For many data journalists, the tricky part of working with climate data is that much of it comes bundled together in multiple dimensions of time and space. For instance, each tiny grid square on a map can hold different information about temperature changes over time.
And this “gridded climate data” often appears exclusively in a file format called NetCDF, which is very different from spreadsheets, and unfamiliar to many reporters.
This is perhaps why several attendees at the recent NICAR21 data journalism conference pointed to a NASA data viewing app, called Panoply, when GIJN informally asked them to name the single most useful tool they’d heard about on the event’s opening day.
That’s saying a lot, because every session of NICAR — which stands for National Institute for Computer-Assisted Reporting, and is organized by Investigative Reporters & Editors — practically bristled with useful data tips and digital tools, as it has done since these annual conferences began in 1989.
In a virtual session titled Using Data to Report on Climate Change, BuzzFeed News science reporter Peter Aldhous showed how gridded NetCDF data files can be easily uploaded into the free Panoply tool, and how you can then use it to make customized maps for more accessible climate stories.
He was joined on the panel by NPR science reporter Rebecca Hersher and ProPublica environment reporter Lisa Song, who shared tips on little-known and niche data sources that can make climate stories both easier to report and understand.
Among their tips were ways to strengthen reporting on sea level rise, measure coastal sea temperatures, model future flooding, track hurricane damage, and customize climate data from around the world.
Sea Level Data
Hersher says the twin challenges of stories about sea level rise are, first, that many audiences don’t realize that ocean levels, and rates of change, are significantly different for local coastlines around the world, and, second, that reporters often don’t know how to find local sea level data.
She says an excellent source for local data can be found within the National Oceanic and Atmospheric Administration’s (NOAA) Tides and Currents Data Trends, which offers global data on local sea levels.
“This data is really easy to access,” says Hersher. “You can easily see, for instance, that sea level rise is a lot faster in Annapolis, Maryland, than it is in Santa Monica, California.”
Hersher points to the Rising Waters series by South Carolina’s Post and Courier outlet as an “awesome” example of how to use NOAA’s localized data, and of climate change coverage in general. The data they found for the coastal city of Charleston showed that the local sea level rose 1.4 inches (3.6 centimeters) from 1990 to 2000; another 2 inches from 2000 to 2010, and a further 2.7 inches from 2010 to 2020. The story found that, as a result, the “nuisance flooding” — flooding at high tide — that happened about four times per year half a century ago now happens about 40 times per year in the city.
A Hidden Sea Temperature Dataset
Hersher says reporters also find it “weirdly hard” to find good data on coastal sea surface temperatures. Although the dataset’s name doesn’t suggest it, the NOAA Coral Reef Watch site is an excellent — and underused — source of this data, she says.
“I think one reason people don’t use this resource is because it’s called ‘Coral Reef Watch,’ but it can be really useful for data not really about coral,” Hersher explains. “You can find what they call daily SST anomalies, which [are] temperature anomalies from normal for the top layer of the water, and in relatively high resolution. NOAA even makes their own gifs of this data that you can just download. I love the scale they use, which a second grader [can understand].”
Hersher used these datasets for an NPR story on the damage caused by the hottest year on record — 2020.
Modeling Future Floods
For local data and visualizations of flood planning, Hersher says she recently stumbled upon hydrological models — which are advanced computer models sometimes commissioned by local authorities to plan for things like flash floods.
“These are all about where water is going to go when certain amounts of water fall,” she notes. “They can be extremely helpful. The problem is you’ll never know these models exist unless you ask — maybe a planning or public works department, or a storm water agency.”
However, Hersher notes that many officials are glad to share hydrological model data with reporters simply because they are expensive to commission, and there is value for them in seeing it fully utilized.
NPR used these models for a major story about flash flood planning in historic Ellicott City in the US state of Maryland.
Although ProPublica’s Lisa Song offered the NICAR21 audience datasets specific to the US — such as the USGS Streamgaging Network, and the National Flood Insurance Program from the Federal Emergency Management Agency (FEMA) — her tips on how to target lesser-known climate data sources could be useful to journalists in many countries.
