Note
Click here to download the full example code
Hail Nowcast (Hong Kong)
This example demonstrates a rule-based hail nowcast for the next 30 minutes, using data from Hong Kong.
Setup
Import all required modules and methods:
import os
import numpy as np
import pandas as pd
import xarray as xr
from pyresample import utils
import cartopy.feature as cf
import cartopy.crs as ccrs
from cartopy.io import shapereader
import matplotlib.pyplot as plt
import matplotlib
from matplotlib.colors import ListedColormap, BoundaryNorm
from swirlspy.qpf import rover
from swirlspy.rad import read_iris_grid, calc_vil
from swirlspy.qpe.utils import timestamps_ending, locate_file
from swirlspy.core.resample import grid_resample
from swirlspy.object import get_labeled_frame, fit_ellipse
from swirlspy.utils import standardize_attr, FrameType
matplotlib.use('agg')
Define the working directories:
# working_dir = os.path.join(os.getcwd(), 'swirlspy/examples')
working_dir = os.getcwd()
radar_dir = os.path.abspath(
os.path.join(working_dir, '../tests/samples/iris/3d/')
)
Define the basemap:
# define the plot function
def plot_base(ax: plt.Axes, extents: list, crs: ccrs.Projection):
# fake the ocean color
ax.imshow(
np.tile(np.array([[[178, 208, 254]]], dtype=np.uint8), [2, 2, 1]),
origin='upper', transform=crs,
extent=extents, zorder=-1
)
# coastline, province and state, color
ax.add_feature(
map_with_province, facecolor=cf.COLORS['land'],
edgecolor='none', zorder=0
)
# overlay coastline, province and state without color
ax.add_feature(
map_with_province, facecolor='none', edgecolor='gray', linewidth=0.5
)
ax.set_title('')
# Logging
start_time = pd.Timestamp.now()
Loading radar data
# Specify the basetime
basetime = pd.Timestamp('201403301930')
# Generate timestamps for the current and past 6 minute
# radar scan
timestamps = timestamps_ending(
basetime,
duration=pd.Timedelta(6, 'm'),
exclude_end=False
)
# Locating the files
located_files = []
for timestamp in timestamps:
located_files.append(locate_file(radar_dir, timestamp))
# Reading the radar data
reflectivity_list = []
for filename in located_files:
reflectivity = read_iris_grid(filename)
reflectivity_list.append(reflectivity)
# Standardize reflectivity xarrays
raw_frames = xr.concat(reflectivity_list,
dim='time').sortby(['y'], ascending=False)
standardize_attr(raw_frames, frame_type=FrameType.dBZ)
initialising_time = pd.Timestamp.now()
Traceback (most recent call last):
File "/tmp/build/docs/swirlspy/swirlspy/examples/hail.py", line 98, in <module>
reflectivity = read_iris_grid(filename)
File "/tmp/build/docs/swirlspy/swirlspy/rad/_iris.py", line 407, in read_iris_grid
raise ValueError("Invalid file") from e
ValueError: Invalid file
Identify regions of interest for hail
Hail has a chance of occurring when two phenomena co-happen:
The 58 dBZ echo top exceeds 3 km.
The Vertically Integrated Liquid (VIL) up to 2 km is less than 5mm.
# getting radar scan for basetime
reflectivity = raw_frames.sel(time=basetime)
# maximum reflectivity at elevations > 3km
ref_3km = reflectivity.sel(
height=slice(3000, None)
).max(dim='height', keep_attrs=True)
# 58 dBZ echo top exceeds 3km
cond_1 = ref_3km > 58
# vil up to 2km
vil_2km = calc_vil(reflectivity.sel(height=slice(1000, 2000)))
# VIL up 2km is less than 5mm
cond_2 = vil_2km < 5
# Region of interest for hail
hail = xr.ufuncs.logical_and(cond_1, cond_2)
The identified regions are then labeled and fitted with minimum enclosing ellipses.
# label image
# image is binary so any threshold between 0 and 1 works
# define minimum size as 4e6 m^2 or 4 km^2
labeled_hail, uids = get_labeled_frame(hail, 0.5, min_size=4e6)
# fit ellipses to regions of interest
ellipse_list = []
for uid in uids:
ellipse, _ = fit_ellipse(labeled_hail == uid)
ellipse_list.append(ellipse)
identify_time = pd.Timestamp.now()
Extrapolation of hail region
First, we obtain the xarray.DataArray to generate the motion field.
In this example, we generate the motion field from two consecutive 3km CAPPI radar scans closest to basetime.
# Select radar data at 3km
frames = raw_frames.sel(height=3000).drop('height')
Obtain the motion field by ROVER
# ROVER
motion = rover(frames)
motion_time = pd.Timestamp.now()
Next, we extract the motion vector at the centroid of each ellipse, and calculate the displacement of the ellipse after 30 minutes. Since the distance is expressed in pixels, we need to convert the distance of the motion vector to grid units (in this case, meters).
