','hotpink','skyblue']
ratio = []
l_ind = 0
for l in labels:
ratio.append(len(data_recent.loc[data_recent['type']==l]))
if len(data_recent.loc[data_recent['type']==l]) == 0:
labels[l_ind] = '' #blank
l_ind = l_ind+1
plt.pie(ratio, labels=labels, colors=colors, autopct = '%.1f%%', \
wedgeprops = {'width' : 1, 'edgecolor' : 'grey', 'linewidth' : 3}, textprops={'size' : 30, 'color' : 'black'} )
plt.title('< Recent Aerosol Types : '+data_recent.index[0].strftime("%y%m%d")+' ~ '+data_recent.index[-1].strftime("%y%m%d")+' >', fontsize=20, fontweight='bold')
fig3.savefig(output_dir + 'pie_chart_'+date_today.strftime("%y%m%d")+'.png', dpi=300)
# %%
Raw
aerosol_classify_2023_12_23.py
import json
import pathlib
import requests
import airflow
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import datetime
from datetime import timedelta
import glob
import os
from airflow import DAG
from airflow.operators.python import PythonOperator
from airflow.models import Variable
dag = DAG(
description="AERONET classification by UNIST FineParticle web",
dag_id="aeronet_classify",
start_date=airflow.utils.dates.days_ago(0, second=10),
schedule_interval=datetime.timedelta(minutes=30),
max_active_runs=1,
)
def load_df(file):
print(f"loading file:{file}")
###############bring data######################################
if file.endswith('if_no_html=1'): #만약 aae, eae 파일이면
df = pd.read_table(file, sep=",", skiprows=6)
else:
df = pd.read_table(file, sep=",", skiprows=7)
df = df.drop(df.index[-1])
###############change column name of fmf#######################
if file.endswith('SDA15=1&AVG=10'): #만약 fmf 파일이면
df.rename(columns={'Date_(dd:mm:yyyy)':'Date(dd:mm:yyyy)'}, inplace=True)
df.rename(columns={'Time_(hh:mm:ss)':'Time(hh:mm:ss)'}, inplace=True)
###############remove -999#####################################
df = df.replace(to_replace = -999, value = np.nan)
###############converge date and time##########################
# print(df.columns)
df['times'] = df[['Date(dd:mm:yyyy)','Time(hh:mm:ss)']].apply(lambda x:' '.join(x),axis=1)
df.drop(columns=['Date(dd:mm:yyyy)', 'Time(hh:mm:ss)'], inplace=True)
##############make datetime index##############################
df['times'] = pd.to_datetime(df['times'], format = "%d:%m:%Y %H:%M:%S")
df['times'] = pd.DatetimeIndex(df['times']) + timedelta(hours = 9)
df = df.set_index('times')
print(f"loading file:{file} done.")
return df
def gen_image(finep_url, year, month, day):
os.umask(0)
today = datetime.datetime(int(year),int(month),int(day))
yesterday = today - timedelta(days=1)
sites = ["Socheongcho", "Anmyon", "Seoul_SNU", "Yonsei_University", "Gangneung_WNU", "KORUS_UNIST_Ulsan"]
# sites = ["Yonsei_University"]
for site in sites:
for a_day in [today, yesterday]:
gen_image_for_site(site, finep_url, a_day)
# DB save
print(f"calling finep /handle-aeronet-classify file handling for year={year},month={month},day={day}")
url = f"{finep_url}/handle-aeronet-classify"
print(f"posting {url}")
res = requests.post(url)
if res.status_code != 200:
print(res.text)
raise ValueError(f"Error response: {res.status_code}")
return res.text
def gen_image_for_site(site, finep_url, day):
print(f"gen_image called with: {site},{finep_url},{day}")
