def _do_month(month, axes, data, in_sts, in_ets, kwargs):
"""Place data on this axes"""
axes.get_xaxis().set_visible(False)
axes.get_yaxis().set_visible(False)
pos = axes.get_position()
ndcheight = (pos.y1 - pos.y0)
ndcwidth = (pos.x1 - pos.x0)
fitbox(plt.gcf(),
month.strftime("%B %Y"),
pos.x0,
pos.x1,
pos.y1,
pos.y1 + 0.028,
ha='center')
axes.add_patch(
Rectangle((0., 0.90),
1,
0.1,
zorder=2,
facecolor='tan',
edgecolor='tan'))
sts = datetime.date(month.year, month.month, 1)
ets = (sts + datetime.timedelta(days=35)).replace(day=1)
calendar.setfirstweekday(calendar.SUNDAY)
weeks = len(calendar.monthcalendar(month.year, month.month))
now = sts
row = 0
dy = 0.9 / float(weeks)
dx = 1. / 7.
for i, dow in enumerate(['SUN', 'MON', 'TUE', 'WED', 'THU', 'FRI', 'SAT']):
axes.text(1. / 7. * (i + 0.5),
0.94,
dow,
fontsize=fontscale(ndcwidth / 7. * 0.4),
ha='center',
va='center')
while now < ets:
# Is this Sunday?
if now.weekday() == 6 and now != sts:
row += 1
if now < in_sts or now > in_ets:
now += datetime.timedelta(days=1)
continue
offx = (now.weekday() + 1) if now.weekday() != 6 else 0
axes.text(offx * dx + 0.01,
0.9 - row * dy - 0.01,
str(now.day),
fontsize=fontscale(ndcheight / 5. * 0.25),
color='tan',
va='top')
_do_cell(axes, now, data, row, dx, dy, kwargs)
now += datetime.timedelta(days=1)
def calendar_plot(sts, ets, data, **kwargs):
"""Create a plot that looks like a calendar
Args:
sts (datetime.date): start date of this plot
ets (datetime.date): end date of this plot (inclusive)
data (dict[dict]): dictionary with keys of dates and dicts for
`val` value and optionally `color` for color
kwargs (dict):
heatmap (bool): background color for cells based on `val`, False
cmap (str): color map to use for norm
"""
bounds = _compute_bounds(sts, ets)
# Compute the number of month calendars we need.
# We want 'square' boxes for each month's calendar, 4x3
fig = plt.figure(figsize=(10.24, 7.68))
if 'fontsize' not in kwargs:
kwargs['fontsize'] = 12
if len(bounds) < 3:
kwargs['fontsize'] = 18
elif len(bounds) < 5:
kwargs['fontsize'] = 16
elif len(bounds) < 10:
kwargs['fontsize'] = 14
if kwargs.get('heatmap', False):
kwargs['cmap'] = plt.get_cmap(kwargs.get('cmap', 'viridis'))
maxval = -1000
for key in data:
if data[key]['val'] > maxval:
maxval = data[key]['val']
# Need at least 3 slots
maxval = 5 if maxval < 5 else maxval
kwargs['norm'] = mpcolors.BoundaryNorm(np.arange(0, maxval),
kwargs['cmap'].N)
for month in bounds:
ax = fig.add_axes(bounds[month])
_do_month(month, ax, data, sts, ets, kwargs)
iemlogo(fig)
title = kwargs.get('title')
if title is not None:
fitbox(fig, title, 0.1, 0.99, 0.95, 0.99)
subtitle = kwargs.get('subtitle')
if subtitle is not None:
fitbox(fig, subtitle, 0.1, 0.99, 0.925, 0.945)
return fig
def calendar_plot(sts, ets, data, **kwargs):
"""Create a plot that looks like a calendar
Args:
sts (datetime.date): start date of this plot
ets (datetime.date): end date of this plot (inclusive)
data (dict[dict]): dictionary with keys of dates and dicts for
`val` value and optionally `color` for color
kwargs (dict):
heatmap (bool): background color for cells based on `val`, False
cmap (str): color map to use for norm
"""
bounds = _compute_bounds(sts, ets)
# Compute the number of month calendars we need.
# We want 'square' boxes for each month's calendar, 4x3
fig = plt.figure(figsize=(10.24, 7.68))
if 'fontsize' not in kwargs:
kwargs['fontsize'] = 12
if len(bounds) < 3:
kwargs['fontsize'] = 18
elif len(bounds) < 5:
kwargs['fontsize'] = 16
elif len(bounds) < 10:
kwargs['fontsize'] = 14
if kwargs.get('heatmap', False):
kwargs['cmap'] = plt.get_cmap(kwargs.get('cmap', 'viridis'))
maxval = -1000
for key in data:
if data[key]['val'] > maxval:
maxval = data[key]['val']
# Need at least 3 slots
maxval = 5 if maxval < 5 else maxval
kwargs['norm'] = mpcolors.BoundaryNorm(np.arange(0, maxval),
kwargs['cmap'].N)
for month in bounds:
ax = fig.add_axes(bounds[month])
_do_month(month, ax, data, sts, ets, kwargs)
iemlogo(fig)
title = kwargs.get('title')
if title is not None:
fitbox(fig, title, 0.1, 0.99, 0.95, 0.99)
subtitle = kwargs.get('subtitle')
if subtitle is not None:
fitbox(fig, subtitle, 0.1, 0.99, 0.925, 0.945)
return fig
def _do_month(month, axes, data, in_sts, in_ets, kwargs):
"""Place data on this axes"""
axes.get_xaxis().set_visible(False)
axes.get_yaxis().set_visible(False)
pos = axes.get_position()
ndcheight = (pos.y1 - pos.y0)
ndcwidth = (pos.x1 - pos.x0)
fitbox(
plt.gcf(), month.strftime("%B %Y"),
pos.x0, pos.x1, pos.y1, pos.y1 + 0.028, ha='center'
)
axes.add_patch(Rectangle((0., 0.90), 1, 0.1, zorder=2,
facecolor='tan', edgecolor='tan'))
sts = datetime.date(month.year, month.month, 1)
ets = (sts + datetime.timedelta(days=35)).replace(day=1)
calendar.setfirstweekday(calendar.SUNDAY)
weeks = len(calendar.monthcalendar(month.year, month.month))
now = sts
row = 0
dy = 0.9 / float(weeks)
dx = 1. / 7.
