Decades of Arctic weather data now available in one place

A new open access dataset provides the most comprehensive and consistent record yet of Arctic terrestrial ground-level weather observations.

Spanning more than three decades and covering hundreds of locations across the circumpolar north, this data resource will help researchers better understand how Arctic environments are changing—and how these changes connect to the global climate system.

The dataset includes over 30 different weather variables, from air temperature and snow depth to solar radiation and soil moisture.

It draws together data from 13 major publicly available sources, all of which were previously siloed in, for many, unknown corners of the internet, and in incompatible formats.

The work was led by Laura Helene Rasmussen, a postdoctoral researcher with the iC3 Polar Research Hub. 

.

"We can start comparing the whole Arctic"

Laura explained that:

“Lots of excellent Arctic weather monitoring has been done over the last 30 years.”   

"But if you wanted to use it, you had to first of all know where to find it, which is not obvious. Then you had to go through a huge mess of formats, variables, timestamps and undocumented quirks.”

“I wanted to make this kind of data much more accessible to everyone, and put data on the same format, so we can start comparing the whole Arctic directly.”

A long-overdue Arctic data clean-up

Arctic surface weather data is critically important for understanding how ecosystems, permafrost and carbon emissions are changing in a warming climate.

But harsh field conditions, fragmented archiving systems, and the growing politicization of data from the Russian Arctic have made it hard to work with in situ data across the region.

To fix this, Laura and colleagues compiled raw data from national meteorological services, academic monitoring stations, long-term ecosystem observatories and global databases.

The final dataset includes more than 700 Arctic and sub-Arctic locations. It covers the period 1990 to 2023, and is available in two versions: one raw and one that has undergone extensive quality checking. 

.

“The goal was to standardize and clean the data, but without over-processing it,” explained Laura. “We want users to be able to reuse and adapt it for their own purposes, not to force a particular way of analysing it.” 

The most represented variables are air temperature, relative humidity, precipitation and snow depth. But the dataset also includes harder-to-access measurements such as incoming and outgoing radiation, soil moisture and ground temperature at various depths.

Open data and open code

The research team developed a pipeline in Python to import, reformat and normalize the original data from each source.

A second pipeline performs an optional five-step quality check. These steps include identifying physically impossible values (such as relative humidity above 100%), flagging outliers, standardizing units (like converting snow depth to centimetres), and detecting possible calibration problems.

Notably, all of the code is also publicly available alongside the data. This allows researchers to replicate the process or apply the quality-checking system to new or locally held datasets.

Unlocking new opportunities in Arctic science 

The release of this dataset is expected to accelerate research in many fields, including permafrost dynamics, tundra ecology, carbon emissions, snow and ice processes, and model evaluation.

It also has direct policy relevance for managing climate risks and verifying satellite products and reanalysis datasets.

“Having this kind of harmonized weather data is especially useful for researchers who are skilled Arctic scientists, but maybe were stopped from asking some questions because of a technical barrier. Now you can just focus on your research question.”

The dataset can also help make Arctic science more transparent and reproducible, especially at a time when access to some regions and data sources is becoming increasingly restricted.

Access the dataset and find out more 

The dataset is freely available in the open-access journal Scientific Data. You can also access the dataset and its associated code via Zenodo [link now fixed].

Lead author Laura Helene Rasmussen is an MSCA Postdoctoral Fellow at the iC3 Polar Research Hub in Tromsø. You can find out more about her MSCA project in this blog. A list of her publications is here.

Laura developed the dataset together with Bo Markussen and Susanne Ditlevsen as part of her Danish Data Science Academy postdoc fellowship at the University of Copenhagen.

Credit for both images in this blog: Nature Scientific Data