We started by downloading data on every Marcellus Shale well permit from the Pennsylvania Department of Environmental Protection's website and mapping it so readers could find out where wells might be going in. Later we added data about fines levied on drillers, and on which wells were eventually drilled.
Then we decided to include the state's production data, but we ran into an issue that has had ripple effects from DEP, to drillers, to environmental organizations, to the state Legislature.
When PG Web content producer Laura Schneiderman downloaded DEP's production data, she discovered it says there are 495 more wells producing gas, or ready to produce gas, than DEP has recorded as ever being drilled, and 182 of those wells don't even show up on the state's Marcellus Shale permit list.
These aren't just factoids that only a techno geek could get excited about, either.
The data discrepancies and other significant problems with DEP's Marcellus Shale data have caused problems for information companies, environmental organizations and drillers that rely on it to analyze the industry. The discrepancies have caused headaches for Senate and House staff who have been trying to make accurate projections about how much revenue an impact fee on wells might generate for local governments, and where.
"There has been a frustration over the last six or seven months that DEP does not have information that is always beyond reproach," said Drew Crompton, chief of staff to Senate President Pro Tem Joe Scarnati, R-Jefferson.
'Significant error rate'
Mr. Crompton, who has tried to make sense of DEP's data as the Senate began crafting an impact fee bill last year, said the information problems are so befuddling that it helped delay approval of the bill. Legislators simply haven't been able to get accurate projections on the financial impact.
"Every time I think I've got something locked down, it changes," said Mr. Crompton, who discovered the same data issues that the Post-Gazette did.
Given that the state says that since 2007 there have been about 4,200 wells drilled, 495 wells is about a 12 percent error rate on the wells drilled data that is so widely quoted by politicians, environmentalists and the industry alike.
"That's a significant error rate," said Bruce Stauffer, vice president of geographIT, a Lancaster-based company that provides geographic information services to industry and governments.
His company also ran into the same problem with DEP's data when last year it began putting together Marcellus monitor, the company's interactive mapping tool that it sells to companies and governments.
"It's obvious DEP's data isn't clear and accurate," he said. "Why? I don't know. And I don't think they have the answers."
When the Post-Gazette first asked DEP to explain the discrepancy, the department would not take questions about why there was such a large error.
Because the production data is "reported to DEP by the operators," spokesman Kevin Sunday wrote in an email response to questions in late October, the Post-Gazette should "rely on the production data as to the 'Marcellus' or 'non-Marcellus' classification of a well."
Both industry and environmentalists said that relying only on industry-provided data poses problems.
"You want to be able to rely on state data as the most accurate," said Davitt Woodwell, vice president of the Pennsylvania Environmental Council, which has been studying the Marcellus Shale industry and found data problems itself. "It makes for a better understanding of issues if you have a lot of good data from the state -- the good and the bad."
Range Resources, the largest Marcellus driller in southwest Pennsylvania, has staff go over all the available state data and draws its own conclusions about what its competitors are doing, rather than rely on Pennsylvania DEP's numbers, said Carl Carlson, Range's director of government affairs and a geologist by training.
"In a perfect world" drillers could rely on the state information, said Mr. Carlson, whose company had 104 wells on the list of 495 wells the Post-Gazette identified. "But we have our own internal list our guys go over. Our list is different than the state lists and we think we have a pretty good handle on how many wells have been drilled."
Part of Range's distrust of the state data comes from obvious examples, he said.
Just by comparing a well's listing on the state DEP's wells drilled list (called a "spud report") to a listing on the Pennsylvania Department of Conservation and Natural Resource's Internet Record Imaging System he said: "I've noticed numerous errors about whether a well is horizontal or vertical."
"If you look up a well on IRIS you see a well clearly listed as horizontal, but it's listed as vertical on the spud list," he said.
That conflict -- between vertical and horizontal Marcellus wells -- also posed problems for legislative staffers, who were trying to figure out if impact fees for vertical wells should be cheaper or the same as for horizontal wells.
But we still haven't been able to get an accurate count of vertical Marcellus wells," Mr. Crompton said.
Since October, six drilling companies, which collectively drilled the overwhelming majority of the 495 wells on the list, all confirmed to the Post-Gazette that their wells on the list were drilled into the Marcellus.
While the drilling companies couldn't figure out why most of the wells didn't register by the state as wells drilled, several said that some of the 182 wells that weren't on the permit list appeared to have been vertical Marcellus wells.
Last month, DEP agreed to check the list and it confirmed that those 182 wells do appear to be vertical Marcellus wells that simply were never recorded properly as Marcellus -- most of which are horizontally drilled wells.
The Post-Gazette also ran into a data problem the state only recently took notice of that made checking either the wells drilled or production data impossible.
The most accurate report about the intentions of any well is the well completion report, which is supposed to be filed within 30 days after a well has been completed. It not only says a well is completed, but indicates the depth of the well, as well as how far it went vertically, when it hit the Marcellus Shale rock, how far out the well was perforated, and other information.
DCNR geologists are supposed to receive a completion report for every well drilled in the state from DEP, analyze it for accuracy, and then tabulate the data and scan in the completion report itself and make it available on its Internet Record Imaging System, a site that requires a fee to use.
However, drillers routinely fail to file the reports with DEP. As of December, DCNR had only received 1,222 well completion reports since 2007, compared to about 4,700 wells that apparently have been drilled.
But starting in March last year, for the first time, DEP began fining drillers for failure to file well completion reports on time. At least 11 drillers have been fined up to $5,000 for the violations.
The data problems span both the Ed Rendell and Tom Corbett administrations.
Data collection and reporting errors were "something identified through the transition period in the first few months" of Mr. Corbett's term as governor in early 2011, Mr. Sunday said. "And it's one we hope to clear up and get more consistent at."
"We acknowledge that there are issues in both how the data is presented and how it's coming in," he said.
He said the problem with the wells drilled data appears to be because of the way that information is collected. In some cases it is called in or emailed in by the companies, but in other cases it is called in or entered by state inspectors who visit the sites.
"So, with so many ways of reporting, we have a risk of error," he said.
DEP also has seen confusion about how to categorize vertical Marcellus wells, since most Marcellus wells are horizontal and "we're trying to clear that data up, too," Mr. Sunday said.
John Hanger, DEP secretary the last two years of Mr. Rendell's administration, said he's not surprised that the state is finding problems "with such a high number of reports coming in."
"Record-keeping issues are important, and I'm glad the department is taking it seriously. It's important for the community and for the industry to have accurate information," he said.
Ultimately, the problems may be rectified because of the impact fees, Mr. Crompton said, because all of the House and Senate bills on the issue include extensive provisions for auditing and fact-checking data.
If the impact fee bill passes "there will be quite the mechanisms in place to know where all the wells are and what's happening with them," he said.