for i, dow in enumerate(['SUN', 'MON', 'TUE', 'WED', 'THU',
'FRI', 'SAT']):
axes.text(
1. / 7. * (i + 0.5), 0.94, dow,
fontsize=fontscale(ndcwidth / 7. * 0.4),
ha='center', va='center')
while now < ets:
# Is this Sunday?
if now.weekday() == 6 and now != sts:
row += 1
if now < in_sts or now > in_ets:
now += datetime.timedelta(days=1)
continue
offx = (now.weekday() + 1) if now.weekday() != 6 else 0
axes.text(
offx * dx + 0.01,
0.9 - row * dy - 0.01,
str(now.day), fontsize=fontscale(ndcheight / 5. * 0.25),
color='tan',
va='top'
)
_do_cell(axes, now, data, row, dx, dy, kwargs)
now += datetime.timedelta(days=1)
def _do_cell(axes, now, data, row, dx, dy, kwargs):
"""Do what work is necessary within the cell"""
val = data.get(now, dict()).get("val")
cellcolor = ("None" if kwargs.get("norm") is None or val is None else
kwargs["cmap"](kwargs["norm"]([val]))[0])
offx = (now.weekday() + 1) if now.weekday() != 6 else 0
cellcolor = data.get(now, dict()).get("cellcolor", cellcolor)
rect = Rectangle(
(offx * dx, 0.9 - (row + 1) * dy),
dx,
dy,
zorder=(2 if val is None else 3),
facecolor=cellcolor,
edgecolor="tan" if val is None else "k",
)
axes.add_patch(rect)
if val is None:
return
color = "k"
if not isinstance(cellcolor, str): # this is a string comp here
color = ("k" if
(cellcolor[0] * 256 * 0.299 + cellcolor[1] * 256 * 0.587 +
cellcolor[2] * 256 * 0.114) > 186 else "white")
color = data[now].get("color", color)
# We need to translate the axes NDC coordinates back to the figure coords
bbox = axes.get_position()
sdx = (bbox.x1 - bbox.x0) * dx
sdy = (bbox.y1 - bbox.y0) * dy
x0 = bbox.x0 + offx * sdx
ytop = bbox.y0 + (bbox.y1 - bbox.y0) * 0.9
y0 = ytop - (row + 1) * sdy
fitbox(
plt.gcf(),
val,
x0,
x0 + sdx,
y0,
y0 + sdy * 0.55,
ha="center",
va="center",
color=color,
)
def calendar_plot(sts, ets, data, **kwargs):
"""Create a plot that looks like a calendar
Args:
sts (datetime.date): start date of this plot
ets (datetime.date): end date of this plot (inclusive)
data (dict[dict]): dictionary with keys of dates and dicts for
`val` value and optionally `color` for color
kwargs (dict):
heatmap (bool): background color for cells based on `val`, False
cmap (str): color map to use for norm
"""
bounds = _compute_bounds(sts, ets)
