An ECMWF-style temperature animation over Europe — June → July 2026¶

This showcase reproduces the look ECMWF posts on social media — the glowing blue-to-magenta 2 m-temperature maps on a dark canvas with country borders — for the whole of 1 June → 15 July 2026, one frame per day, straight from earthlens.

The colour is not hand-rolled: cleopatra ships style="2t", the genuine ECMWF Magics palette for 2 m temperature (the GRIB shortName 2t, vendored from ECMWF's own palettes.json). pyramids forwards style= straight through Dataset / DatasetCollection, so a single cube.plot(style="2t") gives the ECMWF colour scale, and add_reference_map(style="ecmwf-dark") drops the dark basemap and borders underneath.

On "till now": ERA5-Land's daily aggregate is not real-time — verified live, its most recent published day as of this writing is 15 July 2026 (about nine days behind). So the animation runs 1 June → 15 July; extend END once newer days publish.

Needs GEE_SERVICE_ACCOUNT / GEE_SERVICE_KEY and pyramids-gis[viz] (cleopatra ≥ 0.26).

Setup¶

pyramids supplies Dataset / DatasetCollection; cleopatra supplies the ECMWF "2t" preset, the dark-basemap inset, and apply_blank_canvas for the dark look; earthlens supplies the unified EarthLens entry point.

In [ ]:

import base64
import os
import warnings
from pathlib import Path

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
from cleopatra.styling.colors import DATA_STYLES, resolve_colormap
from cleopatra.styling.styles import apply_blank_canvas
from IPython.display import HTML
from loguru import logger
from pyramids.dataset import Dataset
from pyramids.dataset.collection import DatasetCollection

from earthlens.core import EarthLens

warnings.filterwarnings("ignore")
logger.remove()

import base64 import os import warnings from pathlib import Path import matplotlib.pyplot as plt import numpy as np import pandas as pd from cleopatra.styling.colors import DATA_STYLES, resolve_colormap from cleopatra.styling.styles import apply_blank_canvas from IPython.display import HTML from loguru import logger from pyramids.dataset import Dataset from pyramids.dataset.collection import DatasetCollection from earthlens.core import EarthLens warnings.filterwarnings("ignore") logger.remove()

Credentials, region and window¶

Credentials come from a repo-root .env (loaded with python-dotenv). EUROPE is a generous continental box; START / END bracket the June-to-mid-July window.

In [ ]:

from dotenv import find_dotenv, load_dotenv

load_dotenv(find_dotenv(usecwd=True))
SERVICE_ACCOUNT = os.environ["GEE_SERVICE_ACCOUNT"]
SERVICE_KEY = os.environ["GEE_SERVICE_KEY"]

EUROPE = {"lat_lim": [34.0, 62.0], "lon_lim": [-12.0, 32.0]}
START, END = "2026-06-01", "2026-07-15"  # 07-15 = latest published ERA5-Land daily day

OUT = Path("out") / "europe_temperature_2026"
OUT.mkdir(parents=True, exist_ok=True)

from dotenv import find_dotenv, load_dotenv load_dotenv(find_dotenv(usecwd=True)) SERVICE_ACCOUNT = os.environ["GEE_SERVICE_ACCOUNT"] SERVICE_KEY = os.environ["GEE_SERVICE_KEY"] EUROPE = {"lat_lim": [34.0, 62.0], "lon_lim": [-12.0, 32.0]} START, END = "2026-06-01", "2026-07-15" # 07-15 = latest published ERA5-Land daily day OUT = Path("out") / "europe_temperature_2026" OUT.mkdir(parents=True, exist_ok=True)

1 · Pull one daily 2 m-temperature tile per day¶

ERA5-Land's daily aggregate is hosted on Earth Engine, so we reach temperature_2m through the gee key. One daily-mean GeoTIFF per day over the continental box. Already-downloaded tiles are reused, so a rerun does not re-hit Earth Engine.

In [ ]:

raw_dir = OUT / "t2m_daily"
raw_dir.mkdir(parents=True, exist_ok=True)

cached = sorted(raw_dir.glob("*.tif"))
if cached:  # reuse a previous run's tiles instead of re-pulling from Earth Engine
    eu_paths = cached
    print(len(eu_paths), "daily tiles (cached)")
else:
    job = EarthLens(
        data_source="gee",
        dataset="ECMWF/ERA5_LAND/DAILY_AGGR",
        variables=["temperature_2m"],
        start=START,
        end=END,
        temporal_resolution="daily",
        scale=10000.0,
        path=raw_dir,
        export_via="url",
        **EUROPE,
    )
    job.authenticate(service_account=SERVICE_ACCOUNT, service_key=SERVICE_KEY)
    eu_paths = job.download(progress_bar=False)
    print(len(eu_paths), "daily tiles")

raw_dir = OUT / "t2m_daily" raw_dir.mkdir(parents=True, exist_ok=True) cached = sorted(raw_dir.glob("*.tif")) if cached: # reuse a previous run's tiles instead of re-pulling from Earth Engine eu_paths = cached print(len(eu_paths), "daily tiles (cached)") else: job = EarthLens( data_source="gee", dataset="ECMWF/ERA5_LAND/DAILY_AGGR", variables=["temperature_2m"], start=START, end=END, temporal_resolution="daily", scale=10000.0, path=raw_dir, export_via="url", **EUROPE, ) job.authenticate(service_account=SERVICE_ACCOUNT, service_key=SERVICE_KEY) eu_paths = job.download(progress_bar=False) print(len(eu_paths), "daily tiles")

2 · Convert kelvin to °C¶

ERA5-Land 2 m temperature is in kelvin with 0 as the sea/fill value; mask everything below 200 K — well under any real surface temperature, so it catches the 0 fill without touching valid data — and subtract 273.15. Each converted day is written into t2m_celsius/ and the day is recorded as a frame label.

