State of the Planet

News from the Columbia Climate School

Solving Urbanization Challenges by Design – The Science of Green Roofs

Patricia J. Culligan is a professor of civil engineering and engineering mechanics at Columbia University and the Vice Dean of Academic Affairs for Columbia Engineering. Her research focuses on the movement of liquid through porous media—an interest that she has applied in diverse contexts, from tracking pathogens in the shallow aquifers of Bangladesh to metal toxicity in rainbow trout. She has worked extensively with The Earth Institute, in particular the Urban Design Lab, which focuses on “a design-based approach to shaping the long-range future of sustainable urbanism.”

In this interview, Culligan discusses her work with the Columbia University Green Roof Consortium to quantify the benefits of green roofs—a new approach to roofing in dense urban environments that has been heralded as a way to insulate buildings, manage sewer overflows from stormwater (by reducing the volume and peak flow of water that flows into sewers during storms) and mitigate the urban heat island effect.

More broadly, green roofs are seen as a key component of “green infrastructure,” a design approach which attempts to make use of natural and biological systems, including constructed wetlands and other ecosystems, to manage the urban water cycle. Green infrastructure has recently been touted by many organizations, including New York City’s Plan NYC, as a cost-effective and more ecological alternative to the conventional “grey infrastructure” approach that makes use of pipes, cisterns and other hard-surface approaches to storage, diversion and dilution of urban water overflows.

Always the scientist, Culligan believes that if green roofs are going to be applied on a wide scale, their benefits need to be quantified—both to provide a better understanding of how they work and to determine which of the competing green roof technologies should be used to achieve different goals.

In part one of this interview Culligan discusses the difficulty of calculating the economic benefits of green roofs, why results don’t necessarily scale between different sized systems and how lowering peak runoff might be more important than overall flow if one is worried about urban water quality.

What is the Columbia University Green Roof Consortium?

It’s a group of researchers and research students at Columbia University who are interested in understanding the role that green roofs might play in urban sustainability– I’m one of the researchers—the others include Stuart Gaffin, from the Center for Climate Systems Research, Wade McGillis, who’s a research professor at the Lamont campus, and Matt Palmer who’s in E3B [Department of Ecology, Evolution and Environmental Biology]. We have strong collaborations and ties with the Office of Environmental Stewardship and work jointly with the Earth Institute’s Urban Design Lab and the Education Center for Sustainable Engineering. Our team of research students includes both undergraduate and graduate students from multiple disciplines.

A green roof monitoring station at Columbia University

We’re interested in quantifying the performance of green roofs in an urban environment from multiple perspectives. We currently have a suite of greenroofs instrumented around the city. The consortium really consists of the research group, the instrumented suite of greenroofs, and a lot of stakeholders–including Con Edison, the EPA, and several schools—all of whom are interested in understanding what impact they could have in ameliorating some of the environmental challenges we face in the city. My piece of that is looking at the potential benefits of green roofs from a water management perspective.

Specifically, we’re interested in the green roofs capacity to detain and retain stormwater and in the quality of water that outflows from the green roof. With respect to detention, one benefit is preventing stormwater from entering the combined sewer system of New York City, which reduces the chance of combined sewer overflow (CSO) events.

With respect to the water quality, that’s an interest that the EPA, among other stakeholders, has. Sometimes stormwater goes straight into local water bodies; the question is, if it passes through a greenroof system before it enters that local water body, is it cleaner, or not?

From initial observations we’ve found that in some areas–for example, in the addition of nitrates to water that passes through some green roof systems–green roofs might not be environmentally beneficial.

This would be nitrates from plant fertilizers?

Yes, exactly.

I understand you’re using a new medium for growing plants on roofs?

Well, the Green Roof Consortium is currently investigating a number of systems. We have one instrumented up at the Fieldston School–that’s a native system. We have one instrumented at Columbia–a pre-cultivated mat system–there’s the Con Edison building in Queens–a tray system–and a government office building, also a tray system. All of the growing media associated with those are different, and many are proprietary. We didn’t develop them.

We also have set up another investigation system that we call test boxes. They’re sort of small-scale green roof systems that may be a few feet by a few feet in plan area. We’re looking at how those small systems detain and retain stormwater, and also at the water quality that runs off of those systems. There are two questions that we’re asking with those experiments.

The first is, can these small-scale pilot systems give you any indication of large-scale performance?

And then the second part is actually testing the potential use of waste materials for the greenroof growing medium orsubstrate—so investigating the idea of designing a “green” greenroof. This idea has been proposed in a lot of European countries and I think there have been some successful pilot tests, but I think in the U.S. most the attention to date has been focused on getting acceptance for the greenroof technology itself, before pushing the envelope further and thinking about making “green” greenroofs.

So the conventional media used is synthetic?

It’s lightweight, and usually consists of soil mixed with non-organic aggregates and organic matter. The non-organic aggregates can be naturally occurring materials like pumice or they can be synthetically produced. There are examples of media that containpolystyreneto keep the weight down. I know of one system that uses waste polystyrene, so that’s an example of a greenroof system that makes beneficial use of a waste product that might otherwise go to a landfill. The use of virgin polystyrene is, of course, not so environmentally friendly.