# Compute the number of month calendars we need.
# We want 'square' boxes for each month's calendar, 4x3
fig = plt.figure(figsize=(10.24, 7.68))
if "fontsize" not in kwargs:
kwargs["fontsize"] = 12
if len(bounds) < 3:
kwargs["fontsize"] = 18
elif len(bounds) < 5:
kwargs["fontsize"] = 16
elif len(bounds) < 10:
kwargs["fontsize"] = 14
if kwargs.get("heatmap", False):
kwargs["cmap"] = get_cmap(kwargs.get("cmap", "viridis"))
maxval = -1000
for key in data:
if data[key]["val"] > maxval:
maxval = data[key]["val"]
# Need at least 3 slots
maxval = 5 if maxval < 5 else maxval
# Need to have more colors than bins
kwargs["norm"] = mpcolors.BoundaryNorm(
np.arange(0, maxval,
int(maxval / 255.0) + 1), kwargs["cmap"].N)
for month in bounds:
ax = fig.add_axes(bounds[month])
_do_month(month, ax, data, sts, ets, kwargs)
iemlogo(fig)
title = kwargs.get("title")
if title is not None:
fitbox(fig, title, 0.1, 0.99, 0.95, 0.99)
subtitle = kwargs.get("subtitle")
if subtitle is not None:
if subtitle.find("\n") > 0: # Allow more room
fitbox(fig, subtitle, 0.1, 0.99, 0.909, 0.949)
else:
fitbox(fig, subtitle, 0.1, 0.99, 0.925, 0.945)
return fig
def plotter(fdict):
""" Go """
ctx = get_autoplot_context(fdict, get_description())
typ = ctx["typ"]
sort = ctx["sort"]
date = ctx["date"]
pgconn = get_dbconn("postgis")
sts = utc(date.year, date.month, date.day)
ets = sts + datetime.timedelta(hours=24)
opts = {
"W": {
"fnadd": "-wfo",
"sortby": "wfo ASC, phenomena ASC, eventid ASC",
},
"S": {
"fnadd": "",
"sortby": "size DESC"
},
"T": {
"fnadd": "-time",
"sortby": "issue ASC"
},
}
phenoms = {"W": ["TO", "SV"], "F": ["FF"], "M": ["MA"]}
# Defaults
thumbpx = 100
cols = 10
mybuffer = 10000
header = 35
# Find largest polygon either in height or width
gdf = read_postgis(
"""
SELECT wfo, phenomena, eventid, issue,
ST_area2d(ST_transform(geom,2163)) as size,
(ST_xmax(ST_transform(geom,2163)) +
ST_xmin(ST_transform(geom,2163))) /2.0 as xc,
(ST_ymax(ST_transform(geom,2163)) +
ST_ymin(ST_transform(geom,2163))) /2.0 as yc,
ST_transform(geom, 2163) as utmgeom,
(ST_xmax(ST_transform(geom,2163)) -
ST_xmin(ST_transform(geom,2163))) as width,
(ST_ymax(ST_transform(geom,2163)) -
ST_ymin(ST_transform(geom,2163))) as height
from sbw_""" + str(sts.year) + """
WHERE status = 'NEW' and issue >= %s and issue < %s and
phenomena IN %s and eventid is not null
ORDER by """ + opts[sort]["sortby"] + """
""",
pgconn,
params=(sts, ets, tuple(phenoms[typ])),
geom_col="utmgeom",
index_col=None,
)
# For size reduction work
df = read_sql(
"""
SELECT w.wfo, phenomena, eventid,
sum(ST_area2d(ST_transform(u.geom,2163))) as county_size
from
warnings_""" + str(sts.year) + """ w JOIN ugcs u on (u.gid = w.gid)
WHERE issue >= %s and issue < %s and
significance = 'W' and phenomena IN %s
GROUP by w.wfo, phenomena, eventid
""",
pgconn,
params=(sts, ets, tuple(phenoms[typ])),
index_col=["wfo", "phenomena", "eventid"],
)
# Join the columns
gdf = gdf.merge(df, on=["wfo", "phenomena", "eventid"])
gdf["ratio"] = (1.0 - (gdf["size"] / gdf["county_size"])) * 100.0
# Make mosaic image
events = len(df.index)
rows = int(events / cols) + 1
if events % cols == 0:
rows -= 1
if rows == 0:
rows = 1
ypixels = (rows * thumbpx) + header
fig = plt.figure(figsize=(thumbpx * cols / 100.0, ypixels / 100.0))
plt.axes([0, 0, 1, 1], facecolor="black")
imagemap = StringIO()
utcnow = utc()
imagemap.write("<!-- %s %s -->\n" %
(utcnow.strftime("%Y-%m-%d %H:%M:%S"), sort))
imagemap.write("<map name='mymap'>\n")
# Write metadata to image
mydir = os.sep.join(
[os.path.dirname(os.path.abspath(__file__)), "../../../images"])
logo = mpimage.imread("%s/logo_reallysmall.png" % (mydir, ))
y0 = fig.get_figheight() * 100.0 - logo.shape[0] - 5
fig.figimage(logo, 5, y0, zorder=3)
i = 0
# amount of NDC y space we have for axes plotting
ytop = 1 - header / float((rows * 100) + header)
dy = ytop / float(rows)
ybottom = ytop
# Sumarize totals
y = ytop
dy2 = (1.0 - ytop) / 2.0
for phenomena, df2 in gdf.groupby("phenomena"):
car = (1.0 - df2["size"].sum() / df2["county_size"].sum()) * 100.0
fitbox(
fig,
("%i %s.W: Avg size %5.0f km^2 CAR: %.0f%%") %
(len(df2.index), phenomena, df2["size"].mean() / 1e6, car),
0.8,
0.99,
y,
y + dy2,
color=COLORS[phenomena],
)
y += dy2
fitbox(
fig,
"NWS %s Storm Based Warnings issued %s UTC" % (
" + ".join([VTEC_PHENOMENA[p] for p in phenoms[typ]]),
sts.strftime("%d %b %Y"),
),
0.05,
0.79,
ytop + dy2,
0.999,
color="white",
)
fitbox(
fig,
"Generated: %s UTC, IEM Autplot #203" %
(utcnow.strftime("%d %b %Y %H:%M:%S"), ),
0.05,
0.79,
ytop,
0.999 - dy2,
color="white",
)
# We want to reserve 14pts at the bottom and buffer the plot by 10km
# so we compute this in the y direction, since it limits us
max_dimension = max([gdf["width"].max(), gdf["height"].max()])
yspacing = max_dimension / 2.0 + mybuffer
xspacing = yspacing * 1.08 # approx
for _, row in gdf.iterrows():
# - Map each polygon
x0 = float(row["xc"]) - xspacing
x1 = float(row["xc"]) + xspacing
y0 = float(row["yc"]) - yspacing - (yspacing * 0.14)
y1 = float(row["yc"]) + yspacing - (yspacing * 0.14)
col = i % 10
if col == 0:
ybottom -= dy
ax = plt.axes(
[col * 0.1, ybottom, 0.1, dy],
facecolor="black",
xticks=[],
yticks=[],
aspect="auto",
)
for x in ax.spines:
ax.spines[x].set_visible(False)
ax.set_xlim(x0, x1)
ax.set_ylim(y0, y1)
for poly in row["utmgeom"]:
xs, ys = poly.exterior.xy
color = COLORS[row["phenomena"]]
ax.plot(xs, ys, color=color, lw=2)
car = "NA"
carColor = "white"
if not pd.isnull(row["ratio"]):
carf = row["ratio"]
car = "%.0f" % (carf, )
if carf > 75:
carColor = "green"
if carf < 25:
carColor = "red"
