""" Stagnation MERRA Timeseries """

# REGIONAL

# Part 2.0: Random forest. Preparation of data frame.

# In this analysis we'll attempt to gather the most relevant meteorological variables (eulerean) that affect the concentrations of PM2.5 in the NEUS during the summer. We speculate that random forest analysis we'll give us insights into this problem. We'll sample MERRA gridpoints in the NEUS and make a point-by-point data frame.

# Load packages, tools, and stuff:

import glob

import pandas as pd

import numpy as np

import matplotlib.pyplot as plt

import shapely.geometry as sgeom

import cartopy

import cartopy.crs as ccrs

import cartopy.io.shapereader as shpreader

from cartopy.mpl.gridliner import LONGITUDE_FORMATTER, LATITUDE_FORMATTER

from cartopy.feature import NaturalEarthFeature, COASTLINE, LAKES

from scipy.interpolate import RectSphereBivariateSpline as RSBS

from netCDF4 import Dataset

# Directory paths to data and analysis directories:

PATH_AOD = '/Users/Rola/Documents/Science/JHU/DATA/MERRAero/'

PATH_MET = '/Users/Rola/Documents/Science/JHU/DATA/MERRA_meteorology/'

PATH_RES = '/Users/Rola/Documents/Science/JHU/ANALYSIS/Stagnation/'

# Load PM2.5 mean file with PANDAS:

PM = pd.read_csv(PATH_AOD+'PM25_neus_mean_series.csv', parse_dates=[0])

PM = PM.set_index('date')

date = pd.DatetimeIndex(PM.index.values)

# ----------------------------------------------------------------------

# Locate gridpoint inside US

fn = Dataset(PATH_AOD+'GOCART_output_20020701_daily_average.nc4',mode='r')

lona = fn.variables['Longitude'][:]

lata = fn.variables['Latitude' ][:]

lata = lata.T

lona = lona.T

fn.close()

mask = (lona>-125) & (lona<-65) & (lata>24) & (lata<50)

LN=lona[mask]

LT=lata[mask]

code = ('AL','AZ','AR','CA','CO','CT','DE','FL','GA','ID','IL','IN','IA','KS','KY','LA','ME','MD','MA','MI','MN','MS','MO','MT','NE','NV','NH','NJ','NM','NY','NC','ND','OH','OK','OR','PA','RI','SC','SD','TN','TX','UT','VT','VA','WA','WV','WI','WY')

cord = sgeom.MultiPoint(list(zip(LN,LT)))

shpfilename = shpreader.natural_earth(category='cultural', resolution='110m', name='admin_1_states_provinces_lakes')

states = shpreader.Reader(shpfilename).records()

CORD=np.empty([LN.size,2])*0.

k=0

for state in states:

name = state.attributes['postal']

if name in code:

print name

for i in range(0,LN.size):

if state.geometry.contains(cord[i]):

CORD[k,:]=[cord[i].x, cord[i].y]

k=k+1

CORD=zip((CORD[CORD[:,0]!=0,0]),(CORD[CORD[:,1]!=0,1]))

CORDS=pd.DataFrame(CORD)

CORDS=CORDS.drop_duplicates()

maskota = np.empty([361,576])*0

for i in range(0,CORDS.shape[0]):

# print i

aux=(lona==CORDS.ix[i,0]) & (lata==CORDS.ix[i,1])

maskota=maskota+aux

maskcount=maskota

maskota=maskota>=1

#-----------------------------------------------------------------------

# Creation of Data Frame

DAT=pd.DataFrame([])

iii=1

for i in date:

yr=i.year # Get date info from index in loop.

mn=i.month

dy=i.day

print yr, mn, dy, iii/982.*100

# Generate meteorology file names to be read.

hgtfname='MERRA*3d*%4.0f%02.0f%02.0f*.nc' %(yr,mn,dy)

sfcfname='MERRA*2d_slv*%4.0f%02.0f%02.0f*.nc' %(yr,mn,dy)

prefname='MERRA*2d_lnd*%4.0f%02.0f%02.0f*.nc' %(yr,mn,dy)

aodfname='GOCART*%4.0f%02.0f%02.0f*average.nc4' %(yr,mn,dy)

# UNIX-like ls of file names to be read. Finds file in directory.

files_hgt=glob.glob(PATH_MET+hgtfname)

files_sfc=glob.glob(PATH_MET+sfcfname)

files_pre=glob.glob(PATH_MET+prefname)

files_aod=glob.glob(PATH_AOD+aodfname)

# 500 hPa variables.

fn = Dataset(files_hgt[0],mode='r')

lon = fn.variables['longitude'][:]

lat = fn.variables['latitude'][:]

hgt = fn.variables['h'][:]

hgt = np.squeeze(hgt)/10

u = fn.variables['u'][:]

u5 = np.squeeze(u)

v = fn.variables['v'][:]

v5 = np.squeeze(v)

latp5=(90+lat)*np.pi/180

lonp5=(lon+180)*np.pi/180

lon5,lat5 = np.meshgrid(lonp5, latp5)

fn.close(); del u, v, lon, lat, fn

# Surface (2m) variables.

fn = Dataset(files_sfc[0],mode='r')

lon = fn.variables['longitude'][:]

lat = fn.variables['latitude' ][:]

u = fn.variables['u2m' ][:]

v = fn.variables['v2m' ][:]

t = fn.variables['t2m' ][:]