Results of Combined Sewer Overflow. Source: American Society of Landscape Architects

You said most of the focus up until now has been on trying to get green roof technology accepted. What’s the status of that?

I think over the last five years or so, it’s become the new green arm of many building projects, including redevelopment and retrofit projects. But I think the pathway that Columbia University took is to some extent typical of some of the trepidation that’s felt about these systems–they put their green roofs on roofs that were in relatively good condition, because they didn’t want to yet rely on the green roof system as a roofing system until they had gained experience with the use of this technology. But ideally, in order for these systems to minimize cost-benefit, you would replace a traditional roof with a greenroof system at the end of its lifecycle–not put a greenroof on a roofing system that was in good condition.

Also, though they were interested enough to try the new system, Columbia chose the lightest weight system on the market. From the perspective of not adding much additional load to the building, it was the obvious choice. But someone like Matt Palmer might say that if you’re interested in ecological restoration in an urban environment, the thicker systems that can support native vegetation might be more effective.

My current understanding is that Columbia University have been pleased with the performance of their green roofs to date, and they’re rapidly gaining confidence with the use of technology. I believe that this is typical of many building owners who have also installed greenroofs I would not underestimate the initiative Columbia undertook in testing this technology, which was very much supported by Nilda Mesa of Columbia’s Office of Environmental Stewardship.

The green roof tray systems are also becoming quite popular. You often have the whole system integrated in each tray, including the growing medium, and you have the option of using sedum (a desert succulent) or alternate native plants as the vegetation. For this approach, you literally put a system of trays on your roof, so it’s not too difficult to upend the trays if you decide for one reason or another that you want to make alterations. Also, if you decide it wasn’t a good choice, you can change to another system, or go back to the traditional roof.

With the tray system, people are also able to put pathways between the trays so people can access the vegetated areas without walking on the plants. The pre-vegetated sedum mats that Columbia installed are like a green carpet of vegetation that covers most of the roof area and cannot be stepped upon. Columbia has created some pathways for us as a research team so we can access our instruments without harming the roofs were are studying but, in general, the mats are not designed to allow people access to the vegetation.

Native systems are a lot more robust, but understanding which native plants would flourish in a growing medium is not as simple as you would think either. Obviously native plants are used to the climate, but issues such as building shadowing that can happen in a dense urban environment can create conditions that do not always mimic the natural system that existed before man invaded.

What’s the state of the science on green roofs now? Do these systems work? Do we have evidence of that?

Well, we now have several years of stormwater data gathered from our small-scale systems and the full-scale roofs we have instrumented.

To date what we’ve seen is if we just look at the retention—basically the total runoff that a green roof released versus the total atmospheric precipitation –- for the small storms, they seem to be able to retain most of the runoff. The native systems performance is superior because they have thicker substrates—the trays systems in this respect also appear to be superior to the pre-cultivated sedum mat systems but as storms get larger the tray and mat systems behave in a similar fashion.

Then, there’s almost a linear increase in total runoff from the green roof with total precipitation, although the runoff from the greenroof still remains less than the precipitation. After this there follows almost a plateau, which is curious, in that the runoff doesn’t seem to increase even though the precipitation does. Following the plateau there appears to be an exponential increase in total runoff with total precipitation. In the larger storms the runoff from the green roof is almost equivalent to the precipitation.

But the system behavior is very complex because it depends on the preceding weather conditions. This means we have to be consistent in our definition of a storm or event. When there has been no rain for the past six hours our group defines it as an individual precipitation event. This means that you can have multiple events in 24 hours, if it has rained, stopped raining for 6 hours, and then started up again. Conversely, for the same amount of total precipitation you could have one event if the rainfall was somewhat continuous within that period. We are seeing evidence of a relationship between greenroof runoff behavior and observed precipitation in the prior 48 hours before an event.

We’ve gathered an awful lot of data. We’ve tried to compare it to software that was developed for modeling landfill performance, because there are similarities between landfill covers and green roof systems, but we don’t get very good predictive results, even though that model will allow us to put in antecedent moisture conditions and model the system layers. We have also yet to find a normalizing relationship that will tie into the behavior of different sized greenroofs together.

Can you explain?

So we’ve three similar roof systems, one on a brownstone, one on a residential building, and one in a small box. If we normalize (i.e., divide) the results we have collected by the different surface area of each system, for example, the normalized results between systems are not comparable. If we normalize by system volume, the same is true. So the scaling between systems is not directly related to simple calculations based on system area or volume

So results from a small-scale system cannot predict results from a larger system by simply taking into account the different areas of the two systems.

That’s right. It’s related to the flow-path length. So think of a raindrop that has to get from one corner of a large greenroof to the roof drain.  In a small-scale box, it’s basically going to drain downwards pretty quickly. Versus a larger roof, where it’s got to travel a reasonable horizontal distance to reach a drainage point.. But while it’s traveling, it can evapotranspirate. So the scaling is not as straightforward as one might think. Many of the prior predictions that people have done are based on an assumption that green roofs scale with area, and we’ve found that’s actually not correct.

