Cleopatra weather styles on real data¶

The reference-derived DATA_STYLES weather presets, applied to genuine gridded fields:

• temperature styles on real ERA5 2 m air temperature (Europe), and
• precipitation styles on a real NOAH land-model precipitation raster, plus
• flow accumulation on a real hydrology raster.

Each preset is applied in one call with ArrayGlyph(field, extent=..., ax=ax).plot(style=style, ...).

Setup — imports¶

ArrayGlyph is cleopatra's array-plotting glyph. The data is bundled as NumPy .npz, so nothing here reads GIS files — only numpy + cleopatra.

In [1]:

import numpy as np
import matplotlib.pyplot as plt
from pathlib import Path
import cleopatra
from cleopatra.glyphs.gridded.array_glyph import ArrayGlyph
print("cleopatra", cleopatra.__version__)
from cleopatra.styling.params import DataStyle

import numpy as np import matplotlib.pyplot as plt from pathlib import Path import cleopatra from cleopatra.glyphs.gridded.array_glyph import ArrayGlyph print("cleopatra", cleopatra.__version__) from cleopatra.styling.params import DataStyle

cleopatra 0.32.0

Load the real fields¶

Three genuine gridded fields, each a NumPy .npz holding the array (or stack) and its [xmin, xmax, ymin, ymax] extent:

• ERA5 2 m temperature — day picks the frame with the largest spread (max - min).
• NOAH precipitation and a flow-accumulation raster.

The print reports each field's shape and value range.

In [2]:

DATA = Path("../../../examples/data")   # kernel CWD is the notebook's own folder

def load(name):
    z = np.load(DATA / name)
    return z["array"].astype(float), [float(v) for v in z["extent"]]

z = np.load(DATA / "europe_t2m.npz", allow_pickle=True)
cel = z["celsius"]
day = int(np.argmax([np.nanmax(f) - np.nanmin(f) for f in cel]))   # most dynamic day
t2m = cel[day].astype(float)
t2m_ext = [float(v) for v in z["extent"]]
precip, precip_ext = load("noah_precip_europe.npz")
facc, facc_ext = load("flow_accumulation.npz")
print("t2m:", t2m.shape, "degC", (round(np.nanmin(t2m),1), round(np.nanmax(t2m),1)),
      "| precip:", precip.shape, (round(np.nanmin(precip),1), round(np.nanmax(precip),1)),
      "| flow_acc:", facc.shape)

DATA = Path("../../../examples/data") # kernel CWD is the notebook's own folder def load(name): z = np.load(DATA / name) return z["array"].astype(float), [float(v) for v in z["extent"]] z = np.load(DATA / "europe_t2m.npz", allow_pickle=True) cel = z["celsius"] day = int(np.argmax([np.nanmax(f) - np.nanmin(f) for f in cel])) # most dynamic day t2m = cel[day].astype(float) t2m_ext = [float(v) for v in z["extent"]] precip, precip_ext = load("noah_precip_europe.npz") facc, facc_ext = load("flow_accumulation.npz") print("t2m:", t2m.shape, "degC", (round(np.nanmin(t2m),1), round(np.nanmax(t2m),1)), "| precip:", precip.shape, (round(np.nanmin(precip),1), round(np.nanmax(precip),1)), "| flow_acc:", facc.shape)

t2m: (111, 164) degC (np.float64(-1.5), np.float64(35.5)) | precip: (25, 40) (np.float64(0.0), np.float64(28.8)) | flow_acc: (13, 14)

A gallery helper¶

gallery lays out one field under many styles in a grid — one ArrayGlyph(field, ax=ax).plot(style=...) per panel:

• nrow is derived from the style count and ncol.
• gext reorders the imshow extent to ArrayGlyph's [xmin, ymin, xmax, ymax] convention.
• each panel draws the field with one preset; add_colorbar=False keeps the grid clean and ticks are hidden.
• any unused trailing axes are hidden, and a bold suptitle labels the whole grid.