#### Bring every products (all point) ####
# AEROSOL OPTICAL DEPTH Category
# aod_file = path+'.lev15' # 1.5 : Aerosol Optical Depth (AOD) with Precipitable Water and Angstrom Parameter
# fmf_file = path+'.ONEILL_lev15' # 1.5 : Spectral Deconvolution Algorithm (SDA) Retrievals --Fine Mode AOD, Coarse Mode AOD, and Fine Mode Fraction
# AEROSOL INVERSION Category
# aae_file = path+'.tab' # 1.5 Almucantar : Absorption AOD
# eae_file = path+'.aod' # 1.5 Almucantar : Extinction AOD
#%% File Path #######################################################################################################
output_dir = '/data/finepart/aeronet-classify/temp/'
os.makedirs(output_dir, exist_ok=True)
date_today = pd.Timestamp(day)
date_before = pd.Timestamp(date_today-timedelta(days=20))
# AEROSOL OPTICAL DEPTH Category
# 1.5 : Aerosol Optical Depth (AOD) with Precipitable Water and Angstrom Parameter
aod_file = f'https://aeronet.gsfc.nasa.gov/cgi-bin/print_web_data_v3?site={site}&year={date_before.year}&month={date_before.month}&day={date_before.day}&year2={date_today.year}&month2={date_today.month}&day2={date_today.day}&AOD15=1&AVG=10'
# 1.5 : Spectral Deconvolution Algorithm (SDA) Retrievals --Fine Mode AOD, Coarse Mode AOD, and Fine Mode Fraction
fmf_file = f'https://aeronet.gsfc.nasa.gov/cgi-bin/print_web_data_v3?site={site}&year={date_before.year}&month={date_before.month}&day={date_before.day}&year2={date_today.year}&month2={date_today.month}&day2={date_today.day}&SDA15=1&AVG=10'
# AEROSOL INVERSION Category
# 1.5 Almucantar : Absorption AOD
aae_file = f'https://aeronet.gsfc.nasa.gov/cgi-bin/print_web_data_inv_v3?site={site}&year={date_before.year}&month={date_before.month}&day={date_before.day}&year2={date_today.year}&month2={date_today.month}&day2={date_today.day}&product=TAB&AVG=10&ALM15=1&if_no_html=1'
# 1.5 Almucantar : Extinction AOD
eae_file = f'https://aeronet.gsfc.nasa.gov/cgi-bin/print_web_data_inv_v3?site={site}&year={date_before.year}&month={date_before.month}&day={date_before.day}&year2={date_today.year}&month2={date_today.month}&day2={date_today.day}&product=AOD&AVG=10&ALM15=1&if_no_html=1'
try:
data_aod = load_df(aod_file)
data_fmf = load_df(fmf_file)
data_aae = load_df(aae_file)
data_eae = load_df(eae_file)
#%% ARRANGE #######################################################################################################
data_AOD = data_aod[['AOD_440nm','AOD_500nm']]
data_FMF = data_fmf[['FineModeFraction_500nm[eta]']]
if site == 'Socheongcho':
data_AOD = data_aod[['AOD_443nm','AOD_500nm']]
data_AAE = data_aae[['Absorption_AOD[443nm]','Absorption_AOD[870nm]','Absorption_Angstrom_Exponent_440-870nm']]
data_EAE = data_eae[['AOD_Extinction-Total[443nm]','AOD_Extinction-Total[870nm]']]
data_AOD = data_AOD.rename(columns={'AOD_443nm':'AOD_440nm'})
data_AAE = data_AAE.rename(columns={'Absorption_AOD[443nm]':'Absorption_AOD[440nm]'})
data_EAE = data_EAE.rename(columns={'AOD_Extinction-Total[443nm]':'AOD_Extinction-Total[440nm]'})
else:
data_AOD = data_aod[['AOD_440nm','AOD_500nm']]
data_AAE = data_aae[['Absorption_AOD[440nm]','Absorption_AOD[870nm]','Absorption_Angstrom_Exponent_440-870nm']]