# getting meters per pixel
area_def = reflectivity.attrs['area_def']
x_d = area_def.pixel_size_x
y_d = area_def.pixel_size_y
# time ratio
# time ratio between nowcast interval and unit time
time_ratio = pd.Timedelta(30, 'm') / pd.Timedelta(6, 'm')
ellipse30_list = []
for ellipse in ellipse_list:
# getting motion in pixels/6 minutes
pu = motion[0].sel(x=ellipse['center'][0],
y=ellipse['center'][1],
method='nearest')
pv = motion[1].sel(x=ellipse['center'][0],
y=ellipse['center'][1],
method='nearest')
# converting to meters / 6 minutes
u = x_d * pu
v = y_d * pv
# get displacement in 30 minutes
dcenterx = u * time_ratio
dcentery = v * time_ratio
# get new position of ellipse after 30 minutes
x30 = ellipse['center'][0] + dcenterx
y30 = ellipse['center'][1] + dcentery
# get new ellipse, only the center is changed
ellipse30 = ellipse.copy()
ellipse30['center'] = (x30, y30)
ellipse30_list.append(ellipse30)
extrapolate_time = pd.Timestamp.now()
Visualisation
# Defining figure
fig = plt.figure(figsize=(12, 6.5))
# number of rows and columns in plot
nrows = 2
ncols = 1
# Defining the crs
crs = area_def.to_cartopy_crs()
# Load the shape of Hong Kong
map_shape_file = os.path.abspath(os.path.join(
working_dir,
'../tests/samples/shape/hk_tms_aeqd.shp'
))
# coastline and province
map_with_province = cf.ShapelyFeature(
list(shapereader.Reader(map_shape_file).geometries()),
area_def.to_cartopy_crs()
)
# Defining extent
x = labeled_hail.coords['x'].values
y = labeled_hail.coords['y'].values
x_d = x[1] - x[0]
y_d = y[1] - y[0]
extents = [x[0], y[0], x[-1], y[-1]]
# Defining ellipse patches
def gen_patches(lst, ls='-'):
patch_list = []
for ellipse in lst:
patch = matplotlib.patches.Arc(
ellipse['center'],
ellipse['b'] * 2,
ellipse['a'] * 2,
angle=ellipse['angle'],
ls=ls
)
patch_list.append(patch)
return patch_list
# 1. Plotting maximum reflectivity above 3km
# Define color scheme
# Defining colour scale and format.
levels = [
-32768, 10, 15, 20, 24, 28, 32, 34, 38, 41,
44, 47, 50, 53, 56, 58, 60, 62
]
cmap = ListedColormap([
'#FFFFFF00', '#08C5F5', '#0091F3', '#3898FF', '#008243', '#00A433',
'#00D100', '#01F508', '#77FF00', '#E0D100', '#FFDC01', '#EEB200',
'#F08100', '#F00101', '#E20200', '#B40466', '#ED02F0'
])
norm = BoundaryNorm(levels, ncolors=cmap.N, clip=True)
ax = fig.add_subplot(ncols, nrows, 1, projection=crs)
plot_base(ax, extents, crs)
ref_3km.where(ref_3km > levels[1]).plot(
ax=ax,
cmap=cmap,
norm=norm,
extend='max',
cbar_kwargs={
'ticks': levels[1:],
'format': '%.3g',
'fraction': 0.046,
'pad': 0.04
}
)
patches = gen_patches(ellipse_list)
for patch in patches:
ax.add_patch(patch)
patches = gen_patches(ellipse30_list, ls='--')
for patch in patches:
ax.add_patch(patch)
ax.set_title('MAX_REF >= 3KM')
# 2. Plotting VIL up to 2km
levels = [
0, 0.05,
0.1, 0.2, 0.4, 0.6, 0.8, 1,
2, 4, 6, 8, 15, 20,
25, 30, 32, 34
]
cmap = matplotlib.colors.ListedColormap([
'#FFFFFF', '#08C5F5', '#0091F3', '#3898FF', '#008243', '#00A433',
'#00D100', '#01F508', '#77FF00', '#E0D100', '#FFDC01', '#EEB200',
'#F08100', '#F00101', '#E20200', '#B40466', '#ED02F0'
])
norm = BoundaryNorm(levels, ncolors=cmap.N, clip=True)
ax = fig.add_subplot(ncols, nrows, 2, projection=crs)
plot_base(ax, extents, crs)
vil_2km.where(vil_2km > levels[1]).plot(
cmap=cmap,
norm=norm,
ax=ax,
extend='max',
cbar_kwargs={
'ticks': levels[1:],
'format': '%.3g',
'fraction': 0.046,
'pad': 0.04
}
)
ax.set_title('VIL <= 2km')
patches = gen_patches(ellipse_list)
for patch in patches:
ax.add_patch(patch)
patches = gen_patches(ellipse30_list, ls='--')
for patch in patches:
ax.add_patch(patch)
suptitle1 = "Hail Nowcast Tracks"
suptitle2 = basetime.strftime('%Y-%m-%d')
suptitle3 = (f"Based @ {basetime.strftime('%H:%MH')}\n"
f"Valid @ {(basetime + pd.Timedelta(30, 'm')).strftime('%H:%MH')}")
fig.text(0., 0.93, suptitle1, va='top', ha='left', fontsize=16)
fig.text(0.57, 0.93, suptitle2, va='top', ha='center', fontsize=16)
fig.text(0.90, 0.93, suptitle3, va='top', ha='right', fontsize=16)
plt.tight_layout()
plt.savefig(
working_dir +
f"/../tests/outputs/hail.png",
dpi=300
)
visualise_time = pd.Timestamp.now()
Checking run time of each component
print(f"Start time: {start_time}")
print(f"Initialising time: {initialising_time}")
print(f"Identify time: {identify_time}")
print(f"Motion field time: {motion_time}")
print(f"Extrapolate time: {extrapolate_time}")
print(f"Visualise time: {visualise_time}")
print(f"Time to initialise: {initialising_time - start_time}")
print(f"Time to identify hail regions: {identify_time - initialising_time}")
print(f"Time to generate motion field: {motion_time - identify_time}")
print(f"Time to extrapolate: {extrapolate_time - motion_time}")
print(f"Time to visualise: {visualise_time - extrapolate_time}")
print(f"Total: {visualise_time - start_time}")
Total running time of the script: ( 0 minutes 0.006 seconds)