In [1]:
from svrimg.utils.get_tables import get_index_table, get_svrgis_table, get_geog
from svrimg.utils.get_images import get_example
df_indexer = get_index_table("all", data_dir="../data/csvs/")
df_svrgis = get_svrgis_table(data_dir="../data/csvs/")
geog = get_geog(data_dir="../data/geog/")
gridrad = get_example()
df_indexer is a file that defines attributes of each image in the dataset. See http://svrimg.org/raw_data.html¶
In [2]:
df_indexer.head()
Out[2]:
df_svrgis is a file that contains all of the SVRGIS attributes. Both files can be joined using the index column.¶
In [3]:
df_svrgis.head()
Out[3]:
geog is the grid information that will help us place the images correctly¶
In [4]:
geog
Out[4]:
gridrad is interpolated gridrad data from April 27th at 1900 UTC. See http://gridrad.org.¶
In [5]:
gridrad
Out[5]:
In [6]:
import matplotlib.pyplot as plt
import numpy as np
%matplotlib inline
plt.rcParams['figure.figsize'] = 10, 10
from svrimg.utils.map_helper import radar_colormap
from matplotlib.colors import BoundaryNorm
import cartopy.crs as ccrs
import cartopy.feature as cfeature
ax = plt.axes(projection=ccrs.LambertConformal(central_longitude=-95))
ax.set_extent([-110, -70, 24, 48])
ax.coastlines()
ax.add_feature(cfeature.STATES)
cmap = radar_colormap()
classes = np.array(list(range(0, 85, 5)))
norm = BoundaryNorm(classes, ncolors=cmap.N)
ax.pcolormesh(geog.CLONG.squeeze(), geog.CLAT.squeeze(), gridrad.REFC_MAX.squeeze(),
norm=norm, cmap=cmap, transform=ccrs.PlateCarree())
Out[6]:
Get a subset of the data where the radar time was '2011-04-27 19:00:00'¶
In [7]:
df_subset = df_indexer[df_indexer.radar_time=='4/27/2011 19:00']
df_subset
Out[7]:
Use the image request utility to download the images associated with the unids and save them in "../data/tor"¶
In [8]:
from svrimg.utils.get_images import request_images
info = request_images(df_subset.index.values, "../data/tor")
info
Out[8]:
Read in the image¶
In [9]:
from svrimg.utils.get_images import read_image
uid = "201104271836z000303011"
fname = info[uid]
im = read_image(fname)
plt.imshow(np.flipud(im), cmap=cmap, norm=norm)
Out[9]:
In [10]:
row = df_indexer.loc[uid]
blank = np.zeros(shape=(899, 1399))
blank[row.ymin:row.ymax+1, row.xmin:row.xmax+1] = im
plt.imshow(np.flipud(blank), cmap=cmap, norm=norm)
Out[10]:
You can see how it fits in the original image by subtracting it from the original data.¶
In [11]:
plt.rcParams['figure.figsize'] = 10, 10
ax = plt.axes(projection=ccrs.LambertConformal(central_longitude=-95))
ax.set_extent([-110, -70, 24, 48])
ax.coastlines()
ax.add_feature(cfeature.STATES)
cmap = radar_colormap()
classes = np.array(list(range(0, 85, 5)))
norm = BoundaryNorm(classes, ncolors=cmap.N)
ax.pcolormesh(geog.CLONG.squeeze(), geog.CLAT.squeeze(), gridrad.REFC_MAX.squeeze() - blank,
norm=norm, cmap=cmap, transform=ccrs.PlateCarree())
Out[11]:
In [12]:
plt.rcParams['figure.figsize'] = 10, 10
ax = plt.axes(projection=ccrs.LambertConformal(central_longitude=-95))
ax.set_extent([-110, -70, 24, 48])
ax.coastlines()
ax.add_feature(cfeature.STATES)
uid = "201104271915z000300950"
fname = info[uid]
im = read_image(fname)
row = df_indexer.loc[uid]
blank = np.zeros(shape=(899, 1399))
blank[row.ymin:row.ymax+1, row.xmin:row.xmax+1] = im
ax.pcolormesh(geog.CLONG.squeeze(), geog.CLAT.squeeze(), gridrad.REFC_MAX.squeeze() - blank,
norm=norm, cmap=cmap, transform=ccrs.PlateCarree())
Out[12]:
Combine into one function¶
In [13]:
from svrimg.utils.get_images import geo_read_image
plt.rcParams['figure.figsize'] = 10, 10
ax = plt.axes(projection=ccrs.LambertConformal(central_longitude=-95))
ax.set_extent([-88, -83, 33, 38])
ax.coastlines()
ax.add_feature(cfeature.STATES)
vals = geo_read_image(df_indexer, info, "201104271915z000300950")
ax.pcolormesh(geog.CLONG.squeeze(), geog.CLAT.squeeze(), vals,
norm=norm, cmap=cmap, transform=ccrs.PlateCarree())
Out[13]:
