Experts on climate change say a huge programme of tree planting is needed if the UK is to have any chance of reducing its carbon emissions to effectively zero. The BBC England data team examined just how many trees have been planted, and whether the government is on course to meet its ambitious targets. And it personalised the story by providing local versions so readers could see how many trees had been planted in their area with government funding.
The story fulfills one of the BBC’s key priorities to report the effects of climate change on the environment.
BBC News ran programming across multiple platforms highlighting both the level of local tree-planting as well as the scale of the challenge ahead in meeting the Committee on Climate Change’s recommendations for carbon cutting. As well as a story on the BBC News website we also provided shorter versions and radio , geographically tailored to different regions of England that told people the number of trees planted in their area and linked to the longer story from our local news topic pages, thanks to natural language generation software, allowing us to give audiences a story tailored to where they live.
Radio colleagues also had access to scripts written using the same natural language generation tools, listing all of the local councils in their areas and the number of government-funded trees planted since 2010. These scripts were written as though they would be ready for a presenter to read out if they so wanted but in practice were used as the building blocks for regional broadcast content, providing the essential local data in an easy to digest way, coupled with the essential national level context. It avoided colleagues needing to wade through long explanatory notes or filter spreadsheets in order to get to the data that mattered to their audiences. We often provide localised figures in our stories but often these will end up focusing on a few places, whereas this project – containing figures for more than 300 individual local authority districts – gave us the chance to use the local, more personal, aspect as a way into the story for the audience.
The main data visualisations were created using R and the BBC’s bbplot and mapping packages, including an annotated time series highlighting planting rates going back to the 1970s and the geographical breakdown of new trees planted in England with government funding since 2010.
The accompanying local stories and local radio scripts were written using Arria’s natural language generation software, with local data filling in the gaps in sentences but also being used to change the structure of each paragraph according to the data it contained. Python was used to combine local authorities into their respective BBC radio station area.
This workflow produced a story for each local authority district in England, with one summary story for all of London.
Each of the stories contained relevant local statistics that were also put in a national and regional context. The local versions of the story linked to the longform article about the UK trend.
The liveblog for Devon and Cornwall, for example, includes an update about Torridge, one on West Devon, one about East Devon, another about South Hams, the Isles of Scilly, and another about Exeter figures
What was the hardest part of this project?
We were looking at a subject and a theme for coverage but also trying to build knowledge about how we can use data and technology to power local journalism in the future.
Just because climate change is of huge interest, particularly for younger audiences, does not mean that every story about it will connect with people.
The project therefore had to be relatable.
We knew that the issue of tree planting and tree removal had been of great interest after protests about felling in Sheffield, Yorkshire.
So the task was to find data that showed what was happening with trees across the country and then make that relevant to our audiences.
This was where the idea of natural language generation came in, using shorter stories on the rates of planting since 2010 as a way into the story for our audiences.
A technical challenge of this was that stories had to be tagged with a relevant location identifier in order to appear on the BBC’s online local topic pages.
As part of the process, BBC News Labs built a tool to preview stories and the related tags before they were sent to the BBC’s online news content management system for publication.
The logistics involved in publishing such a high volume of localised stories in one day was also a challenge, with the output of stories on the BBC publishing system more typical of a large live news event.
What can others learn from this project?
It is possible for data to tell audiences how a story is locally relevant to them.
This is often done with interactive maps, dashboards and postcode lookup tools. However, with natural language generation we have the opportunity to provide entire stories that are locally relevant.
We can unlock public datasets and tell them in a language, style and tone that makes them accessible to audiences.
Far from being any sort of replacement for journalists, this is a means of enhancing the work they would already do to open up the data that was previously only there for those who knew where to find it.
Automated content can be combined with more in-depth reporting to play to the strengths of both, making stories more relevant to a wider range of audiences, without having to rely solely on automated versions of the story.
The story also demonstrates the role that liveblogs and other rolling coverage platforms (e.g. Facebook pages) can play in serving automated content to bring new audiences to in-depth content.
This was an experimental piece for BBC England but the technology has been used by other teams in the BBC’s general election results coverage, with semi-automated stories providing the constituency by constituency results.
The trees project is also a story of collaboration between newsrooms.
The scripts we wrote and sent out to colleagues gave them access to the data relevant for the local audiences they served without requiring them to sort and filter spreadsheets or try to piece their story together from multiple rows of a CSV.