def test_remove_nan_observations(test_coords):
r"""Test remove observations equal to nan function."""
x, y = test_coords[0], test_coords[1]
z = np.array([np.nan, np.nan, np.nan, 1, 1, 1, 1, 1, 1, 1])
x_, y_, z_ = remove_nan_observations(x, y, z)
truthx = np.array([10, 52, 53, 98, 34, 15, 58])
truthy = np.array([94, 98, 66, 14, 24, 60, 16])
truthz = np.array([1, 1, 1, 1, 1, 1, 1])
assert_array_almost_equal(truthx, x_)
assert_array_almost_equal(truthy, y_)
assert_array_almost_equal(truthz, z_)
def test_remove_nan_observations(test_coords):
r"""Test remove observations equal to nan function."""
x, y = test_coords[0], test_coords[1]
z = np.array([np.nan, np.nan, np.nan, 1, 1, 1, 1, 1, 1, 1])
x_, y_, z_ = remove_nan_observations(x, y, z)
truthx = np.array([10, 52, 53, 98, 34, 15, 58])
truthy = np.array([94, 98, 66, 14, 24, 60, 16])
truthz = np.array([1, 1, 1, 1, 1, 1, 1])
assert_array_almost_equal(truthx, x_)
assert_array_almost_equal(truthy, y_)
assert_array_almost_equal(truthz, z_)
def crtanje_m():
to_proj = ccrs.AlbersEqualArea(central_longitude=-1., central_latitude=10.)
#load cordinates
fname = '/home/martin/Master_rad/PredtandfilaGrid.dat'
#col_names = ['index','lon','lat','country','altitude'] ovo koristimo ako nemama definisane imena kolona
#load temp
df = pd.read_fwf(fname, na_values='MM')
lon = df['lon'].values
lat = df['lat'].values
xp, yp, _ = to_proj.transform_points(ccrs.Geodetic(), lon, lat).T
data1 = pd.read_csv('/home/martin/Master_rad/CARPATGRID_TA_M.ser',
sep='\s+')
y = int(input('Unesite godinu: ' ' '))
m = int(input('Unesite mesec: ' ' '))
x1 = data1.loc[y, m]
x_masked, y_masked, t = remove_nan_observations(xp, yp, x1.values)
tempx, tempy, temp = interpolate(x_masked,
y_masked,
t,
interp_type='barnes',
minimum_neighbors=8,
search_radius=150000,
hres=30000)
temp = np.ma.masked_where(np.isnan(temp), temp)
levels = list(range(-20, 20, 1))
cmap = plt.get_cmap('viridis')
norm = BoundaryNorm(levels, ncolors=cmap.N, clip=True)
fig = plt.figure(figsize=(20, 10))
view = fig.add_subplot(1, 1, 1, projection=to_proj)
view.set_extent([27.0, 16.9, 49.5, 44.5])
view.add_feature(cfeature.STATES.with_scale('50m'))
view.add_feature(cfeature.OCEAN)
view.add_feature(cfeature.COASTLINE.with_scale('50m'))
view.add_feature(cfeature.BORDERS, linestyle=':')
mmb = view.pcolormesh(tempx, tempy, temp, cmap=cmap, norm=norm)
fig.colorbar(mmb, shrink=.4, pad=0.02, boundaries=levels)
view.set_title('Srednja temperatura')
plt.show()
usecols=(2, 3, 4, 5, 18, 19),
names=[
'latitude', 'longitude', 'slp', 'temperature',
'wind_dir', 'wind_speed'
],
na_values=-99999)
###########################################
# Project the lon/lat locations to our final projection
lon = data['longitude'].values
lat = data['latitude'].values
xp, yp, _ = to_proj.transform_points(ccrs.Geodetic(), lon, lat).T
###########################################
# Remove all missing data from pressure
x_masked, y_masked, pres = remove_nan_observations(xp, yp, data['slp'].values)
###########################################
# Interpolate pressure using Cressman interpolation
slpgridx, slpgridy, slp = interpolate(x_masked,
y_masked,
pres,
interp_type='cressman',
minimum_neighbors=1,
search_radius=400000,
hres=100000)
##########################################
# Get wind information and mask where either speed or direction is unavailable
wind_speed = (data['wind_speed'].values * units('m/s')).to('knots')
wind_dir = data['wind_dir'].values * units.degree
print(e)
return None
return lon, lat, value
from_proj = ccrs.Geodetic()
