ERA5 time series¶

In this notebook, we will explore how to:

Extract ERA5 reanalysis time series data for a single point from the C3S Climate Data Store (CDS)
Calculate temporal statistics (e.g. daily minima, maxima and means)
Visualise your results with earthkit-plots

Components of earthkit¶

This tutorial uses the following earthkit components - click any logo to open the package documentation:

earthkit-data earthkit-transforms earthkit-plots

Note: some of the examples in this notebook require an optional dependency - the Python package reverse-geocode. This lets you easily look up the nearest location to a latitude-longitude point. To install it, uncomment the cell below and run it.

[1]:

#!pip install reverse-geocode

[2]:

import earthkit.data as ekd
import earthkit.plots as ekp
import earthkit.transforms as ekt

2. Plotting a time series¶

As well as geographical plots, earthkit-plots also has support for statistical plots. To get started, we can instantiate a TimeSeries() object just like we would a Map().

[6]:

chart = ekp.TimeSeries()

# Plot our hourly data as a line
chart.line(data)

chart.show()

../_images/tutorials_time-series-solutions_11_0.png

Notice that earthkit-plots identified the time dimension of the data and put it on the x-axis by default, reducing the padding on the left and right edges to contain the data.

3. Customising the plot¶

Now let’s:

convert the units to celsius
change the line colour
add a title and a label for the y-axis

[7]:

chart = ekp.TimeSeries()

chart.line(
    data,
    units="celsius",
    color="red",
)

# The label string is automatic, based on the data and the units
chart.ylabel()

# You can use metadata templating in titles just like with map plots
chart.title("ERA5 hourly {variable_name}")

chart.show()

../_images/tutorials_time-series-solutions_13_0.png

4. Changing the plot orientation¶

In some cases, you may wish to flip your axes. With earthkit-plots, this is as straightforward as choosing which variable/coordinate/dimension you would like on each axis.

Note that you only need to specify one axis, like below where we specify y="valid_time". Try experimenting with x="t2m" - it works just the same.

[8]:

chart = ekp.TimeSeries(size=(4, 8))

chart.line(
    data,
    y="valid_time", # put valid_time on the y-axis
    units="celsius",
    color="red",
)

chart.xlabel()

chart.title("ERA5 hourly {variable_name}")

chart.show()

/Users/ecm7348/Documents/Coding/earthkit-plots/src/earthkit/plots/sources/__init__.py:280: UserWarning: Unit conversion failed for y: K -> celsius. Error: ufunc 'multiply' cannot use operands with types dtype('<M8[ns]') and dtype('int64'). Returning original values.
  warnings.warn(

../_images/tutorials_time-series-solutions_15_1.png

5. Enhancing ticks and formatting¶

Now let’s add a nicer title and format the time ticks better for clarity.

Here we are adding more metadata keys ot the title, including:

latitude and longitude, with magic formatters (%Lt and %Ln) to easily add cardinal directions
location (requires the reverse-geocode library to have been installed at the top of this notebook!), which can be formatted to show the closest city (%c) or country (%C) to the latitude and longitude point

We are also formatting the x (time) axis with:

frequency="M": show one tick every month
format="%B": show the full month name
period=True: place the labels in the center of each period

[9]:

chart = ekp.TimeSeries()

chart.line(data, units="celsius", color="red")

chart.title("ERA5 hourly {variable_name} at {latitude:%Lt} {longitude:%Ln} ({location:%c}, {location:%C})")

chart.ylabel()

chart.xticks(
    frequency="M",
    format="%B",
    period=True,
)

chart.show()

../_images/tutorials_time-series-solutions_17_0.png

6. Adding daily statistics¶

Finally, let’s do something a bit more interesting by calculating daily minimum and maximum temperatures, and adding them to the plot. We can do this easily with earthkit-transforms:

[10]:

daily_max = ekt.temporal.daily_max(data).rename({"t2m": "daily max"})
daily_min = ekt.temporal.daily_min(data).rename({"t2m": "daily min"})

chart = ekp.TimeSeries()

chart.line(daily_max, units="celsius", label="{name}")
chart.line(data, units="celsius", label="hourly")
chart.line(daily_min, units="celsius", label="{name}")

chart.title("ERA5 hourly {variable_name} at {latitude:%Lt} {longitude:%Ln} ({location:%c}, {location:%C})")

chart.ylabel()

chart.xticks(
    frequency="M",
    format="%B",
    period=True,
)

chart.legend()

chart.show()

../_images/tutorials_time-series-solutions_19_0.png

7. Produce a time series plot for a different location for summer 2025.¶

Let’s do a time series plot for Bologna, Italy - home to ECMWF’s computing facilities.

