Einstein (the first superstar recognized by just one name) said, “Everything should be made as simple as possible, but not simpler.” This is sage advice for us all as we are about to embark on a national sausage-making discussion concerning the appropriate standard(s) to use for the control of stormwater runoff volume through the use of green infrastructure. In a recent discussion EPA stated that is looking for input on ways to make things appropriate and consistent, yet flexible—a laudable goal…if we can attain it.
Of course, the devil is in the details. So, to add a bit of spice to the mix, let’s talk about a few of those details.
There seem to be three main approaches vying for nomination as the “go-to default” standard for a volume-based green infrastructure criterion: percent capture, depth capture, and hydromodification.
Percent capture can be summarized as “capture and remove the “˜XX’ percent storm.” “XX” has been put forward as anything from about 80% to 95%. Seems simple enough, right?
Einstein might cringe.
First of all, what do we mean by percent capture? EPA guidance for federal compliance with the Energy Independence and Security Act Section 438 suggests (1) lining daily rainfall data up small to large, (2) deleting all daily values < 0.1 inch, and (3) assigning a percent-smaller-than value to the remaining daily values. The stated supposition is that this appropriately mimics the natural hydrology of any site (EPA 841-B-09-001, December 2009).
On the face of it, and with some thought, you might arrive at the conclusion that this is not all it seems. Setting aside the sort of arbitrary assumption that this procedure correlates consistently to percent volume capture—which it does by accident—there are two significant problems with it when it comes to appropriateness and consistency. I will use Tennessee data to illustrate.
First is the idea that there is a missing time element necessary to sound design. The 95% storm in Nashville is 1.78 inch. The standard implies that I have to capture 1.78 inches per daily rainfall amount regardless of when it falls or how hard. I have arbitrarily broken all storm events at midnight. If I remove the arbitrary midnight break (as nature might chose to do) and capture the 1.78-inch depth on a per-storm basis rather than a calendar-day basis, my capture rate goes from about 94% to 85%. If I use the Tennessee standard of the first runoff with a 72-hour inter-event dry period it drops to 74%. However, we know that, except for cisterns, there is in-storm infiltration—but how much is never discussed.
We also have to consider the idea that the same percent storm applied to different places gives inappropriate and inconsistent results. In Tennessee, the 95% storm is 2.03 inches in Memphis, 1.78 inches in Nashville, 1.45 inches in Knoxville, and 1.25 inches in Tri-Cities. If the state of Tennessee required the 95% capture for all four cities for absolutely identical sites—say type C soil—Memphians would have a hard time stuffing 8 pounds of sausage in a 6-pound skin, while those in Tri-cities might skate by easily. In Phoenix it is less than an inch—talk about a retirement destination.
This seems simpler than possible.
Depth of capture has similar problems. If we assume, for example, that the standard is to fully capture the first 1 inch of rainfall, then here is the problem. First of all, if it rains really, really hard, nature cannot capture the first inch. Some amount of runoff would constitute hydrologic mimicry. Continuous simulation of hourly rainfall for 40 years in Nashville demonstrated, for example, that only type A soil can capture the first inch of rainfall all the time, type B soil almost all the time, type C soil some of the time, and type D soil almost never. No natural soil complex extant in Nashville can capture an instantaneous 1 inch of rainfall (“kerplunk”). None.
How is it then that we require a type D soil site to handle rainfall just like a type A soil site?
As in the previous example, how does the “time” factor into this criterion? The median time it takes to accumulate 1 inch of rainfall after 72 hours of no rain in Nashville is over 30 hours. In Philadelphia’s test year (2005), three storms attain 1 inch in an hour or less, five in six hours or less, and all the rest stretch out much longer.
Yet we are requiring designers to capture 1 inch of instantaneous rainfall regardless of the underlying natural soil type. Is this hydrologic mimicry? Nope—it can end up looking more like grey infrastructure painted green. When instant volume capture is the name of the game, then grey tanks work best. Trees are of no value, having only about 0.07 inch of instantaneous capture.
A modification of this method transforms the instantaneous depth into a treatment volume and then gives “credits” for all sorts of things that may not have much instantaneous depth capability but are good green approaches nonetheless. The resultant mashup of two ideas may work fine, though I have found lots of only-part-time stormwater practitioners scratching their heads over its complexity.
This too seems simpler than possible…or harder.
Hydromodification (or some of its earlier channel protection cousins practiced around the country) shows great theoretical promise if most of the problem a community faces is the unnatural erosion of urban stream banks and beds. In this case, matching the accumulation of “work” or flow as a surrogate for sediment transport capacity over time (impulse analysis or stream power as it was called in the 80s) to some target value is the goal.
Of course, this can be very complex and site specific–requiring fairly complex geomorphic and sediment transport continuity analysis in each case or stream type. In some places, trying to return the stream to a natural (pre-settlement) condition is neither possible nor advisable. And in many places, it is not applicable or of interest for good reasons that some hydromod evangelists might not appreciate. In nearly every case, actual volume reduction on the scale required is fairly expensive, involving very-extended detention treatment of portions of the runoff in an attempt to mimic a more natural hydrograph recession curve for about the 0.5Q2 through Q10—flows well above those handled by green infrastructure.
If this approach is used to the exclusion of actual volume treatment or removal (i.e., green infrastructure), then it cannot solve the overall pollution removal (remember that old fashioned goal?) issue—only the lower flow regime runoff hydrograph mimicry one. Pollution removal is secondary. But when done well in communities with lots of space to grow it can push streams back toward stability.
Is there another way? What is the Holy Grail of volume-based hydrology?
There are a lot smarter people than me on both coasts (I have names if you want them) who are wrestling with this and, perhaps, coming to new conclusions. But I do have a couple thoughts to throw into the pot.
First of all, you must calibrate any approach locally. How does nature perform her feat in your area? Can she be copied? Should she be?
Deep sandy soils? Lots of rainfall never runs off. Considerable interflow pathways? Lots of rainfall returns to the stream system in the hours and days following rainfall. Type D soil over bedrock? Pavement has little impact.
Second, I think there might be a better parameter than either depth or percent, and that is average annual runoff, as reflected in a single coefficient—say Rv. We have found through lots of continuous simulation modeling that when we attain the target Rv we also tend to mimic the hydrology across a broad spectrum of storms. And it is easy and inexpensive to establish what the target Rv should be for any set of natural conditions. In Nashville, for example, type C soil with grass and trees (the typical Nashville lawn) has an annual Rv of 0.2 to 80% capture, and the trendline shows that a 1-inch storm of median duration is typically fully captured. But not every 1-inch storm is captured. This meets the state standard and the design criterion is easy to figure—any engineer who has ever done a Rational Method C Factor calculation can handle it.
There may be some real problems with this approach not yet encountered. The jury is still out, though several of us put on our wily developer hats and tried unsuccessfully to break it.
Find out more about the approach adopted in Nashville here.I think in the coming couple of years, if we all push our pieces of the puzzle to the middle and let others help put them into the proper place, a clearer and fuller picture of a simple, flexible, locally calibrated, and yet appropriate and consistent criterion-setting approach may emerge…simple as possible…not simpler.