to_proj = ccrs.AlbersEqualArea(central_longitude=-97.0000, central_latitude=38.0000)
levels = list(range(-20, 20, 1))
cmap = plt.get_cmap('magma')
norm = BoundaryNorm(levels, ncolors=cmap.N, clip=True)
x, y, temp = station_test_data('air_temperature', from_proj, to_proj)
x, y, temp = remove_nan_observations(x, y, temp)
x, y, temp = remove_repeat_coordinates(x, y, temp)
###########################################
# Scipy.interpolate linear
# ------------------------
gx, gy, img = interpolate(x, y, temp, interp_type='linear', hres=75000)
img = np.ma.masked_where(np.isnan(img), img)
view = basic_map(to_proj)
mmb = view.pcolormesh(gx, gy, img, cmap=cmap, norm=norm)
plt.colorbar(mmb, shrink=.4, pad=0, boundaries=levels)
###########################################
# Natural neighbor interpolation (MetPy implementation)
# -----------------------------------------------------
# `Reference <https://github.com/Unidata/MetPy/files/138653/cwp-657.pdf>`_
def plot_map(df,
date,
value,
cbarlabel='',
title='',
cmap="Spectral_r",
zmin=0,
zmax=100,
save=False):
cities = [
'Williston', 'Minot', 'Grand Forks', 'Fargo', 'Bismarck', 'Devils Lake'
]
fig, ax = plt.subplots(figsize=(15, 10))
to_plot = df[df["Date"] == date]
lon = to_plot['Longitude (deg)'].values
lat = to_plot['Latitude (deg)'].values
#make basemap map
print("Making map...")
m = Basemap(llcrnrlon=min(lon),
llcrnrlat=min(lat),
urcrnrlon=max(lon),
urcrnrlat=max(lat),
projection='merc',
resolution='h')
#Convert lon, lat to map coordinates and mask any missing points
xp, yp = m(lon, lat)
x_masked, y_masked, z = remove_nan_observations(xp, yp,
to_plot[value].values)
print("Interpolating data...")
#Uncomment one of the interpolation methods below. rbf is chosen as a default
#linear
#gridx, gridy, z = interpolate(x_masked, y_masked, z, interp_type='linear', hres=5000)
#rbf
gridx, gridy, z = interpolate(x_masked,
y_masked,
z,
interp_type='rbf',
hres=5000,
rbf_func='linear',
rbf_smooth=0)
#cressman
#gridx, gridy, z = interpolate(x_masked, y_masked, z, interp_type='cressman', minimum_neighbors=1, hres=5000,
#search_radius=100000)
#barnes
#gridx, gridy, z = interpolate(x_masked, y_masked, z, interp_type='barnes', hres=5000,
#search_radius=100000)
#Natural neighbor
#gridx, gridy, z = interpolate(x_masked, y_masked, z, interp_type='natural_neighbor', hres=5000)
z = np.ma.masked_where(np.isnan(z), z)
#Find max and min of z if not input manually
if zmax == 'max':
zmax = np.max(z)
if zmin == 'min':
zmin = np.min(z)
#normalize colormap
cmap = plt.get_cmap(cmap)
norm = Normalize(vmin=zmin, vmax=zmax)
#Draw map lines
m.drawcoastlines()
m.drawcountries()
m.drawstates()
#Set contour line parameters. must be integers
low = int(round(np.min(z)))
high = int(round(np.max(z)))
step = int(round((high - low) / 10))
#Set step at least 2 so there are not contour lines set for every increase of 1 (unless that's what you want)
if step < 2:
step = 2
#plot colormesh
mmb = m.pcolormesh(gridx, gridy, z, cmap=cmap, norm=norm)
cbar = m.colorbar(mmb, location='bottom')
cs = m.contour(gridx,
gridy,
z,
colors='k',
levels=list(range(low, high, step)),
alpha=0.3)
plt.clabel(cs, inline=1, fontsize=12, fmt='%i')
#Plot city markers
geolocator = Nominatim()
for city in cities:
loc = geolocator.geocode(city)
x, y = m(loc.longitude, loc.latitude)
m.plot(x,
y,
marker='o',
color='None',
markeredgecolor='k',
markersize=8)
plt.text(x + 10000, y + 5000, city, fontsize=12, alpha=0.7)
m.drawcounties()
cbar.set_label(cbarlabel)
plt.title(title, fontsize=18)
if save:
# Make sure a directory exists for the gdd data
print("Checking for file pathways...")