[11]:

dataset = "reanalysis-era5-single-levels-timeseries"
request = {
    "variable": ["2m_temperature"],
    "location": {"longitude": 11.25, "latitude": 44.5},
    "date": ["2025-06-01/2025-09-30"],
    "data_format": "netcdf"
}

ds = ekd.from_source("cds", dataset, request)
data = ds.to_xarray()

2026-04-15 17:56:34,221 INFO [2026-02-16T00:00:00] - To generate this ERA5 hourly time series dataset, **homogenisation conventions have been applied to the ERA5 source GRIB data** to ensure consistency, usability, and alignment across chosen variables and time steps. The processed data were then written to an **ARCO Zarr archive**, enabling efficient cloud-optimised access and scalable data retrieval. Please refer to the [user guide](https://confluence.ecmwf.int/x/R6cfHg) for details.

- The dataset presented here is a subset of selected parameters from the full [CDS ERA5 hourly data on single levels (1940–present)](https://cds.climate.copernicus.eu/datasets/reanalysis-era5-single-levels?tab=overview). **Requirements for additional parameters may be considered**. Please raise your request with ECMWF Support [here](https://jira.ecmwf.int/plugins/servlet/desk/portal/1/create/202).
2026-04-15 17:56:34,222 INFO Request ID is c98831a4-a284-4f8b-9894-2ef4f6fac397
2026-04-15 17:56:34,274 INFO status has been updated to accepted
2026-04-15 17:56:47,769 INFO status has been updated to running
2026-04-15 17:56:55,417 INFO status has been updated to successful

[12]:

daily_max = ekt.temporal.daily_max(data).rename({"t2m": "daily max"})
daily_min = ekt.temporal.daily_min(data).rename({"t2m": "daily min"})

chart = ekp.TimeSeries()

chart.line(daily_max, units="celsius", label="{name}")
chart.line(data, units="celsius", label="hourly")
chart.line(daily_min, units="celsius", label="{name}")

chart.title("ERA5 hourly {variable_name} at {latitude:%Lt} {longitude:%Ln} ({location:%c}, {location:%C})")

chart.ylabel()

chart.xticks(
    frequency="M",
    format="%B",
    period=True,
)

chart.legend()

chart.show()

../_images/tutorials_time-series-solutions_22_0.png

8. Calculate the daily mean temperature, and plot that between the daily min and daily max (instead of the hourly temperature).¶

[13]:

daily_mean = ekt.temporal.daily_mean(data).rename({"t2m": "daily mean"})

chart = ekp.TimeSeries()

chart.line(daily_max, units="celsius", label="{name}")
chart.line(daily_mean, units="celsius", label="{name}")
chart.line(daily_min, units="celsius", label="{name}")

chart.title("ERA5 hourly {variable_name} at {latitude:%Lt} {longitude:%Ln} ({location:%c}, {location:%C})")

chart.ylabel()

chart.xticks(
    frequency="M",
    format="%B",
    period=True,
)

chart.legend()

chart.show()

../_images/tutorials_time-series-solutions_24_0.png

9. Request `total_precipitation` from ERA5 and calculate the daily mean precipitation and plot as a bar chart.¶

[14]:

dataset = "reanalysis-era5-single-levels-timeseries"
request = {
    "variable": ["total_precipitation"],
    "location": {"longitude": -1, "latitude": 51.5},
    "date": ["2025-06-01/2025-09-30"],
    "data_format": "netcdf"
}

ds = ekd.from_source("cds", dataset, request)
data = ds.to_xarray()

2026-04-15 17:56:56,976 INFO [2026-02-16T00:00:00] - To generate this ERA5 hourly time series dataset, **homogenisation conventions have been applied to the ERA5 source GRIB data** to ensure consistency, usability, and alignment across chosen variables and time steps. The processed data were then written to an **ARCO Zarr archive**, enabling efficient cloud-optimised access and scalable data retrieval. Please refer to the [user guide](https://confluence.ecmwf.int/x/R6cfHg) for details.

- The dataset presented here is a subset of selected parameters from the full [CDS ERA5 hourly data on single levels (1940–present)](https://cds.climate.copernicus.eu/datasets/reanalysis-era5-single-levels?tab=overview). **Requirements for additional parameters may be considered**. Please raise your request with ECMWF Support [here](https://jira.ecmwf.int/plugins/servlet/desk/portal/1/create/202).
2026-04-15 17:56:56,977 INFO Request ID is b2d38f51-c307-47e7-b006-3f81f2449435
2026-04-15 17:56:57,019 INFO status has been updated to accepted
2026-04-15 17:57:18,088 INFO status has been updated to successful

[15]:

import numpy as np
daily_precip = ekt.temporal.daily_mean(ds)
daily_precip = daily_precip.assign_coords(valid_time=daily_precip.valid_time+np.timedelta64(12, "h"))

[16]:

chart = ekp.TimeSeries()

chart.bar(daily_precip, units="mm")

chart.title("ERA5 hourly {variable_name} at {latitude:%Lt} {longitude:%Ln} ({location:%c}, {location:%C})")

chart.ylabel()

chart.xticks(
    frequency="M",
    format="%B",
    period=True,
)

chart.show()

../_images/tutorials_time-series-solutions_28_0.png