- Look for stream gage data for both river depth and stream flow rates from government science agencies, and use these two datasets in combination to show changes over time. “These gages are used in extensive networks — things they put in rivers that measure the height of the water, and other gages that measure the stream flow per second,” says Song. “You can use both kinds of data to do some interesting analysis on flooding trends and historical trends.” ProPublica used gage data extensively for their collaborative investigative project with Reveal — titled Flood Thy Neighbor — that found “life-and-death decisions [about levee protection] are dictated less by sound science than by economics, politics, and luck.”
- In comparing stream gage data for historical trends, be careful, however, to check with the relevant survey official to ensure you’re making an ‘apples-to-apples’ comparison. “They might have moved a gauge a mile upstream or downstream, and needed to recalibrate their zero-points,” she notes. “We have found that they don’t always remember to calibrate the data properly.”
- Look for niche datasets — especially from academic sources — that can show climate change damage and costs that are otherwise difficult to measure. For instance, in one story about the huge costs of replenishing beaches damaged by hurricanes, Song and her colleagues used a niche dataset maintained by Western Carolina University that counts the truckloads of sand used by the US Army Corps of Engineers to rebuild beaches eroded by increasingly violent storms.
- Look for flood insurance or emergency assistance data to measure patterns. In the US, for instance, Song says FEMA produces detailed data for its National Flood Insurance Program, which provides federal relief for households repeatedly struck by flood damage. While this vast dataset is hard to access, Song says reporters can request parts of that data from the Natural Resources Defense Council (NRDC) environmental group, which successfully sued to obtain the database.
- Reach out to press-friendly organizations that actively seek to help reporters, like Climate Central — an independent organization built around climate scientists and journalists that offers a partnership program for radio, TV, and digital newsrooms. “Climate Central is a good general source of climate data, and they also have an entire program that tries to help local reporters — and broadcast reporters in particular — access and visualize all kinds of climate data,” Song says.
- Look for real-time sampling data after environmental disasters from advocacy groups, academic institutions, and private companies. Song says that following Hurricane Harvey in the US in 2017 one private company drove its own mobile air monitoring lab around an oil refinery damaged in the storm. She says the data the company found on benzene in the air was more transparent and comprehensive than the data by the US government watchdog, the Environmental Protection Agency.
- When you find a great gas emissions database, dig deep. Song points to a major 2019 story by the LA Times which showed that California’s wildfires cancelled out all the environmental gains made by the state’s emissions reduction laws as a great example of digging deep into comprehensive greenhouse gas emissions datasets.
Handling “Gridded” Climate Data
To illustrate how prevalent NetCDF files are in climate science, BuzzFeed’s Peter Aldhous did a search of one major climate science source — the National Center for Atmospheric Research Climate Data Guide. He found that 134 of the archive’s 214 total datasets were presented in the NetCDF format.
He says these files often show both information and time periods for each map grid cell, such as 2 degrees latitude by 2 degrees longitude. For instance, he notes the NASA GISTEMP analysis publishes surface temperature information for these hundreds of grid squares monthly — for the whole planet — with the archive going back to 1880.
“If you’re familiar with spreadsheets and databases, NetCDF is probably a weird format to think about working in — that dimension of time and two dimensions of space,” says Aldhous. “But there is software to make it really easy to look at this format — Panoply. It’s multi-platform — you need Java on your computer, but otherwise that’s about all. You can download it easily, and start running it.”
Once you’ve opened a NetCDF file in Panoply, Aldhous suggests that you:
- Select the data layer or layers you’re interested in, and create a plot.
- Use Panoply’s default longitude and latitude plot program to create a map.
- Use Panoply’s user-friendly filters to customize your map’s scale and color palettes.
- Have fun choosing options you think will best illustrate your dataset, including overlays of country borders, and projection types for the map itself.
- Label your map, and save it either as an image or as an animation.
“It’s pretty easy to use, and it allows you to access the raw data on things like sea surface temperatures,” Aldhous explains.
For reporters with some coding experience, Aldhous recommends processing gridded climate data in the Raster package in the R programming language.
“I do most of my climate work in R, and I use the Raster package — it’s incredibly powerful for working with this data,” he says. “I hope those who have worked with code will look at this package and think ‘that doesn’t look too scary.’”
Rowan Philp is a reporter for GIJN. Rowan was formerly chief reporter for South Africa’s Sunday Times. As a foreign correspondent, he has reported on news, politics, corruption, and conflict from more than two dozen countries around the world.