# Draw Text!
issue = row["issue"]
s = "%s.%s.%s.%s" % (
row["wfo"],
row["phenomena"],
row["eventid"],
issue.strftime("%H%M"),
)
# (w, h) = font10.getsize(s)
# print s, h
ax.text(
0,
0,
s,
transform=ax.transAxes,
color="white",
va="bottom",
fontsize=7,
)
s = "%.0f sq km %s%%" % (row["size"] / 1000000.0, car)
ax.text(
0,
0.1,
s,
transform=ax.transAxes,
color=carColor,
va="bottom",
fontsize=7,
)
# Image map
url = ("/vtec/#%s-O-NEW-K%s-%s-%s-%04i") % (
sts.year,
row["wfo"],
row["phenomena"],
"W",
row["eventid"],
)
altxt = "Click for text/image"
pos = ax.get_position()
mx0 = pos.x0 * 1000.0
my = (1.0 - pos.y1) * ypixels
imagemap.write(
('<area href="%s" alt="%s" title="%s" '
'shape="rect" coords="%.0f,%.0f,%.0f,%.0f">\n') %
(url, altxt, altxt, mx0, my, mx0 + thumbpx, my + thumbpx))
i += 1
faux = plt.axes([0, 0, 1, 1], facecolor="None", zorder=100)
for i in range(1, rows):
faux.axhline(i * dy, lw=1.0, color="blue")
imagemap.write("</map>")
imagemap.seek(0)
if gdf.empty:
fitbox(fig, "No warnings Found!", 0.2, 0.8, 0.2, 0.5, color="white")
df = gdf.drop("utmgeom", axis=1)
return fig, df, imagemap.read()
def plotter(fdict):
""" Go """
pgconn = get_dbconn('asos')
ctx = get_autoplot_context(fdict, get_description())
station = ctx['zstation']
h1 = int(ctx['h1'])
h2 = int(ctx['h2'])
varname = ctx['v']
tzname = ctx['_nt'].sts[station]['tzname']
df = read_sql("""
WITH data as (
SELECT valid at time zone %s + '10 minutes'::interval as localvalid,
date_trunc(
'hour', valid at time zone %s + '10 minutes'::interval) as v,
tmpf, dwpf, sknt, drct, alti, relh, random() as r,
coalesce(mslp, alti * 33.8639, 1013.25) as slp
from alldata where station = %s and report_type = 2
and extract(hour from valid at time zone %s + '10 minutes'::interval)
in (%s, %s)),
agg as (
select *, extract(hour from v) as hour,
rank() OVER (PARTITION by v ORDER by localvalid ASC, r ASC) from data
)
SELECT *, date(
case when hour = %s
then date(v - '1 day'::interval)
else date(v) end) from agg WHERE rank = 1
""",
pgconn,
params=(tzname, tzname, station, tzname, h1, h2,
h2 if h2 < h1 else -1),
index_col=None)
if df.empty:
raise NoDataFound("No data was found.")
if varname == 'q':
df['pressure'] = mcalc.add_height_to_pressure(
df['slp'].values * units('millibars'),
ctx['_nt'].sts[station]['elevation'] * units('m')).to(
units('millibar'))
# compute mixing ratio
df['q'] = mcalc.mixing_ratio_from_relative_humidity(
df['relh'].values * units('percent'), df['tmpf'].values *
units('degF'), df['pressure'].values * units('millibars')) * 1000.
# pivot
df = df.pivot(index='date', columns='hour', values=varname).reset_index()
df = df.dropna()
df['doy'] = pd.to_numeric(pd.to_datetime(df['date']).dt.strftime("%j"))
df['year'] = pd.to_datetime(df['date']).dt.year
df['week'] = (df['doy'] / 7).astype(int)
df['delta'] = df[h2] - df[h1]
(fig, ax) = plt.subplots(1, 1)
if ctx['opt'] == 'no':
ax.set_xlabel("Plotted lines are smoothed over %.0f days" %
(ctx['smooth'], ))
ax.set_ylabel(
"%s %s Difference" %
(PDICT[varname], "Accumulated Sum" if ctx['opt'] == 'yes' else ''))
if ctx['opt'] == 'no':
# Histogram
H, xedges, yedges = np.histogram2d(df['doy'].values,
df['delta'].values,
bins=(50, 50))
ax.pcolormesh(xedges,
yedges,
H.transpose(),
cmap=plt.get_cmap(ctx['cmap']),
alpha=0.5)
# Plot an average line
gdf = df.groupby('doy').mean().rolling(ctx['smooth'],
min_periods=1,
center=True).mean()
y = gdf['delta'] if ctx['opt'] == 'no' else gdf['delta'].cumsum()
ax.plot(gdf.index.values,
y,
label='Average',
zorder=6,
lw=2,
color='k',
linestyle='-.')