data_EAE = data_eae[['AOD_Extinction-Total[440nm]','AOD_Extinction-Total[870nm]']]
#merge AOD-FMF, AAE-EAE (depend on time index)
data_AOD_FMF = data_AOD.join(data_FMF, how='left')
data_AAE_EAE = data_AAE.join(data_EAE, how='left')
#latest date
date_latest = data_AAE_EAE.index[-1]
#%% Calculte SAE #######################################################################################################
data_AAE_EAE = data_AAE_EAE.rename(columns={'Absorption_Angstrom_Exponent_440-870nm':'AAE'})
data_AAE_EAE['scattering440'] = data_AAE_EAE['AOD_Extinction-Total[440nm]']-data_AAE_EAE['Absorption_AOD[440nm]']
data_AAE_EAE['scattering870'] = data_AAE_EAE['AOD_Extinction-Total[870nm]']-data_AAE_EAE['Absorption_AOD[870nm]']
data_AAE_EAE['SAE'] = -np.log(data_AAE_EAE['scattering440']/data_AAE_EAE['scattering870'])/np.log(440/870)
#%% Make Daily data ######################################################################################################
data_AOD_FMF = data_AOD_FMF.resample('1d').mean()
#std of AAE and SAE
data_AAE_std = data_AAE_EAE['AAE'].resample('1d').std()
data_SAE_std = data_AAE_EAE['SAE'].resample('1d').std()
data_AAE_SAE = data_AAE_EAE.resample('1d').mean()
data_AAE_SAE['SAE_std'] = data_SAE_std
data_AAE_SAE['AAE_std'] = data_AAE_std
data_AAE_SAE = data_AAE_SAE[['AAE','SAE','AAE_std','SAE_std']]
### total 7 products ###
data = data_AOD_FMF.join(data_AAE_SAE, how='left')
#AAE-SAE criteria
for index, row in data.iterrows():
bc = np.nan
if row['AOD_440nm'] <= 0.4: ##classification AOD minimum criteria
data.at[index, 'type'] = 'Low'
data.at[index, 'color'] = 'white'
if row['AOD_440nm'] > 0.4:
if row['SAE'] <= 0 and row['AAE'] >= 2:
data.at[index, 'type'] = 'Dust'
data.at[index, 'color'] = 'gold'
if row['SAE'] <= 1.5 and row['AAE'] >= 1.5 and row['FineModeFraction_500nm[eta]'] < 0.4:
data.at[index, 'type'] = 'Dust'
data.at[index, 'color'] = 'gold'
if row['SAE']<=1.5 and row['AAE']>=1.5 and row['FineModeFraction_500nm[eta]']>=0.4:
data.at[index, 'type'] = 'BC+BrC'
data.at[index, 'color'] = 'tan'
if row['SAE']<1 and row['AAE']<1:
data.at[index, 'type'] = 'Uncertain'
data.at[index, 'color'] = 'skyblue'
if row['SAE']>=1 and row['AAE']<1:
data.at[index, 'type'] = 'NA'
data.at[index, 'color'] = 'hotpink'
if row['SAE']<1.5 and row['AAE']>=1.0 and row['AAE']<1.5 and row['AAE']>row['SAE'] and row['FineModeFraction_500nm[eta]']>0.6:
bc = 1
if row['SAE']<1.5 and row['AAE']>=1.0 and row['AAE']<1.5 and row['AAE'] > row['SAE'] and row['FineModeFraction_500nm[eta]']<=0.6:
data.at[index, 'type'] = 'Uncertain'
data.at[index, 'color'] = 'skyblue'
if row['SAE']>=1 and row['AAE']>=1.0 and row['AAE']<1.5 and row['AAE']<=row['SAE']:
bc = 1
if row['SAE']>=1.5 and row['AAE']>1.5:
bc = 1
if bc == 1:
data.at[index, 'bc'] = 1
if bc == 1 and row['AAE']>=2:
data.at[index, 'type'] = 'BrC'
data.at[index, 'color'] = 'brown'
if bc == 1 and 1.5<=row['AAE']<2:
data.at[index, 'type'] = 'BC+BrC'
data.at[index, 'color'] = 'tan'
if bc == 1 and row['AAE']<1.5:
data.at[index, 'type'] = 'BC'
data.at[index, 'color'] = 'dimgray'
#%% ##############################################################################
#%% Date #######################################################################################################