q = fn.variables['qv2m'][:]

s = fn.variables['slp' ][:]

u2 = np.squeeze(u)

v2 = np.squeeze(v)

t2 = np.squeeze(t)

q2 = np.squeeze(q)

s2 = np.squeeze(s)

u2 = u2[1:-1,1:-1]

v2 = v2[1:-1,1:-1]

t2 = t2[1:-1,1:-1]

q2 = q2[1:-1,1:-1]

s2 = s2[1:-1,1:-1]

latp2=(90+lat[1:-1])*np.pi/180

lonp2=(lon[1:-1]+180)*np.pi/180

fn.close(); del u, v, t, q, s, lon, lat, fn

# Precipitation flux.

fn = Dataset(files_pre[0],mode='r')

lon = fn.variables['longitude'][:]

lat = fn.variables['latitude' ][:]

prec = np.squeeze(fn.variables['prectot'][:])

prec = prec[1:-1,1:-1]

lonr,latr = np.meshgrid(lon,lat) # same shape as 2m variables

fn.close(); del lon, lat, fn

# MERRAero PM2.5 and AOD.

fn = Dataset(files_aod[0],mode='r')

lona = fn.variables['Longitude' ][:]

lata = fn.variables['Latitude' ][:]

AOD = fn.variables['TOTEXTTAU_avg'][:]

# 1e9 converts to micrograms per cubic meter.

# 1.375 factor converts SO4 to AmmSO4.

PM1 = fn.variables['DUSMASS25_avg'][:]*1e9

PM2 = fn.variables['SSSMASS25_avg'][:]*1e9

PM3 = fn.variables['SO4SMASS_avg' ][:]*1e9*1.375

PM4 = fn.variables['OCSMASS_avg' ][:]*1e9

PM5 = fn.variables['BCSMASS_avg' ][:]*1e9

PM25 = PM1+PM2+PM3+PM4+PM5 # By Definition from MERRAero

PM1 = PM1.T

PM2 = PM2.T

PM3 = PM3.T

PM4 = PM4.T

PM5 = PM5.T

PM25 = PM25.T

lata = lata.T

lona = lona.T

latai = (90+lata)*np.pi/180

lonai = (lona+180)*np.pi/180

fn.close(); del fn

clon=-98.5795 # central lat/lon for imaging. Dead-center of US.

clat= 39.8282

# plt.imshow(PM25); plt.show()

# ------------------------------------------------------------------

# Interpolates 144x288 500 hPa and 361x540) meteorology into 361x576

# MERRAero aerosol grid.

def interpol(latitude,longitude,variable):

return interpolated_vairable

u5i = interpol(latp5,lonp5,u5)

v5i = interpol(latp5,lonp5,v5)

hgti = interpol(latp5,lonp5,hgt)

prei = interpol(latp2,lonp2,prec)

u2i = interpol(latp2,lonp2,u2)

v2i = interpol(latp2,lonp2,v2)

t2i = interpol(latp2,lonp2,t2)

q2i = interpol(latp2,lonp2,q2)

s2i = interpol(latp2,lonp2,s2)

# ------------------------------------------------------------------

PER=pd.DataFrame.from_items([('u_500hPa',u5i[maskota]),('v_500hPa',v5i[maskota]),('hgt_500hPa',hgti[maskota]),('precip',prei[maskota]),('u_2m',u2i[maskota]),('v_2m',v2i[maskota]),('temp_2m',t2i[maskota]),('spec_humidity',q2i[maskota]),('slp',s2i[maskota]),('dust',PM1[maskota]),('sea_salt',PM2[maskota]),('AmmSO4',PM3[maskota]),('OC',PM4[maskota]),('BC',PM5[maskota]),('PM25',PM25[maskota]),('lat',lata[maskota]),('lon',lona[maskota])])

PER['date']=i

DAT=pd.concat([DAT,PER],axis=0)

del PER

iii=iii+1

DAT=DAT.set_index(DAT['date'])

DAT.drop('date', inplace=True, axis=1)

DAT.to_csv(PATH_RES+'Stagnation_random_forest_matrix_usa.csv', index=True)

# --- Stagnation Field Plot --------------------------------------

geo_axes = plt.subplot(111, projection=cartopy.crs.Miller())

clon=-98.5795 # central lat/lon for imaging. Dead-center of US.

clat= 39.8282

geo_axes.set_extent([clon-35,clon+35, clat-15, clat+15]) #US

#geo_axes.set_extent([-84,-66, 36, 48]) # NEUS

geo_axes.scatter(lona[maskota], lata[maskota], 10, transform=ccrs.PlateCarree(), marker='o',facecolor='k', edgecolors='k')

COUNTRIES = NaturalEarthFeature(category='cultural', scale='10m', facecolor='none', name='admin_0_countries_lakes')

geo_axes.add_feature(COUNTRIES, linewidth=0.5)

shpfilename = shpreader.natural_earth(category='cultural', resolution='10m', name='admin_1_states_provinces_lakes')

states = shpreader.Reader(shpfilename).records()

code = ('MD','VA','DE','NJ','PA','WV','RI','VT','NH','NY','MA','ME','CT')

for state in states:

name = state.attributes['postal']

if name in code:

geo_axes.add_geometries(state.geometry, ccrs.PlateCarree(), facecolor='none', edgecolor='blue', linewidth=1)

plt.show()

#-----------------------------------------------------------------