This article is from the archive of our partner .

In a letter responding to the State Department's draft environmental assessment for the Keystone XL pipeline, the EPA finds several areas it deems insufficient. Perhaps the third time's the charm on State doing something about it.

The pipeline has become a focal point of the national — and, to some extent, global — environmental movement. Proposed by TransCanada, Keystone XL would shunt diluted bitumen, a sludge from which petroleum products can be developed, from Alberta, Canada, to the Gulf Coast. It's a 1,179-mile project, running through the heart of the country. To TransCanada and fossil fuel advocates, it represents a needed upgrade to the country's infrastructure. To environmentalists, it represents a major escalation of the systems behind and contributors to climate change.

Because the pipeline crosses an international barrier, the State Department is tasked will final permit authority over the project. And since it's in charge, it also has to complete an environmental impact analysis. In March, the agency released its latest draft, which indicated that State felt comfortable that the project wouldn't dramatically increase greenhouse gas emissions and that a new proposed route that avoids the sand hills of Nebraska reduced the risk of a spill polluting a critical aquifer.

Each time an agency completes an environmental impact statement (EIS), the EPA reviews and comments on it. Three times the EPA has reviewed State's assessments for Keystone. And three times — in 2010, 2011, and yesterday — the EPA has found the assessments to be either insufficient or inadequate. In this case, "The EPA review has identified significant environmental impacts that must be avoided in order to provide adequate protection for the environment," in part because "the draft EIS does not contain sufficient information."

There are four specific reasons that the EPA faults the current iteration of State's draft EIS. (One, asking for more details on providing water to at-risk communities in the event of a spill, is fairly straightforward.)

It underestimates how much greenhouse gas will be created. This is a critical argument for opponenets of the pipeline. The product derived from the tar sands in Alberta, bitumen, is much more carbon-intensive to extract; State estimates it requires 81 percent more carbon dioxide emission to get out of the ground. But State's analysis relies upon the assumption that the bitumen will be extracted regardless of whether or not the pipeline is built — an assumption that has been challenged. Without the pipeline, it may not be economically feasible to extract the product. And if the product is never extracted, its carbon footprint drops from an estimated 935 million metric tons of carbon dioxide over 50 years to a flat zero.

It downplays the risk of spills. In 2010, pipeline company Enbridge experienced a major spill near the Kalamazoo River in Michigan, the largest on-land spill in American history. What spilled is diluted bitumen, dilbit, the same product that Keystone XL would carry. And it's much harder to clean up.

In that spill, oil sands crude sank to the bottom of the Kalamazoo River, mixing with the river bottom's sediment and organic matter, making the oil difficult to find and recover. After almost three years of recovery efforts, EPA recently determined that dredging of bottom sediments will be required to protect public health and welfare and the environment. This determination was based in large part on demonstrations that the oil sands crude associated with the Enbridge spill will not appreciably biodegrade.

One problem Enbridge faced was that it took 17 hours for the pipeline to be shut off after the company ignored spill warnings. The EPA is suggesting that monitors for spill detection be greatly upgraded, in order to minimize the amount of dilbit that might leak — and that TransCanada be required to regularly monitor groundwater near the pipeline.

It doesn't consider alternate, safer routes. The main concern regarding a spill is what's called the Ogallala Aquifer, a massive underground repository of fresh water covering the Plains states, replenished slowly as rain and surface water seeps back down into it. In eastern Nebraska, that seeping happens more quickly through the Sandhills, which act as a natural sponge. When TransCanada first proposed running the pipeline through that area, the state — including its Republican governor — objected.

The pipeline's new route avoids the Sandhills, but still goes over the aquifer. The EPA would prefer it not, regardless of the economic impact.

The Keystone Corridor Alternatives were determined not to be reasonable alternatives primarily on the basis that these routes are longer than the proposed Project's route, and that additional pipeline miles would be needed to connect to Bakken MarketLink project, which would allow the proposed Project to also transport crude from North Dakota and Montana. As we have indicated in the past, we believe these alternative routes could further reduce risks to groundwater resources.

One criticism, popular among pipeline opponents wasn't included in the EPA's review: revelations that two consulting firms with ties to companies that stand to profit from the pipeline contributed analysis to the draft EIS. Nonetheless, the Natural Resources Defense Council's Anthony Swift said the EPA's response was "exactly right."

The extent to which the State Department incorporates the EPA's feedback remains to be seen. History may provide some guide. In 2011, the EPA's response to a draft EIS asked for it to include analysis of the effects on wetlands and migratory birds. Oh, and also concerns it had about oil spills, greenhouse gas emissions, and the effects of a spill on at-risk communities. The EPA rated that draft EO-2, meaning it had environmental objections and that the report lacked sufficient information. That's the same rating it issued yesterday.

This article is from the archive of our partner The Wire.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.