One of the biggest stories of 2021 was the desperate efforts by migrants to cross the English Channel in small boats – a story which made headlines around the world in November following the deaths of at least 27 people when their dinghy sank. The i paper used a pioneering method of analysing maritime and aviation tracking data obtained through open sources, combined with traditional journalism methods, to reveal how the migrant boat sinking unfolded. This was part of a wider project to understand the extent of the small boat crisis using data obtained under FOI laws.
Our stories on the migrant boat crossings have had more than 40,000 page views in 2021 and have fed into the UK and international debate on the issue. With human rights bodies calling for an inquiry into the November dinghy sinking, they provide evidence of what took place. At a time of our investigation, there was little information in the public domain and little transparency about what had happened in the Channel. Our investigation was able, for the first time, to give an account of how the tragedy unfolded, miles out to sea, over a period of 24 hours. The UK Maritime and Coastguard Agency initially declined to confirm details of its operations as the tragedy was unfolding but, following our investigation, confirmed that it rescued three other boats during a search operation launched on the night of the sinking. i’s data reporter Tom Saunders analysed and mapped the location data, which was obtained and explained by investigations reporter Dean Kirby. We also used traditional reporting techniques, with i’s chief reporter Cahal Milmo travelling to Calais to interview migrants, and interviewing British and French officials. Economics editor David Parsley and deputy political editor Arj Singh also working on the initial story. The investigation has shown what can be achieved by reporters using tracking data alongside other reporting methods – and will act as a model for others to follow. The story followed an investigation in the summer, using data obtained from the Maritime and Coastguard Agency under FOI, which was able to show for the first time the full extent of the migrant crossings and how the people smugglers were operating. In that project, we interviewed migrants who had survived the crossing, as well French police sources and human rights groups in the UK, France and Tunisia.
We used QGIS to analyse and clip elements that were relevant to the investigation which could be found within the bounding box of the data we received. This included maritime boundaries along with land borders in SHP file format. We clipped these files to the boundary box of the data we received then exported them for use within R.
We then used R and the readr package to read the AIS data as a delimited text file. Using base R we can convert the timestamps to POSIXct. We then filtered for different types of ships that may be relevant to the investigation such as lifeboats, military boats and border force ships.
By orientating the data by timestamp and ship we were then able to plot the path of the relevant ships throughout the time period. This was done through simple features and RGDAl packages which allowed us to plot the delimited file on top of geospatial data.
Then using the geosphere package we were able to use the Haversine formula to calculate the distance between points of interest such as the distance between the mayday call and the flight path of the helicopter.
By splitting the time into 3-hour sections we then were able to analyse the path of ships to identify the different types of rescue manoeuvers that both the aircraft and ships took while looking for the dingy.
What was the hardest part of this project?
The hardest part of this project was combing through thousands of data points in one of the world’s busiest shipping lanes to identify the ships that were relevant to our investigation and then plotting that data in order to analyse in detail the paths taken by the different rescue crews.
It was also difficult to combine the delimited AIS data with the geospatial data in a meaningful way. It was an essential part of our analysis, but the raw data was not easy to incorporate particularly when you consider that we had to plot in three dimensions, taking into account not just the longitude and latitude of each shape but the time and the orientation of each ship at each given moment. We worked under considerable time pressure to break a new, exclusive line on a story which was being probed by journalists around the world. Our findings created additional pressures as they were initially at odds with what officials were saying, but ultimately led to them confirming our findings.
What can others learn from this project?
Other journalists can learn that data journalism and geospatial analysis has just as much application in breaking news environments as it does in a slower cycle of longer-form investigative journalism. The data analysis of the dinghy sinking that we carried out was essential to building out the rest of the piece and unlike many investigations, it was the first part of the puzzle, rather than the last thing to be included in the finished article.
On top of this, journalists can also realise the value of quick and efficient data analysis as a way to further breaking news stories. We were able to work on this analysis in two separate directions. Firstly, the data was analysed and rudimentary graphics were distributed to journalists on the scene who could use it as a tool for directing their enquiries while on a separate track we worked on presenting the data in a legible format that would work for both our print and online audiences.