if not os.path.isdir("./maps"):
os.mkdir('maps')
print("Saving figure...")
fig.savefig('./maps/' + str(title) + '.png', format='png')
return m
if proj_from is not None and proj_to is not None:
proj_points = proj_to.transform_points(proj_from, lon, lat)
return proj_points[:, 0], proj_points[:, 1], value
return lon, lat, value
###########################################
# Get pressure information using the sample station data
xp, yp, pres = station_test_data(['slp'], from_proj, to_proj)
###########################################
# Remove all missing data from pressure
pres = np.array([p[0] for p in pres])
xp, yp, pres = remove_nan_observations(xp, yp, pres)
###########################################
# Interpolate pressure as usual
slpgridx, slpgridy, slp = interpolate(xp,
yp,
pres,
interp_type='cressman',
minimum_neighbors=1,
search_radius=400000,
hres=100000)
###########################################
# Get wind information
x, y, wind = station_test_data(['wind_speed', 'wind_dir'], from_proj, to_proj)
if proj_from is not None and proj_to is not None:
proj_points = proj_to.transform_points(proj_from, lon, lat)
return proj_points[:, 0], proj_points[:, 1], value
return lon, lat, value
###########################################
# Get pressure information using the sample station data
xp, yp, pres = station_test_data(['slp'], from_proj, to_proj)
###########################################
# Remove all missing data from pressure
pres = np.array([p[0] for p in pres])
xp, yp, pres = remove_nan_observations(xp, yp, pres)
###########################################
# Interpolate pressure as usual
slpgridx, slpgridy, slp = interpolate(xp, yp, pres, interp_type='cressman',
minimum_neighbors=1, search_radius=400000, hres=100000)
###########################################
# Get wind information
x, y, wind = station_test_data(['wind_speed', 'wind_dir'], from_proj, to_proj)
###########################################
# Remove bad data from wind information
wind_speed = np.array([w[0] for w in wind])
wind_dir = np.array([w[1] for w in wind])
print(e)
return None
return lon, lat, value
from_proj = ccrs.Geodetic()
to_proj = ccrs.AlbersEqualArea(central_longitude=-97.0000, central_latitude=38.0000)
levels = list(range(-20, 20, 1))
cmap = plt.get_cmap('magma')
norm = BoundaryNorm(levels, ncolors=cmap.N, clip=True)
x, y, temp = station_test_data('air_temperature', from_proj, to_proj)
x, y, temp = remove_nan_observations(x, y, temp)
x, y, temp = remove_repeat_coordinates(x, y, temp)
###########################################
# Scipy.interpolate linear
# ------------------------
gx, gy, img = interpolate(x, y, temp, interp_type='linear', hres=75000)
img = np.ma.masked_where(np.isnan(img), img)
view = basic_map(to_proj)
mmb = view.pcolormesh(gx, gy, img, cmap=cmap, norm=norm)
plt.colorbar(mmb, shrink=.4, pad=0, boundaries=levels)
###########################################
# Natural neighbor interpolation (MetPy implementation)
# -----------------------------------------------------
# `Reference <https://github.com/Unidata/MetPy/files/138653/cwp-657.pdf>`_