The peak reduction also interests us. Peak flow is dramatically retained for most of these systems, over a wide range of storms–a 90 percent reduction in many cases.

What does that mean?

Well, you have the cumulative, or total, flow, which is the amount of water that comes off from the greenroof. That depends on the size of the event itself, it depends on the duration of the event, and it depends on the preceding events.

But then, there’s a peak runoff. So if you look a typical storm, there’s a buildup in intensity, and then you have a peak in intensity, when it’s raining as hard as it can, and then it drops off again. So the green roof also has a peak runoff. So with every storm, you see a peak runoff. Usually it’s delayed a little bit in time compared to the peak rainfall intensity, but not so much, really. They tend to coincide fairly well.

But we found that the reduction in peak intensity seems to be fairly constant over a quite wide range of storms. As opposed to the cumulative runoff, which shows a different behavior. But the reduction in peak runoff from the green roofs compared to peak rainfall intensity is dramatic. So the question is, if you’re interested in stormwater management, what parameter controls? Reducing the peak intensity in runoff or the total runoff?

Do we know the answer to that question?

We don’t really. All of the greenroof cost-benefit analysis that I have see published, , assume that greenroof performance scales with area and that it’s the cumulative runoff that matters.

We’re finding that performance does not scale with area, and I suspect that a reduction in peak runoff–for some storm types–is going to be much more important than cumulative runoff in controlling combined sewer overflow events.

We’ve also started looking at various economic models, because it is not clear to us how to do the cost-benefit analysis. What we’ve been doing for our initial analyses is looking at the number of dollars per volume reduction in CSOs–so dollars per gallon–and we can compare that with the dollars per gallon reduction in CSOs for gray infrastructure, i.e. stormwater detention basins, as an alternative.

What we’ve found is that you can divide the New York City sewer sheds into three categories: one where the CSO problem is not so severe, one where it’s of medium severity, and one where it’s pretty severe. And for the CSO watersheds where it’s not so severe, greenroofs are not an economic solution if you look at this method of cost-benefit analysis. For the ones that are medium severe, our calculations indicate that they’re probably a little bit more economic than gray infrastructure. For the very sever structure they’re much more economical.

But you’ve got to think “is this the right way of doing it?” Because the city is not penalized based on its CSO events, it’s penalized for not meeting water quality criteria. Okay, so then you say, maybe, dollars per eliminated CSO event is a better metric for exploring the cost-benefits of greenroofs. And most of the CSO events happen for the small storms that the greenroofs detain. So with this metric as an indicator greenroofs might be deemed quite economical.

CSO events happen for the small storms? I always think of them as happening with the big storms.

No, they happen with the small storms too, depending on the sewershed configuration. In some NYC sewershed 60 to 70 percent of CSO events are associated with small rainfall events. However, most of the gallons come out of the big storms. So that’s why The cost-benefit analysis changes when you trade dollars per CSO gallon eliminated for dollars per CSO event eliminated in the calculations.

If you’re skewing your analysis towards gallons, greenroofs only look cost effective if you’re in a sewershed that has above-average problems. But if you’re skewing it towards number of events, then maybe green roofs become more cost-effective citywide.

Then the other thing is, if it’s water quality you’re looking at, one should really be looking at the concentration of fecal matter that gets put into the water body. With the big events, it’s likely very diluted. “Dilution is the solution to pollution” sort of works for that one! But for the small events, where you’ve just got enough stormwater to trigger the event, then you’re looking at less dilution. So maybe you should look at dollars per milligram per liter of fecal matter that is eliminated from entering a waterbody

And then there’s yet another way of looking at it. If greenroofs could reduce 25 percent of the stormwater entering the sewage system, the system might be able to handle 25 percent more sanitary sewage without much additional investment in sewage infrastructure. This might enable population increase in some neighborhoods at a lower public infrastructure cost. So, the benefit of this needs an entirely different metric. We’re really interested in not just the science, but the justification of greenroofs as a potential urban strategy.

Continued in Part 2

Columbia Water Center demonstrates research-based solutions to global freshwater scarcity.  Follow Columbia Water Center on Facebook and Twitter

Notify of

Inline Feedbacks
View all comments
13 years ago

This is a fantastic approach to water conservation,am writing from Kenya which is believed to be water scarcity country by all definitions.In my rural homes we usually use grass thatched huts,does these grass thatched roofs qualify to be green roofs. very little research have been carried out on this especially the quality of water from such roofs.What could be done on this?

Pat Widmayer
Pat Widmayer
13 years ago

YES! Your roof definitely is green in the sense that the materials to make it are natural as well as the distance materials have to travel to their destination. This would hardly increase your human footprint. I am in the Horticulture and Crop Sciences Department at The Ohio State University, and have not yet uncovered anything on the quality of the water as well. I am specifically interested in the water chemistry, particularly on global level because I’m sure much of the water chemistry if effected by acid rain falling hundreds even thousands of miles before falling to the earth…