# Plot selected year
for i in range(1, 5):
year = ctx.get("y%s" % (i, ))
if year is None:
continue
df2 = df[df['year'] == year]
if not df2.empty:
gdf = df2.groupby('doy').mean().rolling(ctx['smooth'],
min_periods=1,
center=True).mean()
y = gdf['delta'] if ctx['opt'] == 'no' else gdf['delta'].cumsum()
ax.plot(gdf.index.values, y, label=str(year), lw=2, zorder=10)
ax.set_xticks((1, 32, 60, 91, 121, 152, 182, 213, 244, 274, 305, 335, 365))
ax.set_xticklabels(calendar.month_abbr[1:])
ax.set_xlim(1, 366)
ax.grid(True)
ax.legend(loc='best', ncol=5)
sts = datetime.datetime(2000, 6, 1, h1)
ets = datetime.datetime(2000, 6, 1, h2)
title = ("%s [%s] %s Difference (%.0f-%.0f)\n"
"%s minus %s (%s) (timezone: %s)") % (
ctx['_nt'].sts[station]['name'], station, PDICT[varname],
df['year'].min(), df['year'].max(), ets.strftime("%-I %p"),
sts.strftime("%-I %p"),
"same day" if h2 > h1 else "previous day", tzname)
fitbox(fig, title, 0.05, 0.95, 0.91, 0.99, ha='center')
return fig, df
def plotter(fdict):
""" Go """
pgconn = get_dbconn('postgis')
ctx = get_autoplot_context(fdict, get_description())
wfo = ctx['station']
phenomena = ctx['phenomena']
significance = ctx['significance']
if ctx['season'] == 'all':
months = range(1, 13)
elif ctx['season'] == 'water_year':
months = range(1, 13)
elif ctx['season'] == 'spring':
months = [3, 4, 5]
elif ctx['season'] == 'spring2':
months = [4, 5, 6]
elif ctx['season'] == 'fall':
months = [9, 10, 11]
elif ctx['season'] == 'summer':
months = [6, 7, 8]
elif ctx['season'] == 'winter':
months = [12, 1, 2]
else:
ts = datetime.datetime.strptime("2000-" + ctx['season'] + "-01",
'%Y-%b-%d')
# make sure it is length two for the trick below in SQL
months = [ts.month, 999]
nt = NetworkTable("WFO")
(fig, ax) = plt.subplots(1, 1)
tzname = nt.sts[wfo]['tzname']
df = read_sql("""
WITH data as (
SELECT extract(year from issue) as yr, eventid,
min(issue at time zone %s) as minissue,
max(expire at time zone %s) as maxexpire from warnings WHERE
phenomena = %s and significance = %s
and wfo = %s and
extract(month from issue) in %s GROUP by yr, eventid),
events as (
select count(*) from data),
timedomain as (
SELECT generate_series(minissue,
least(maxexpire, minissue + '24 hours'::interval)
, '1 minute'::interval)
as ts from data
),
data2 as (
SELECT extract(hour from ts)::int * 60 + extract(minute from ts)::int
as minute, count(*) from timedomain
GROUP by minute ORDER by minute ASC)
select d.minute, d.count, e.count as total from data2 d, events e
""", pgconn, params=(
tzname, tzname, phenomena, significance, wfo, tuple(months)),
index_col='minute')
if df.empty:
raise ValueError("No Results Found")
df['frequency'] = df['count'] / df['total'] * 100.
ax.bar(df.index.values, df['frequency'].values, ec='b', fc='b',
align='center')
ax.grid()
if df['frequency'].max() > 70:
ax.set_ylim(0, 101)
ax.set_xticks(range(0, 25 * 60, 60))
ax.set_xlim(-0.5, 24 * 60 + 1)
ax.set_xticklabels(["Mid", "", "", "3 AM", "", "", "6 AM", "", "", '9 AM',
"", "", "Noon", "", "", "3 PM", "", "", "6 PM",
"", "", "9 PM", "", "", "Mid"])
ax.set_xlabel("Timezone: %s (Daylight or Standard)" % (tzname,))
ax.set_ylabel("Percentage [%%] out of %.0f Events" % (df['total'].max(), ))
title = "[%s] %s :: Time of Day Frequency" % (wfo, nt.sts[wfo]['name'])
subtitle = "%s (%s.%s) [%s]" % (
vtec.get_ps_string(phenomena, significance),
phenomena, significance, MDICT[ctx['season']]
)
fitbox(fig, title, 0.05, 0.95, 0.95, 0.99, ha='center')
fitbox(fig, subtitle, 0.05, 0.95, 0.91, 0.945, ha='center')
return fig, df
def plotter(fdict):
""" Go """
pgconn = get_dbconn("asos")
ctx = get_autoplot_context(fdict, get_description())
station = ctx["zstation"]
date = ctx["date"]
opt = ctx["opt"]
varname = ctx["v"]
tzname = ctx["_nt"].sts[station]["tzname"]
# Resolve how to limit the query data
limiter = ""
if opt == "day":
limiter = (f" and to_char(valid at time zone '{tzname}', 'mmdd') = "
f"'{date.strftime('%m%d')}' ")
subtitle = (f"For Date of {date.strftime('%-d %b')}, "
f"{date.strftime('%-d %b %Y')} plotted in bottom panel")
datefmt = "%I %p"
elif opt == "week":
limiter = f" and extract(week from valid) = {date.strftime('%V')} "
subtitle = (
f"For ISO Week of {date.strftime('%V')}, "
f"week of {date.strftime('%-d %b %Y')} plotted in bottom panel")
datefmt = "%-d %b"
elif opt == "month":
limiter = f" and extract(month from valid) = {date.strftime('%m')} "
subtitle = (f"For Month of {date.strftime('%B')}, "
f"{date.strftime('%b %Y')} plotted in bottom panel")
datefmt = "%-d"
else:
subtitle = f"All Year, {date.year} plotted in bottom panel"
datefmt = "%-d %b"
# Load up all the values, since we need pandas to do some heavy lifting
obsdf = read_sql(
f"""
select valid at time zone 'UTC' as utc_valid,
extract(year from valid at time zone %s) as year,
extract(hour from valid at time zone %s +
'10 minutes'::interval)::int as hr, {varname}
from alldata WHERE station = %s and {varname} is not null {limiter}
and report_type = 2 ORDER by valid ASC
""",
pgconn,
params=(tzname, tzname, station),
index_col=None,
)
if obsdf.empty:
raise NoDataFound("No data was found.")