In [ ]:

celsius_dir = OUT / "t2m_celsius"
celsius_dir.mkdir(parents=True, exist_ok=True)

celsius_paths = []
for path in sorted(eu_paths):
    field = Dataset.read_file(str(path))
    field = field.apply(lambda kelvin: np.where(kelvin > 200, kelvin - 273.15, np.nan))
    cpath = celsius_dir / Path(path).name
    field.to_file(str(cpath))
    celsius_paths.append(cpath)

labels = [
    pd.to_datetime(Path(p).stem.split("_")[-1]).strftime("%b %d")
    for p in sorted(eu_paths)
]
len(celsius_paths), labels[0], "->", labels[-1]

celsius_dir = OUT / "t2m_celsius" celsius_dir.mkdir(parents=True, exist_ok=True) celsius_paths = [] for path in sorted(eu_paths): field = Dataset.read_file(str(path)) field = field.apply(lambda kelvin: np.where(kelvin > 200, kelvin - 273.15, np.nan)) cpath = celsius_dir / Path(path).name field.to_file(str(cpath)) celsius_paths.append(cpath) labels = [ pd.to_datetime(Path(p).stem.split("_")[-1]).strftime("%b %d") for p in sorted(eu_paths) ] len(celsius_paths), labels[0], "->", labels[-1]

3 · Animate with the ECMWF "2t" preset + a dark base map¶

This is the ECMWF-social-media recipe, one call at a time:

• cube.plot(style="2t") — the genuine ECMWF Magics 2 m-temperature palette.
• apply_blank_canvas(..., facecolor="black") — the dark-animation canvas.
• save_animation(..., fps=5, crf=23) — writes the frames out as an mp4.
• add_reference_map(style="ecmwf-dark", ...) — the dark basemap with country borders underneath.

The "2t" palette is flat-saturated at both ends by design, with the gradient packed into the middle, so we widen its range to [-10, 50] °C — that puts the real 5–38 °C daily-mean spread into the gradient band instead of the saturated caps, and it reads like ECMWF's. DATA_STYLES is cleopatra's shared module-level table, so the two injected keys are removed again in finally to avoid leaking a mutated "2t" style into any other cell or notebook in the same session.

In [ ]:

west, east = EUROPE["lon_lim"]
south, north = EUROPE["lat_lim"]

# The ECMWF "2t" 2 m-temperature palette (cleopatra's temperature_2m preset),
# stretched to a fixed -10..50 degC range so the heat spread lands in the gradient
# band rather than saturating the palette's flat ends.
t2m_cmap = resolve_colormap(next(iter(DATA_STYLES["temperature_2m"].values()))["cmap"])
cube = DatasetCollection.from_files(celsius_paths)
glyph = cube.plot(cmap=t2m_cmap, vmin=-10, vmax=50, figsize=(8.5, 7.6))
apply_blank_canvas(glyph.ax, facecolor="black")  # the ECMWF/CAMS dark-animation look

# The reference map's extent is in degrees, matching the bbox the frames were
# cropped to, so the inset lines up with the data underneath it.
glyph.add_reference_map(style="dark", extent=[west, south, east, north])

mp4 = OUT / "europe_t2m_2026.mp4"
glyph.save_animation(str(mp4), fps=5, crf=23)
plt.close("all")

mp4

west, east = EUROPE["lon_lim"] south, north = EUROPE["lat_lim"] # The ECMWF "2t" 2 m-temperature palette (cleopatra's temperature_2m preset), # stretched to a fixed -10..50 degC range so the heat spread lands in the gradient # band rather than saturating the palette's flat ends. t2m_cmap = resolve_colormap(next(iter(DATA_STYLES["temperature_2m"].values()))["cmap"]) cube = DatasetCollection.from_files(celsius_paths) glyph = cube.plot(cmap=t2m_cmap, vmin=-10, vmax=50, figsize=(8.5, 7.6)) apply_blank_canvas(glyph.ax, facecolor="black") # the ECMWF/CAMS dark-animation look # The reference map's extent is in degrees, matching the bbox the frames were # cropped to, so the inset lines up with the data underneath it. glyph.add_reference_map(style="dark", extent=[west, south, east, north]) mp4 = OUT / "europe_t2m_2026.mp4" glyph.save_animation(str(mp4), fps=5, crf=23) plt.close("all") mp4

Embed the MP4 straight into the notebook as a base64 data URI so the animation travels with the file.

In [ ]:

encoded = base64.b64encode(mp4.read_bytes()).decode()
HTML(
    f'<video src="data:video/mp4;base64,{encoded}" width="760" '
    'autoplay loop muted playsinline controls></video>'
)

encoded = base64.b64encode(mp4.read_bytes()).decode() HTML( f'<video src="data:video/mp4;base64,{encoded}" width="760" ' 'autoplay loop muted playsinline controls></video>' )

Notes¶

• Swap the preset. style="2t" is 2 m temperature; cleopatra vendors other ECMWF Magics palettes keyed by GRIB shortName. Inspect from cleopatra.styling.colors import DATA_STYLES; sorted(DATA_STYLES).
• Lighter look. Drop apply_blank_canvas and switch to add_reference_map(style="ecmwf-light") for the white-background variant.
• Longer / smoother. Extend END as newer ERA5-Land days publish, or raise fps for a faster playback. Lower scale for a sharper (heavier) movie.
• Same recipe elsewhere. Change EUROPE to any box and re-run — the whole animation is one cube.plot(style=...) → apply_blank_canvas → add_reference_map → save_animation sequence.