data_recent = data[date_today-timedelta(days=10):date_today]
data_recent = data_recent.dropna(axis=0, subset=['AAE', 'AOD_440nm']) #if AAE is none, delete row
#%%1. Make Grid ##############################################################################
fig, ax = plt.subplots(figsize = (12,12))
#Data Plotting
plt.scatter(data_recent['SAE'], data_recent['AAE'], c=data_recent['color'], s=300, edgecolors='red', marker='o', zorder=2)
plt.errorbar(data_recent['SAE'], data_recent['AAE'], xerr=data_recent['SAE_std'], yerr=data_recent['AAE_std'], color='None', \
ecolor='red',elinewidth=1.5, capsize=8, zorder=1)
#Write down Dates
for t in range (0, np.size(data_recent.index)):
if np.isnan(data_recent['SAE'][t]) == False and -0.5<data_recent['SAE'][t]<2.5 and 0<data_recent['AAE'][t]<3:
plt.text(data_recent['SAE'][t]+0.03, data_recent['AAE'][t]+0.03, data_recent.index[t].strftime("%m%d"), fontsize=15, fontweight='bold')
plt.xlim(-0.5,2.51)
plt.xticks(np.arange(-0.5,2.501,0.5),fontsize=15)
plt.ylim(0,3)
plt.yticks(np.arange(0,3.01,1),fontsize=15)
plt.xlabel('SAE 440-870nm', fontsize=25)
plt.ylabel('AAE 440-870nm', fontsize=25)
plt.title('< Aerosol Type Classification [ '+site+' ] >', fontsize=20, fontweight='bold')
#draw lines
plt.plot([1,1],[0,1],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([-0.5,2.5],[1,1],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([-0.5,2.5],[1.5,1.5],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([1.5,1.5],[1.5,3],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([1.5,2.5],[1.5,1.5],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([1.5,2.5],[2,2],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([-0.5,0],[2,2],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([0,0],[2,3],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([1,1.5],[1,1.5],'-',color='grey',linewidth=1,alpha = 0.8)
#type names
plt.text(-0.1,0.45,'Uncertain', fontsize=30)
plt.text(-0.28,0.33,'large and low absorbing', fontsize=18, color='grey')
plt.text(-0.4,2.43,'Dust', fontsize=30)
plt.text(1.63,0.45,' NA', fontsize=30)
plt.text(1.13,0.33,' small and low (non) absorbing', fontsize=18, color='grey')
plt.text(1.88,1.18,'BC', fontsize=30)
plt.text(1.7,1.06,'black carbon', fontsize=18, color='grey')
plt.text(1.7,1.68,'BC + BrC', fontsize=30)
plt.text(1.88,2.43,'BrC', fontsize=30)
plt.text(1.72,2.31,'brown carbon', fontsize=18, color='grey')
plt.text(-0.1,1.2,'Uncertain / BC', fontsize=30)
plt.text(0.3,2.2,'Dust / BC + BrC', fontsize=30)
#fmf criteria
plt.text(-0.1,1.06,'FMF<0.6', fontsize=12, color='blue')
plt.text(0.68,1.06,'FMF>0.6', fontsize=12, color='blue')
plt.text(-0.1,1.15,'-------------------------- --------', fontsize=16, color='blue')
plt.text(0.3,2.06,'FMF<0.4', fontsize=12, color='blue')
plt.text(0.75,2.06,'FMF>0.4', fontsize=12, color='blue')
plt.text(0.3,2.15,'------------ -------------------------', fontsize=16, color='blue')
# _-_ is seperator between site and the rest of filename
fig.savefig(output_dir + site + '_-_criteria_'+date_today.strftime("%Y%m%d")+'.png', dpi=300)
#%% 2. Make Legend ##############################################################################
