Finger pointing frustrations
Well, the New York Times finally decided to dial in to the ongoing seismic mess in Oklahoma. And while the coverage highlights the potential conflicts of interest and ability of the oil and gas industry in that state to dampen if not entirely prevent criticism of its operations, it doesn’t exactly shed a lot of light on the problems of saying why there are these earthquakes and it doesn’t help folks to understand how these earthquakes might be connected to the wells in question. Rather than get into the mud and argue the details in Oklahoma (you can read the 2014 Science paper if you want to see how that is done), consider a much simpler system.
For the past 24 years, the Bureau of Reclamation in southwest Colorado has been pumping saline water into a deep aquifer (the goal is to reduce the salinity of Colorado River water reaching Mexico). Unlike most (if not all) of the injection wells associated with oil and gas work, the seismicity of this region was monitored prior to the beginning of injection and has continued to be monitored ever since.
Not surprisingly, small earthquakes started showing up immediately when injection started, and more or less the region influenced by injection has expanded over time (see plot below from Block et al., Seism. Res. Lett., 2014)
What is clear from this work is that the effects of the injected fluid are far from simple. In particular, the generation of seismicity north of the valley without much in the way of intervening events shows just how complex these systems can get. Drop another 10,000 wells on top of this and see how easy it is to figure out which well is responsible for which earthquake.
Now the New York Times’s explanation for this was, um, a bit opaque: “The mechanics of wastewater-induced earthquakes are straightforward: Soaked with enough fluid, a layer of rock expands and gets heavier. Earthquakes can occur when the pressure from the fluid reaches a fault, either through direct contact with the soaked rock or indirectly, from the expanding rock.” Most of this is somewhere between wrong and misleading.
There are two ways to get earthquakes from messing with fluids: changing the stress field in the earth and changing the pore pressure. In the first, the weight of a mass of water will increase the vertical normal stress–basically increase the weight on deeper layers. With the extra weight, these layers will want to thin vertically and expand horizontally (try squishing a ball of Silly Putty more from the top and bottom than the sides and you will get the same result). If the stress field is extensional, you can generate earthquakes this way. Some earthquakes induced by filling large reservoirs (e.g., Lake Mead) are caused in this manner.
More common, and probably most relevant in Oklahoma, is the change in pore pressure. If you were to slice open a rock at depth and push on it to keep it from deforming, the force you would have to use would be fighting two aspects of the rock: how much the mineral material in the rock was pushing out, and how much any fluid in the pores in the rock was pushing out. This second parameter, the pore pressure, is the likely culprit here. Injecting fluid will change the pore pressure in the surrounding rocks, and just as atmospheric pressures change without moving the entire atmosphere, so can adding fluid in one area affect the pressure elsewhere without the need for the newly added fluid to travel all the way to some area that has a pressure change. If the pore pressure increases, the normal stress on the mineral part of the rock lowers. But any shear stresses acting to try to make the rocks slide across a fault are unaffected (water cannot support a shear stress), so the lower normal stress on the rock combined with the continuing shear stress means that the rock will more easily fail on any given fault line. It is a little like hydroplaning when driving a car in a rainstorm; the water is absorbing some of your weight, which reduces the frictional forces that allow you to change direction or slow down. Slipping is a lot easier.
So one big challenge is to pin the tail on the right donkey (something that 2014 paper sought to do). Ideally we could anticipate when this might happen, but it seems we have thousands of injection wells running in this country that seem to have no impact at all (well, at least not yet). So while it seems nearly certain that some injection wells are causing significant earthquakes, and we may be gaining the ability more and more to figure out which wells are most likely the culprit, we are still quite a ways from knowing ahead of time if a well will cause earthquakes. In the interim, it would be best if regulators and injection companies kept the possibility of having to shut down a well in every well’s operational plan.