In [3]:

def gallery(styles, field, extent, title, ncol=3):
    styles = list(styles)
    nrow = int(np.ceil(len(styles) / ncol))
    # squeeze=False -> axes is always a 2-D array, so .ravel() works for any grid
    fig, axes = plt.subplots(nrow, ncol, figsize=(13, 3.3 * nrow), squeeze=False)
    gext = [extent[0], extent[2], extent[1], extent[3]]   # -> ArrayGlyph [xmin, ymin, xmax, ymax]
    for i, ax in enumerate(axes.ravel()):
        if i >= len(styles):
            ax.set_visible(False)                          # hide leftover empty panels
            continue
        glyph = ArrayGlyph(field, extent=gext, ax=ax)
        glyph.plot(add_colorbar=False, title=styles[i], title_size=9, data_style=DataStyle(style=styles[i]))
        ax.set_xticks([])
        ax.set_yticks([])
    fig.suptitle(title, fontweight="bold")
    fig.tight_layout()
    plt.show()

def gallery(styles, field, extent, title, ncol=3): styles = list(styles) nrow = int(np.ceil(len(styles) / ncol)) # squeeze=False -> axes is always a 2-D array, so .ravel() works for any grid fig, axes = plt.subplots(nrow, ncol, figsize=(13, 3.3 * nrow), squeeze=False) gext = [extent[0], extent[2], extent[1], extent[3]] # -> ArrayGlyph [xmin, ymin, xmax, ymax] for i, ax in enumerate(axes.ravel()): if i >= len(styles): ax.set_visible(False) # hide leftover empty panels continue glyph = ArrayGlyph(field, extent=gext, ax=ax) glyph.plot(add_colorbar=False, title=styles[i], title_size=9, data_style=DataStyle(style=styles[i])) ax.set_xticks([]) ax.set_yticks([]) fig.suptitle(title, fontweight="bold") fig.tight_layout() plt.show()

Temperature styles — real ERA5 2 m temperature (°C, Europe)¶

These presets assume °C, which matches ERA5. The two temperature_flame* styles tie opacity to value (hot glows, cool fades) — the CAMS aerosol technique recoloured for heat.

In [4]:

temp_styles = ["temperature_2m", "air_temperature", "mean_temperature_2m",
               "min_temperature_2m", "max_temperature_2m", "dewpoint_temperature_2m",
               "potential_temperature", "temperature", "temperature_flame", "temperature_flame_amber"]
gallery(temp_styles, t2m, t2m_ext, "Temperature styles on real ERA5 2 m temperature")

temp_styles = ["temperature_2m", "air_temperature", "mean_temperature_2m", "min_temperature_2m", "max_temperature_2m", "dewpoint_temperature_2m", "potential_temperature", "temperature", "temperature_flame", "temperature_flame_amber"] gallery(temp_styles, t2m, t2m_ext, "Temperature styles on real ERA5 2 m temperature")

Precipitation styles — real NOAH precipitation¶

The precipitation family (total / convective / large-scale precip, rain & snowfall rates) on a real precipitation field.

In [5]:

precip_styles = ["total_precipitation", "precipitation", "convective_precipitation",
                 "large_scale_precipitation", "precipitation_rate", "convective_rainfall_rate",
                 "large_scale_rainfall_rate", "rainfall", "snowfall", "snowfall_rate_water_equivalent"]
gallery(precip_styles, precip, precip_ext, "Precipitation styles on real NOAH precipitation")

precip_styles = ["total_precipitation", "precipitation", "convective_precipitation", "large_scale_precipitation", "precipitation_rate", "convective_rainfall_rate", "large_scale_rainfall_rate", "rainfall", "snowfall", "snowfall_rate_water_equivalent"] gallery(precip_styles, precip, precip_ext, "Precipitation styles on real NOAH precipitation")

Flow accumulation — real hydrology raster¶

flow_accumulation uses a symmetric-log norm with value-linked opacity, so the channel network stands out against faded low-accumulation cells.

In [6]:

gallery(["flow_accumulation"], facc, facc_ext, "flow_accumulation on a real flow-accumulation raster", ncol=1)

gallery(["flow_accumulation"], facc, facc_ext, "flow_accumulation on a real flow-accumulation raster", ncol=1)