fig2, ax2 = plt.subplots(figsize = (6,12))
plt.xlim(0,2.8)
plt.ylim(0,4)
plt.text(1.2,2.8,'< Aerosol Types >', fontsize=20, fontweight='bold')
plt.scatter(0.8,2.5, c='gold',s=300, edgecolor='red')
plt.text(1.1,2.47,'Dust', fontsize=20)
plt.scatter(0.8,2.3, c='dimgray',s=300, edgecolor='red')
plt.text(1.1,2.27,'Black Carbon', fontsize=20)
plt.scatter(0.8,2.1, c='tan',s=300, edgecolor='red')
plt.text(1.1,2.07,'Black + Brown Carbon', fontsize=20)
plt.scatter(0.8,1.9, c='brown',s=300, edgecolor='red')
plt.text(1.1,1.87,'Brown Carbon', fontsize=20)
plt.scatter(0.8,1.7, c='hotpink',s=300, edgecolor='red')
plt.text(1.1,1.67,'Non-Absorbing', fontsize=20)
plt.scatter(0.8,1.5, c='skyblue',s=300, edgecolor='red')
plt.text(1.1,1.47,'Uncertain', fontsize=20)
plt.scatter(0.8,1.2, c='white',s=300, edgecolor='red')
plt.text(1.1,1.17,'Low (AOD440 < 0.4)', fontsize=20)
plt.scatter(0.73,1.0, c='red',s=300, marker=1)
plt.text(1.1,0.97,'Daily Std', fontsize=20)
plt.axis(False)
fig2.savefig(output_dir + site + '_-_legend_'+date_today.strftime("%Y%m%d")+'.png', dpi=300)
#%%3. Make Pie Chart ########################################################################################
if data_recent.count()[0] == 0 : #if data is absent, show 'no data available'
fig3, ax3 = plt.subplots(figsize = (12,12))
plt.xlim(0,2)
plt.ylim(0,2)
plt.axis(False)
plt.text(0.5,1,'No Data Available', fontsize=20)
fig3.savefig(output_dir + site + '_-_piechart_'+date_today.strftime("%Y%m%d")+'.png', dpi=300)
# fig3.savefig(output_dir + site + '_-_piechart.png', dpi=300)
else:
fig3, ax3 = plt.subplots(figsize = (12,12))
labels = ['Dust','BC','BC+BrC','Low','NA','Uncertain']
colors = ['gold','dimgray','tan','white','hotpink','skyblue']
ratio = []
l_ind = 0
for l in labels:
ratio.append(len(data_recent.loc[data_recent['type']==l]))
if len(data_recent.loc[data_recent['type']==l]) == 0:
labels[l_ind] = '' #blank
l_ind = l_ind+1
plt.pie(ratio, labels=labels, colors=colors, autopct = '%.1f%%', \
wedgeprops = {'width' : 1, 'edgecolor' : 'grey', 'linewidth' : 3}, textprops={'size' : 30, 'color' : 'black'} )
plt.title('< Recent Aerosol Types : '+data_recent.index[0].strftime("%Y%m%d")+' ~ '+data_recent.index[-1].strftime("%Y%m%d")+' >', fontsize=20, fontweight='bold')
fig3.savefig(output_dir + site + '_-_piechart_'+date_today.strftime("%Y%m%d")+'.png', dpi=300)
plt.close("all")
except:
fig, ax = plt.subplots(figsize = (12,12))
plt.xlim(-0.5,2.51)
plt.xticks(np.arange(-0.5,2.501,0.5),fontsize=15)
plt.ylim(0,3)
plt.yticks(np.arange(0,3.01,1),fontsize=15)
plt.xlabel('SAE 440-870nm', fontsize=25)
plt.ylabel('AAE 440-870nm', fontsize=25)
plt.title('< Aerosol Type Classification [ '+site+' ] >', fontsize=20, fontweight='bold')
#draw lines
plt.plot([1,1],[0,1],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([-0.5,2.5],[1,1],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([-0.5,2.5],[1.5,1.5],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([1.5,1.5],[1.5,3],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([1.5,2.5],[1.5,1.5],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([1.5,2.5],[2,2],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([-0.5,0],[2,2],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([0,0],[2,3],'-',color='grey',linewidth=1,alpha = 0.8)