# Assign percentiles
obsdf["quantile"] = obsdf[["hr", varname]].groupby("hr").rank(pct=True)
# Compute actual percentiles
qtile = (obsdf[["hr", varname
]].groupby("hr").quantile(np.arange(0, 1.01,
0.05)).reset_index())
qtile = qtile.rename(columns={"level_1": "quantile"})
(fig, ax) = plt.subplots(2, 1)
cmap = get_cmap(ctx["cmap"])
for hr, gdf in qtile.groupby("hr"):
ax[0].plot(
gdf["quantile"].values * 100.0,
gdf[varname].values,
color=cmap(hr / 23.0),
label=str(hr),
)
ax[0].set_xlim(0, 100)
ax[0].grid(True)
ax[0].set_ylabel(PDICT[varname])
ax[0].set_xlabel("Percentile")
ax[0].set_position([0.13, 0.55, 0.71, 0.34])
cax = plt.axes([0.86, 0.55, 0.03, 0.33],
frameon=False,
yticks=[],
xticks=[])
cb = ColorbarBase(cax, cmap=cmap)
cb.set_ticks(np.arange(0, 1, 4.0 / 24.0))
cb.set_ticklabels(["Mid", "4 AM", "8 AM", "Noon", "4 PM", "8 PM"])
cb.set_label("Local Hour")
thisyear = obsdf[obsdf["year"] == date.year]
if not thisyear.empty:
ax[1].plot(thisyear["utc_valid"].values,
thisyear["quantile"].values * 100.0)
ax[1].grid(True)
ax[1].set_ylabel("Percentile")
ax[1].set_ylim(-1, 101)
ax[1].xaxis.set_major_formatter(
mdates.DateFormatter(datefmt, tz=pytz.timezone(tzname)))
if opt == "day":
ax[1].set_xlabel(f"Timezone: {tzname}")
title = ("%s %s %s Percentiles\n%s") % (
station,
ctx["_nt"].sts[station]["name"],
PDICT[varname],
subtitle,
)
fitbox(fig, title, 0.01, 0.99, 0.91, 0.99, ha="center", va="center")
return fig, qtile
def plotter(fdict):
""" Go """
pgconn = get_dbconn("asos")
ctx = get_autoplot_context(fdict, get_description())
station = ctx["zstation"]
month = ctx["month"]
varname = ctx["var"]
tzname = ctx["_nt"].sts[station]["tzname"]
if ctx.get("sdate") and ctx.get("edate"):
date_limiter = (
" and (to_char(valid at time zone '%s', 'mmdd') >= '%s'"
" %s to_char(valid at time zone '%s', 'mmdd') <= '%s')") % (
tzname,
ctx["sdate"].strftime("%m%d"),
"or" if ctx["sdate"] > ctx["edate"] else "and",
tzname,
ctx["edate"].strftime("%m%d"),
)
title = "between %s and %s" % (
ctx["sdate"].strftime("%-d %b"),
ctx["edate"].strftime("%-d %b"),
)
if ctx["sdate"] == ctx["edate"]:
date_limiter = (
"and to_char(valid at time zone '%s', 'mmdd') = '%s'") % (
tzname, ctx["sdate"].strftime("%m%d"))
title = "on %s" % (ctx["sdate"].strftime("%-d %b"), )
else:
if month == "all":
months = range(1, 13)
elif month == "fall":
months = [9, 10, 11]
elif month == "winter":
months = [12, 1, 2]
elif month == "spring":
months = [3, 4, 5]
elif month == "summer":
months = [6, 7, 8]
elif month == "octmar":
months = [10, 11, 12, 1, 2, 3]
else:
ts = datetime.datetime.strptime("2000-" + month + "-01",
"%Y-%b-%d")
# make sure it is length two for the trick below in SQL
months = [ts.month, 999]
date_limiter = (
" and extract(month from valid at time zone '%s') in %s") % (
tzname, tuple(months))
title = MDICT[month]
if ctx.get("hour") is not None:
date_limiter += (
f" and extract(hour from valid at time zone '{tzname}' "
f"+ '10 minutes'::interval) = {ctx['hour']}")
dt = datetime.datetime(2000, 1, 1, ctx["hour"])
title += " @" + dt.strftime("%-I %p")
(agg, dbvar) = varname.split("_")
if agg in ["max", "min"]:
titlelabel = "Top"
sorder = "DESC" if agg == "max" else "ASC"
df = read_sql(
f"""
WITH data as (
SELECT valid at time zone %s as v, {dbvar} from alldata
WHERE station = %s {date_limiter})
SELECT v as valid, {dbvar} from data
ORDER by {dbvar} {sorder} NULLS LAST LIMIT 100
""",
pgconn,
params=(ctx["_nt"].sts[station]["tzname"], station),
index_col=None,
)
else:
titlelabel = "Most Recent"
op = ">=" if agg == "above" else "<"
threshold = float(ctx.get("threshold", 100))
df = read_sql(
f"SELECT valid at time zone %s as valid, {dbvar} from alldata "
f"WHERE station = %s {date_limiter} and {dbvar} {op} {threshold} "
"ORDER by valid DESC LIMIT 100",
pgconn,
params=(ctx["_nt"].sts[station]["tzname"], station),
index_col=None,
)
if df.empty:
raise NoDataFound("Error, no results returned!")
