Data “stormy-telling”: How data stories could help during weather events
This was a mild winter by every measure on the East Coast of the U.S., where I live. We had a few cold spells, but we also hit 70 degrees in February more than once. We hardly had any snow, which meant the skiing was terrible, which was the only real downside. My wife would say that’s not even a downside!
Then came March. It came in like a lion, as the saying goes. We had a series of nor’easters, rip-roaring windstorms coming in from the northeast rather than the west, which is the source for most of our weather. This means the wind and moisture are coming off the ocean, which often makes for the fiercest storms. You can have hurricane-force winds with a nor’easter and if it includes snow, these storms can dump the white stuff by the foot.
Anyway, we had a nor’easter on 2 March, and then another one on 7 March, another one 12 March, and then yet another one 21 March (affectionately dubbed a “four-easter” since it was the fourth one in three weeks). The second one is the one that affected me the most. We lost power for days. And it got me wondering: Who else was without power during the storm? Where were the power crews at the moment I needed to know? When would my power be restored? This was an opportunity for a data story that my power company definitely missed out on.
My questions were clear—and pretty obvious. So how would you solve this with a data story or even a simple data dashboard of some kind that might help? The most obvious answer is that you would want to produce a map where someone could see the outages. This makes perfect sense. And my power company did that! Good start, but is this map helpful?
I will argue that it is not. First of all, this is as detailed a map as I can get. I can see my town. And I can see that in my town, 2,500–5,000 people were without power. But that’s not very useful. Mildly interesting? Sure. Useful? No. And, as the color scale indicates, I can see the total number, which makes the one pink town (Framingham) look like it’s a disaster. However, when you roll over the towns on the map, you see that only about 29 percent of Framingham’s residents were without power, whereas in my town, Hopkinton, it was 54 percent.
So even this mildly interesting map didn’t immediately communicate the most relevant number to me (the curiosity is more about percentage than total number, for me.)
But that’s not where this map really failed me. What I wanted to know was where specifically was the outage in my area. If I could zoom in on Hopkinton and see street by street which areas were without power, it would have given me a real sense not just of the size of the impact, but the locality of the impact. If every street within three miles of my house were out, and if the map even indicated a transformer that was blown and/or a power sub-station facing issues, it would have given me a much better sense of how long I might expect to be without power. And, even more helpful, it could have shown me little dots indicating the location of crews in real time, so I could see if they were working in my area. This would really help me make a guess about how long before I see my lights (and internet!) turned back on.
Perhaps it could even have included a future-forward prediction (like weather maps that show how a storm will move in time beyond now) of when they expected crews to move on to other locations, along with estimated repair times. They could have included lots of caveats, so no one would hold the company to those timelines explicitly. This transparency would reap dividends though there will always be some cranks (like those who criticize weather reporters when the storms shift at the last second) who might complain when some predictions end up being too rosy.
All of the data I’m suggesting is available. The company knows where it has outages and you can bet it’s all in a computer somewhere. The reporting system for outages is fully automated–via computer and/or phone. And you also know they know where their crews are. I’m sure there are multiple databases with everything they would need.
So, a data dashboard is an obvious opportunity. What about a real data story? What might that look like? What purpose might that serve?
Imagine this: The company produces an article with interactive and animated visuals that they run on their website and make available to local newspapers as a tool. The story is an after-action review of the storm and their process of returning power to the 350,000 people who lost power. This is an outline of how it could flow, along with a description of the associated visuals it could include.
- On 7 March, a strong nor’easter hit Massachusetts with top wind speeds exceeding 70 miles per hour. Over a period of 12 hours, 350,000 people lost power.
(Visual: an animated map showing red dots that appear as time progresses showing reported outages, filling up the visual with an overwhelming display of the number of places with outages.)
- Our crews were already out in force as the storm began, still cleaning up from the storm on 2 March. We had staffed up to handle that previous storm, increasing our crews on the ground by X percent to a total of Y linemen.
(Visual: a static map showing green and blue dots indicating crews in place as the storm began, already working on other outages. The green dots indicate full-time employees and the blue dots indicate the additional crew working for the company. A second map could even indicate where those crew came from; often they are from out of state, brought in to help in an emergency. A third visual would be an overall percentage of people with power—still below 100 percent from the previous storm.)
- As the storm waned and we were able to begin work (our crews cannot work until winds are below 40 miles per hour), our first priority was to support emergency services to clear roads and dangerous wires in the towns in our service area.
(Visual: animated and interactive map showing where crews were working, with road names visible if the user zooms in so they can see the major roads affected. The overall percentage of users with power would remain visible, slowly creeping up as time progresses.)
- Once the emergency services support was complete, we began the task of working to restore power throughout our service area. Our priority is to work on X first, Y second, and Z third. For instance, when there is a power sub-station out, affecting at least XX people, that is our first point of attack.
(Visual: static map indicating a real example of one of these types of impacts, showing the number of crew there and how long it took to fix. Overall percentage visual remains and animates to show this work progressing. Perhaps a small animated map showing this type of work progressing statewide with small dots changing from red dots to green checkmarks, indicating successful repair.)
This story outline could continue for a few more steps, outlining the entire process through the repair priorities list. Each step would show the timeline of progress, and would clearly illustrate how decisions are made, how crews are moved around, and the amount of effort it takes to get things up and running.
The added bonus of working on this type of output would be establishing a process and technologies that could be leveraged to make the real-time dashboard available during the storm and in the immediate aftermath to help people see what’s happing in their neighborhoods as described above.
The after-action report would be great PR, demonstrating the company’s commitment to community safety and well-being, and would be about 6,472% better than the bland emails they kept sending that essentially just said, “We’re working on it. Please be patient. It’s gonna be awhile…” without any useful information.
Learn more from Bill Shander in his session “From numbers to narrative: Data storytelling and visualization for the communication pro” at the IABC World Conference, happening 3–6 June 2018. Register today.
About the author
Bill Shander is an information designer, helping clients turn their data into compelling visual and interactive experiences. Clients include the World Bank, United Nations, International Monetary Fund, American Express, PricewaterhouseCoopers, and Facebook. He is the founder of Beehive Media, a Boston-based data visualization and information design consultancy. Shander teaches data storytelling, information design and data visualization on LinkedIn Learning & Lynda.com and in workshops around the world.