plt.plot([1,1.5],[1,1.5],'-',color='grey',linewidth=1,alpha = 0.8)
#type names
plt.text(-0.1,0.45,'Uncertain', fontsize=30)
plt.text(-0.28,0.33,'large and low absorbing', fontsize=18, color='grey')
plt.text(-0.4,2.43,'Dust', fontsize=30)
plt.text(1.63,0.45,' NA', fontsize=30)
plt.text(1.13,0.33,' small and low (non) absorbing', fontsize=18, color='grey')
plt.text(1.88,1.18,'BC', fontsize=30)
plt.text(1.7,1.06,'black carbon', fontsize=18, color='grey')
plt.text(1.7,1.68,'BC + BrC', fontsize=30)
plt.text(1.88,2.43,'BrC', fontsize=30)
plt.text(1.72,2.31,'brown carbon', fontsize=18, color='grey')
plt.text(-0.1,1.2,'Uncertain / BC', fontsize=30)
plt.text(0.3,2.2,'Dust / BC + BrC', fontsize=30)
#fmf criteria
plt.text(-0.1,1.06,'FMF<0.6', fontsize=12, color='blue')
plt.text(0.68,1.06,'FMF>0.6', fontsize=12, color='blue')
plt.text(-0.1,1.15,'-------------------------- --------', fontsize=16, color='blue')
plt.text(0.3,2.06,'FMF<0.4', fontsize=12, color='blue')
plt.text(0.75,2.06,'FMF>0.4', fontsize=12, color='blue')
plt.text(0.3,2.15,'------------ -------------------------', fontsize=16, color='blue')
fig.savefig(output_dir + site + '_-_criteria_'+date_today.strftime("%Y%m%d")+'.png', dpi=300)
fig2, ax2 = plt.subplots(figsize = (6,12))
plt.xlim(0,2.8)
plt.ylim(0,4)
plt.text(1.2,2.8,'< Aerosol Types >', fontsize=20, fontweight='bold')
plt.scatter(0.8,2.5, c='gold',s=300, edgecolor='red')
plt.text(1.1,2.47,'Dust', fontsize=20)
plt.scatter(0.8,2.3, c='dimgray',s=300, edgecolor='red')
plt.text(1.1,2.27,'Black Carbon', fontsize=20)
plt.scatter(0.8,2.1, c='tan',s=300, edgecolor='red')
plt.text(1.1,2.07,'Black + Brown Carbon', fontsize=20)
plt.scatter(0.8,1.9, c='brown',s=300, edgecolor='red')
plt.text(1.1,1.87,'Brown Carbon', fontsize=20)
plt.scatter(0.8,1.7, c='hotpink',s=300, edgecolor='red')
plt.text(1.1,1.67,'Non-Absorbing', fontsize=20)
plt.scatter(0.8,1.5, c='skyblue',s=300, edgecolor='red')
plt.text(1.1,1.47,'Uncertain', fontsize=20)
plt.scatter(0.8,1.2, c='white',s=300, edgecolor='red')
plt.text(1.1,1.17,'Low (AOD440 < 0.4)', fontsize=20)
plt.scatter(0.73,1.0, c='red',s=300, marker=1)
plt.text(1.1,0.97,'Daily Std', fontsize=20)
plt.axis(False)
fig2.savefig(output_dir + site + '_-_legend_'+date_today.strftime("%Y%m%d")+'.png', dpi=300)
fig3, ax3 = plt.subplots(figsize = (12,12))
plt.xlim(0,2)
plt.ylim(0,2)
plt.axis(False)
plt.text(0.5,1,'No Data Available', fontsize=30)
fig3.savefig(output_dir + site + '_-_piechart_'+date_today.strftime("%Y%m%d")+'.png', dpi=300)
plt.close("all")
# %%
gen_image = PythonOperator(
task_id="gen_image",
python_callable=gen_image,
op_kwargs={
"finep_url": Variable.get("finep_url"),
"year": "{{ execution_date.year }}",
"month": "{{ execution_date.month }}",
"day": "{{ execution_date.day }}"
},
execution_timeout = datetime.timedelta(minutes=30),
retries = 5, # 최대 재시도 횟수
retry_delay = datetime.timedelta(minutes=1), # 재시도 간격
dag=dag,
)
gen_image
@practice
Copy link
Author

practice commented Jan 2, 2024

https://gist.github.com/practice/25e6080e14057cbce7fd2d2cf656892e 에서 이어 가겠습니다.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment