Pennsylvania Violation Data
The Pennsylvania Department of Environmental Protection (DEP) responded to our request for oil and gas violation information with a list of 9,370 violations from January 1, 2007 through September 30, 2010. That works out to almost seven oil and gas violations statewide per day. Certainly, the DEP has been taking its job very seriously in trying to regulate the oil and gas industry within the Commonwealth.
At the same time, it makes one wonder what violations occur that the DEP is not aware of.
A Closer Look at the Numbers
The 9,370 violations came from a total of 3,661 wells. Of that total, 2,075 violations are flagged as Marcellus Shale, from 592 distinct wells. And of the Marcellus wells total, there were 1,497 violations from 399 wells that were flagged as horizontal wells. All horizontal wells in this dataset are also Marcellus Shale wells.
That means that for non-Marcellus Shale wells, there were 7,295 violations from 3,069 wells, for a frequency of just 2.38 violations per well. The highest frequency of violations per well was 37, for API# 131-20020, a horizontally drilled Marcellus Shale well in the town of Washington in Wyoming County, PA.
In terms of average violations per well, the reported data indicates 2.38 violations per non-Marcellus well, and 3.51 violations per Marcellus well, for an overall average of 2.56 violations per well.
Please keep in mind that all of these values only represent wells that have violations, and therefore it should not be construed that the average Marcellus well will have three and a half DEP violations. This certainly does, however, raise the question of why an offending horizontal Marcellus Shale well would have an average of 1.37 more violations per well than its non-Marcellus counterpart, but it does not provide the answer. Perhaps it reflects that more can go wrong on horizontal Marcellus Shale wells, or that they are more tightly regulated. Perhaps a handful of individual sites like the one in Washington have so many violations that they skew the data in some way. There could be numerous scenarios to explain the discrepancy.
Obtaining Geographic Location Information
To date, the primary means of linking oil and gas data to a geographic location has been using the dataset compiled by PASDA, which includes the information of over 123,000 oil and gas locations within the state. Unfortunately, the PASDA data was insufficient for the violation data, as there were thousands of records that did not match. I have reduced that number significantly to by adding the list of known permits since 2007 (which also includes geographic information) to the PASDA list. This process created duplicate entries, which was worked around by obtaining the average longitude and average latitude of each API number.
This process does not eliminate any errors inherent in the data. In one example that I looked at, there were three listings for one unique well number, one of which was off by 0.3 degrees of latitude, or about 20 miles. In this scenario, it would be tempting to correct the one to conform to the other two, but in reality, it isn’t clear which one is correct. And for that reason, auditing the dataset—which is now over 128,000—for errors is not especially productive. For the moment, it is sufficient to say that there are some errors present in the location data, which are hopefully minimal in scope.
>Thanks, Matt, for your analysis. Have same Q as Conrad and Anonymous above: Any clues on specifics of violations, like basic categories (clerical vs operational like bad well bore casing cementing, spills affecting streams, etc) plus a few specific example of operational violations? Assume DEP didn't send them to you. Any idea where we can link to DEP database to get these specifics on violations?
The repeat violators would be interesting to see which companies, how many violations with dates and specific nature of violations. I say this because repeated violations speaks to a pattern of potential negligence that might best to treated with suspension of permit for x number of months rather than fines since we know gas co's have more money than God and fines are little disincentive to bad performance.
>Is there a way to determine the nature of these violations?
>Thanks, that link should be fixed now.
>The 'permits' link is broken
>Matt this is a masterful assessment and visualization of a huge dataset–it should be used to inform policy and siting requirements for all oil and gas wells. It would now be interesting to look at visualizations by well type and try to get the denominator in by looking at all wells–do Marcellus wells have more violations than non-Marcellus wells in total? What are the major categories of violations for Marcellus and non-Marcellus wells? Are there more serious violations for Marcellus than non-Marcellus wells? So many questions now. So what you did was the first step in inguiry–
I really would like people commenting on this blog as I see a way to ask questions about datasets as a group which is what science is really all about–Dan