ylabels = []
fmt = "%.0f" if dbvar in ["tmpf", "dwpf"] else "%.2f"
hours = []
y = []
lastval = -99
ranks = []
currentrank = 0
rows2keep = []
for idx, row in df.iterrows():
key = row["valid"].strftime("%Y%m%d%H")
if key in hours or pd.isnull(row[dbvar]):
continue
rows2keep.append(idx)
hours.append(key)
y.append(row[dbvar])
lbl = fmt % (row[dbvar], )
lbl += " -- %s" % (row["valid"].strftime("%b %d, %Y %-I:%M %p"), )
ylabels.append(lbl)
if row[dbvar] != lastval or agg in ["above", "below"]:
currentrank += 1
ranks.append(currentrank)
lastval = row[dbvar]
if len(ylabels) == 10:
break
if not y:
raise NoDataFound("No data found.")
fig = plt.figure(figsize=(8, 6))
ax = plt.axes([0.1, 0.1, 0.5, 0.8])
ax.barh(
range(len(y), 0, -1),
y,
ec="green",
fc="green",
height=0.8,
align="center",
)
ax2 = ax.twinx()
ax2.set_ylim(0.5, 10.5)
ax.set_ylim(0.5, 10.5)
ax2.set_yticks(range(1, len(y) + 1))
ax.set_yticks(range(1, len(y) + 1))
ax.set_yticklabels(["#%s" % (x, ) for x in ranks][::-1])
ax2.set_yticklabels(ylabels[::-1])
ax.grid(True, zorder=11)
ax.set_xlabel("%s %s" % (METRICS[varname], UNITS[dbvar]))
ab = ctx["_nt"].sts[station]["archive_begin"]
if ab is None:
raise NoDataFound("Unknown station metadata.")
fitbox(
fig,
("%s [%s] %s 10 Events\n%s %s (%s) (%s-%s)") % (
ctx["_nt"].sts[station]["name"],
station,
titlelabel,
METRICS[varname],
ctx.get("threshold") if agg in ["above", "below"] else "",
title,
ab.year,
datetime.datetime.now().year,
),
0.01,
0.99,
0.91,
0.99,
ha="center",
)
fig.text(
0.98,
0.03,
"Timezone: %s" % (ctx["_nt"].sts[station]["tzname"], ),
ha="right",
)
return fig, df.loc[rows2keep]
def plotter(fdict):
""" Go """
pgconn = get_dbconn("asos")
ctx = get_autoplot_context(fdict, get_description())
station = ctx["zstation"]
h1 = int(ctx["h1"])
h2 = int(ctx["h2"])
varname = ctx["v"]
tzname = ctx["_nt"].sts[station]["tzname"]
df = read_sql(
"""
WITH data as (
SELECT valid at time zone %s + '10 minutes'::interval as localvalid,
date_trunc(
'hour', valid at time zone %s + '10 minutes'::interval) as v,
tmpf, dwpf, sknt, drct, alti, relh, random() as r,
coalesce(mslp, alti * 33.8639, 1013.25) as slp
from alldata where station = %s and report_type = 2
and extract(hour from valid at time zone %s + '10 minutes'::interval)
in (%s, %s)),
agg as (
select *, extract(hour from v) as hour,
rank() OVER (PARTITION by v ORDER by localvalid ASC, r ASC) from data
)
SELECT *, date(
case when hour = %s
then date(v - '1 day'::interval)
else date(v) end) from agg WHERE rank = 1
""",
pgconn,
params=(
tzname,
tzname,
station,
tzname,
h1,
h2,
h2 if h2 < h1 else -1,
),
index_col=None,
)
if df.empty:
raise NoDataFound("No data was found.")
