Retrieving data from remote archives#
This tutorial covers the retrieval of data from the ICOS Carbon Portal and the CEDA archives.
0. Using the tutorial object store#
To avoid adding the example data we use in this tutorial to your normal
object store, we need to tell OpenGHG to use a separate sandboxed object
store that we’ll call the tutorial store. To do this we use the
use_tutorial_store function from openghg.tutorial. This sets the
OPENGHG_TUT_STORE environment variable for this session and won’t
affect your use of OpenGHG outside of this tutorial.
from openghg.tutorial import use_tutorial_store
use_tutorial_store()
1. ICOS#
It’s easy to retrieve atmospheric gas measurements from the ICOS Carbon
Portal using
OpenGHG. To do so we’ll use the retrieve_atmospheric function from
openghg.retrieve.icos.
Checking available data#
You can find the stations available in ICOS using their map interface. Click on a site to see it’s information, then use it’s three letter site code to retrieve data. You can also use the search page to find available data at a given site.
Using retrieve_atmospheric#
First we’ll import retrieve_atmospheric from the retrieve submodule, then
we’ll retrieve some data from Weybourne (WAO). The function will
first check for any data from WAO already stored in the object
store, if any is found it is returned, otherwise it’ll retrieve the data
from the ICOS Carbon Portal, this may take a bit longer.
from openghg.retrieve.icos import retrieve_atmospheric
wao_data = retrieve_atmospheric(site="WAO", species="ch4")
Now we can inspect wao_data, an ObsData object to see what was
retrieved.
wao_data
We can see that we’ve retrieved ch4 data that covers 2013-04-01 -
2015-07-31. Quite a lot of metadata is saved during the retrieval
process, including where the data was retrieved from (dobj_pid in
the metadata), the instruments and their associated metadata and a
citation string.
You can see more information about the instruments by going to the link
in the instrument_data section of the metadata
metadata = wao_data.metadata
instrument_data = metadata["instrument_data"]
citation_string = metadata["citation_string"]
Here we get the instrument name and a link to the instrument data on the ICOS Carbon Portal.
instrument_data
And we can easily get the citation string for the data
citation_string
Viewing the data#
As with any ObsData object we can quickly plot it to have a look.
NOTE: the plot created below may not show up on the online documentation version of this notebook.
wao_data.plot_timeseries()
Data levels#
Data available on the ICOS Carbon Portal is made available under three different levels (see docs).
- Data level 1: Near Real Time Data (NRT) or Internal Work data (IW).
- Data level 2: The final quality checked ICOS RI data set, published by the CFs,
to be distributed through the Carbon Portal.
This level is the ICOS-data product and free available for users.
- Data level 3: All kinds of elaborated products by scientific communities
that rely on ICOS data products are called Level 3 data.
By default level 2 data is retrieved but this can be changed by passing
data_level to retrieve_icos. Below we’ll retrieve some more
recent data from WAO.
wao_data_level1 = retrieve_atmospheric(site="WAO", species="CH4", data_level=1)
wao_data_level1
You can see that we’ve now got data from 2021-07-01 - 2022-04-24. The ability to retrieve different level data has been added for convenienve, choose the best option for your workflow.
NOTE: level 1 data may not have been quality checked.
wao_data_level1.plot_timeseries(title="WAO - Level 1 data")
Forcing retrieval#
As ICOS data is cached by OpenGHG you may sometimes need to force a retrieval from the ICOS Carbon Portal.
If you retrieve data using retrieve_icos and notice that it does not
return the most up to date data (compare the dates with those on the
portal) you can force a retrieval using force_retrieval.
new_data = retrieve_atmospheric(site="WAO", species="CH4", data_level=1, force_retrieval=True)
Here you may notice we get a message telling us there is no new data to process, if you force a retrieval and there is no newer data you’ll see this message.
2. CEDA#
To retrieve data from CEDA you can use the retrieve_surface function
from openghg.retrieve.ceda. This lets you pull down data from CEDA, process
it and store it in the object store. Once the data has been stored
successive calls will retrieve the data from the object store.
NOTE: For the moment only surface observations can be retrieved and it is expected that these are already in a NetCDF file. If you find a file that can’t be processed by the function please open an issue on GitHub and we’ll do our best to add support that file type.
To pull data from CEDA you’ll first need to find the URL of the data. To
do this use the CEDA data browser and
copy the link to the file (right click on the download button and click
copy link / copy link address). You can then pass that URL to
retrieve_surface, it will then download the data, do some
standardisation and checks and store it in the object store.
We don’t currently support downloading restricted data that requires a login to access. If you’d find this useful please open an issue at the link given above.
Now we’re ready to retrieve the data.
from openghg.retrieve.ceda import retrieve_surface
url = "https://dap.ceda.ac.uk/badc/gauge/data/tower/heathfield/co2/100m/bristol-crds_heathfield_20130101_co2-100m.nc?download=1"
hfd_data = retrieve_surface(url=url)
Now we’ve got the data, we can use it as any other ObsData object,
using data and metadata.
hfd_data.plot_timeseries()
Retrieving a second time#
The second time we (or another user) retrieves the data it will be pulled
from the object store, this should be faster than retrieving from CEDA.
To get the same data again use the site, species and inlet
arguments.
hfd_data2 = retrieve_surface(site="hfd", species="co2")
hfd_data2
3. Cleanup#
If you’re finished with the data in this tutorial you can cleanup the
tutorial object store using the clear_tutorial_store function.
from openghg.tutorial import clear_tutorial_store
clear_tutorial_store()