Digging for Data on Pipelines

Gathering the news usually means a lot of hard work - developing contacts, researching precedents, asking the right questions at the right moment. One skill that is more important all the time is knowing how to find and use big blocks of data. Here at CBC News we are blessed to have some top-of-class "data journalists" renowned for their ability to examine and analyze complex numbers and patterns.

This week, they helped us better understand the safety record of the labyrinth of oil and gas pipelines that criss-cross Canada.

There is no shortage of debate about where and when pipelines should go, from the Obama administration's long-delayed decision on Keystone XL to the recent debate over Enbridge's push to reverse Line 9.

But for all that loud and rancorous debate, we realized it would be difficult for many Canadians to assess those arguments. It's hard to find public information about safety-related incidents other than the ones that make big headlines.

So our project began in August of 2012 with an access-to-information request to the National Energy Board for data on all "reportable incidents" since 2000 involving pipelines that cross provincial and international borders. This week, after many months of data crunching and verification, we published the results in a searchable map -- the centrepiece of a series of stories looking at pipeline safety and the availability of information from government regulators.

Our map shows it all: 1,047 incidents reported by the industry to the NEB, from exploding batteries and accidental fires to leaks of oil and gas.

It wasn't easy getting to this point. The 405 pages of data sent to the CBC only included incidents that were classified by the NEB as "reportable," meaning not all the information the industry sent to the regulator was given to us. (A "reportable incident" includes unintended fires/explosions, crude oil spills over 1,500 litres, gas leaks of any amount, and cases where a worker was fatally or seriously injured.)

The documents we got were also extremely difficult to sort and analyze. In the end, we had to use commercial optical character recognition software to convert images of text into editable documents.

As the CBC's Amber Hildebrandt and Michael Pereira reported, the original documents had many holes and redactions. The type of information provided by companies varied wildly from incident to incident. Fields for such things as "spill amount" and "substance" were often left blank. When possible, our researchers completed those fields based on details provided in the incident summaries.

The federal regulator has acknowledged to CBC News that their pipeline incident database wasn't always kept up to date. The NEB said it has other documents that contain the information necessary to properly track companies under its purview.

Where the NEB had published more recent information on an incident, we updated that on the map. We also tracked down publicly available reports on investigations into some of the biggest incidents and added those links to provide Canadians with more context.

Before we went live with the map, CBC gave a sample of the updated data to the federal regulator to check our methodology. The NEB identified one mistake made by us as well as one mistake made by them, plus they volunteered a few extra details for some of the incidents. We added all that information to the map.

One group of pipeline companies has suggested that we employed too broad a definition of an "incident," and the result is that the safety numbers look worse than they really are.

We stand by our approach of relying on official NEB information.

In fact, in the spirit of transparency, CBC also opened up the data for this project to the public. We posted the final spreadsheet as a downloadable file  that showcases both the NEB's original information plus columns filled out by the CBC. The original NEB documents were also published in two files.

Our incident map and the stories it spurred raise important questions about public access to pipeline safety data in Canada. As we learned, this country lags far behind the United States in terms of transparency.

After all, if it's this tough for some of the best data journalists in the country to determine what's happening to Canada's pipelines, what hope does the public have? We think the publication of our map and the raw data are the first steps to greater openness on this side of the border.

Brodie Fenlon
Managing Editor, CBCNews.ca