if varname == "q":
df["pressure"] = mcalc.add_height_to_pressure(
df["slp"].values * units("millibars"),
ctx["_nt"].sts[station]["elevation"] * units("m"),
).to(units("millibar"))
# compute mixing ratio
df["q"] = (mcalc.mixing_ratio_from_relative_humidity(
df["relh"].values * units("percent"),
df["tmpf"].values * units("degF"),
df["pressure"].values * units("millibars"),
) * 1000.0)
# pivot
df = df.pivot(index="date", columns="hour", values=varname).reset_index()
df = df.dropna()
df["doy"] = pd.to_numeric(pd.to_datetime(df["date"]).dt.strftime("%j"))
df["year"] = pd.to_datetime(df["date"]).dt.year
df["week"] = (df["doy"] / 7).astype(int)
df["delta"] = df[h2] - df[h1]
(fig, ax) = plt.subplots(1, 1)
if ctx["opt"] == "no":
ax.set_xlabel("Plotted lines are smoothed over %.0f days" %
(ctx["smooth"], ))
ax.set_ylabel(
"%s %s Difference" %
(PDICT[varname], "Accumulated Sum" if ctx["opt"] == "yes" else ""))
if ctx["opt"] == "no":
# Histogram
H, xedges, yedges = np.histogram2d(df["doy"].values,
df["delta"].values,
bins=(50, 50))
ax.pcolormesh(
xedges,
yedges,
H.transpose(),
cmap=get_cmap(ctx["cmap"]),
alpha=0.5,
)
# Plot an average line
gdf = (df.groupby("doy").mean().rolling(ctx["smooth"],
min_periods=1,
center=True).mean())
y = gdf["delta"] if ctx["opt"] == "no" else gdf["delta"].cumsum()
ax.plot(
gdf.index.values,
y,
label="Average",
zorder=6,
lw=2,
color="k",
linestyle="-.",
)
# Plot selected year
for i in range(1, 5):
year = ctx.get("y%s" % (i, ))
if year is None:
continue
df2 = df[df["year"] == year]
if not df2.empty:
gdf = (df2.groupby("doy").mean().rolling(ctx["smooth"],
min_periods=1,
center=True).mean())
y = gdf["delta"] if ctx["opt"] == "no" else gdf["delta"].cumsum()
ax.plot(gdf.index.values, y, label=str(year), lw=2, zorder=10)
ax.set_xticks((1, 32, 60, 91, 121, 152, 182, 213, 244, 274, 305, 335))
ax.set_xticklabels(calendar.month_abbr[1:])
ax.set_xlim(1, 366)
ax.grid(True)
ax.legend(loc="best", ncol=5)
sts = datetime.datetime(2000, 6, 1, h1)
ets = datetime.datetime(2000, 6, 1, h2)
title = ("%s [%s] %s Difference (%.0f-%.0f)\n"
"%s minus %s (%s) (timezone: %s)") % (
ctx["_nt"].sts[station]["name"],
station,
PDICT[varname],
df["year"].min(),
df["year"].max(),
ets.strftime("%-I %p"),
sts.strftime("%-I %p"),
"same day" if h2 > h1 else "previous day",
tzname,
)
fitbox(fig, title, 0.05, 0.95, 0.91, 0.99, ha="center")
return fig, df
def _make_plot(
station,
df,
units,
nsector,
rmax,
hours,
months,
sname,
level,
bins,
tzname,
**kwargs,
):
"""Generate a matplotlib windrose plot
Args:
station (str): station identifier
df (pd.DataFrame): observations
drct (list): list of wind directions
units (str): units of wind speed
nsector (int): number of bins to use for windrose
rmax (float): radius of the plot
hours (list): hour limit for plot
month (list): month limit for plot
sname (str): station name
level (int): RAOB level in hPa of interest
bins (list): values for binning the wind speeds
tzname (str): Time zone this plot is produced in.
Returns:
matplotlib.Figure
"""
wu = WINDUNITS[units]
# Filters the missing values
df2 = df[df["drct"] >= 0]
direction = df2["drct"].values * mpunits("degree")
if "speed" in df2.columns:
speed = df2["speed"].values * wu["units"]
else:
speed = df2["sknt"].values * mpunits("knots")
if not hasattr(bins, "units"):
bins = wu["bins"] * wu["units"]
if level is not None:
bins = RAOB_BINS[units] * wu["units"]
if len(df2.index) < 5:
wp = WindrosePlot()
wp.ax.text(
0.5,
0.5,
"Not Enough Data For Plot.",
ha="center",
transform=wp.ax.transAxes,
)
return wp.fig
wp = plot(direction, speed, bins=bins, nsector=nsector, rmax=rmax)
# Now we put some fancy debugging info on the plot
tlimit = "[Time Domain: "
if len(hours) == 24 and len(months) == 12:
tlimit = ""
if len(hours) < 24:
if len(hours) > 4:
tlimit += "%s-%s" % (
datetime(2000, 1, 1, hours[0]).strftime("%-I %p"),
datetime(2000, 1, 1, hours[-1]).strftime("%-I %p"),
)
else:
for h in hours:
tlimit += "%s," % (datetime(2000, 1, 1, h).strftime("%-I %p"),)
if len(months) < 12:
for h in months:
tlimit += "%s," % (datetime(2000, h, 1).strftime("%b"),)
if tlimit != "":
tlimit += "]"
label = ("[%s] %s%s\n" "Windrose Plot %s\n" "Time Bounds: %s") % (
station,
sname if sname is not None else "((%s))" % (station,),
"" if level is None else " @%s hPa" % (level,),
tlimit,
_time_domain_string(df, tzname),
)
fitbox(wp.fig, label, 0.14, 0.99, 0.92, 0.99, ha="left")
label = ("Summary\nobs count: %s\nMissing: %s\nAvg Speed: %.1f %s") % (
len(df.index),
len(df.index) - len(df2.index),
speed.m.mean(),
units,
)
wp.fig.text(0.96, 0.11, label, ha="right", fontsize=14)
if not kwargs.get("nogenerated", False):
wp.fig.text(
0.02,
0.1,
"Generated: %s" % (datetime.now().strftime("%d %b %Y"),),
verticalalignment="bottom",
fontsize=14,
)
# Denote the direction blowing from
lbl = ("Calm values are < %.1f %s\nArrows indicate wind direction.") % (
bins.m[0],
units,
)
wp.fig.text(0.02, 0.125, lbl, va="bottom")
return wp.fig
