Application Story

Paul D. Wolfe and Katie Felver
University of Oregon
Department of Architecture
Eugene, OR, 97403-1206

Green Roof

Fig. 1: Indigo (12W) Building Green Roof, note roof shielding and installed weather stations

Green roofs are a technology increasing in use all over the world and their effect on urban stormwater mitigation and thermal insulation can improve the performance of a building; however, weather conditions have an effect on these performance ideals. This study analyzes two different urban green roofs in Portland, Oregon for thermal properties, hydrological properties, and the influence of shading and solar radiation on the roof surface.

Weather stations were installed and recorded the following: soil temperature, ambient temperature, RH, soil moisture, precipitation, and wind speed. After a brief testing period, solar exposure is seen to show decreased thermal performance; the roof maintains a more normal and lower temperature in the shade. High levels of exposure to weather can impact the hydrological performance of a roof. The discharge from one roof shows a lag time of up to two days, illustrating successful stormwater mitigation.


Green Roof

Fig. 2: Cyan Building Green Roof

Portland, Oregon, already renowned for the city's commitment to the environment and the sustainability movement, has begun to address growing issues of stormwater runoff and stormwater management methods. There is a need to prevent Combined Sewer System overflow events that dump both stormwater and sewage into the Willamette River and eventually the Pacific Ocean via the Columbia River. Portland's Bureau of Environmental Services (BES) lists methods for stormwater management, one of which is the green-roof, or eco-roof(1). With 20% of Portland's surface area tied up in roofscape, the green roof presents a logical and potentially effective method of reducing stormwater runoff and Combined Sewer System overflow events. In addition to the benefits of stormwater management, the eco-roof should help to reduce the urban heat-island effect by reducing roof surface temperatures and re-radiation, potentially encouraging the infiltration of local biota into former ecological dead-zones. Better understanding of the effectiveness of green roof systems in an urban environment is needed and can be achieved by analyzing environmental issues and their effect on green roof performance.

In this study, two different roofs in the City Center of Portland, Oregon were monitored and analyzed. These are the Indigo Building, also known as 12W, by ZGF Architects and Gerding Edlen Developers, and the Cyan Building by THA Architecture, GBD Architects, and Gerding Edlen Developers.

Cyan Building Indigo Building
Completed Fall 2009 Summer 2009
Sqft 378,220 370,000 (approx)
Height 16 Stories 22 Stories
Green Roof 6,000 (approx) 4,800 (approx)
LEED Rating Gold Platinum (anticipated)

Both buildings have extensive green roofs. An extensive green roof is classified by a growth media depth of less than 6" and plant types that are typically sedum, mosses, herbs, and grasses(2).

2.1 Roof Comparison

The Indigo Building extensive green roof is on the top floor, 23 stories above the ground, and varies greatly in soil depth, solar shading, and wind conditions. In contrast, the Cyan Building extensive green roof is on a three-story portion of the building. It is consistent in depth, shading, and wind conditions, and has more mature vegetation. For this reason, a direct comparison between the two roofs would yield only speculative conclusions. Instead this study looks at both roofs but in different views that use the individual properties of each roof as an opportunity to draw different conclusions about performance. This study is intended to be continued throughout a full year to measure roof conditions in different seasons, comparing the effect of the rainy winter to the dry summer.

2.2 Indigo Roof Research Questions

At the Indigo Building, the conditions of the green roof create a variety of environmental microclimates at the roof surface. The microclimates include varying degrees of solar shade, including near 100% solar radiation exposure and near 100% solar shade. Two weather stations set up in these opposing conditions would yield data that could be analyzed to determine the impact that solar shading has on green roof performance. The results of this comparison could be used to understand the consequences of placing building installations, equipment, and photovoltaic panels on a green roof surface.

How does shading from direct solar radiation affect the precipitation absorption, thermal insulation properties, and moisture storage rate of a green roof?

2.3 Cyan Building Research Questions

The homogeneity of the Cyan Building green roof provided opportunity to study the hydrological effects of a green roof. Using one weather station, the rainfall and subsequent change in moisture content of the soil could be monitored, and then used to estimate the discharge rate of the green roof. Flow measurement of the total green roof runoff is not possible due to the construction of the roof drainage system, but can be estimated using a simplified water balance equation.

What is weather impact of an unshielded, new extensive green roof in the Portland, Oregon climate zone?

Using simplified hydrologic equations, what is the discharge from the Cyan green roof?


Sedum: Genus representing nearly 400 types of leafy succulents; ranges from herbs to shrubs, often used on extensive roofs.

Evapotranspiration: Type of water loss from a hydrological system due to evaporation and transpiration from vegetation.

Discharge: Outflow from a hydrological system.

4.1 Set-up and Installation

Beginning in the winter, weather base stations were set up on the individual roofs to measure various environmental characteristics of each green roof.

On the Indigo building roof, two Onset HOBO Wireless Weather Stations (U30) were installed. One station was set up in near 100% direct solar exposure and one station was set up in near 100% shade. The shading percentage was established by using the Solmetric Suneye. An allowance was made on the siting of the 100% shaded weather station so that the photovoltaic panel required to keep the battery charged was exposed to a portion of direct sunlight at some point during each day.

Parameter/Sensor Units
Soil Moisture m3/m3
Wind Speed mph
Precipitation inches
Soil Temperature °F
Ambient Temperature °F
Relative Humidity %

Once the weather station and tripod were set up with the appropriate mounting and securing equipment, the sensors were installed on and around the weather station. Interference from surrounding rooftop equipment and architectural features were not eliminated from the sensors range as they are integral to the rooftop assembly and therefore affect the green roof (Fig.1). The soil temperature and soil moisture sensors were placed within direct proximity of one another and at relatively the same depth (1") within the growth media of the roof. This was done to ensure that the reading returned from the sensors would be applicable to one another.

On the Cyan building, the base station used was a Onset HOBO Weather Station (U22). All the same sensors were used with the U22 and installed in the same manner as those listed earlier.

4.2 Logging and Data Extraction

Once all the weather stations had been set up and made secure, they were launched in order to begin logging data. In the case of the two U30 units, a preconfigured setup using a logging period of 1 minute was loaded to the logger using the HOBOware software package and a laptop computer. This was done in the field. For the U22 unit, the setup was preloaded onto the device using the HOBOware software in the lab and set with a trigger (in this case, the "log" button on the interior of the unit). For both types of units, the logging can be confirmed via an LED indicator on the interior of the unit that blinks once every 30 seconds to indicate logging in process.

The data can be extracted on a rough schedule but can still be compared with all previous and subsequent sets of data. To extract the data, a handheld Onset U-Shuttle device is used for the U22 and a laptop is needed for the U30. In the case of the U30 stations, the laptop is connected to the HOBOware software. The software will prompt the user to continue logging or stop. In both methods, the prompts must be identified in order for the loggers to continue to log. In all cases the extracted files should be backed up in multiple locations to avoid loss of data.

  1. Data Analysis
  2. Data collected from the U30 units is already in the HOBOware format but data obtained from the U22 unit is on the Shuttle. The readout is downloaded from the U-Shuttle device using a computer and the HOBOware software package. From the HOBOware software the data can be exported in table format for analysis in Microsoft Excel. Graphs and data points can be quickly created and analyzed in Excel and used to chart the changes in individual parameters over time.

5.1 Initial Observations and Issues

The initial data collection point was for a 12-day period from February 8, 2010 (when the weather stations were installed and launched) to February 19, 2010. The weather stations were calibrated to record data at an interval of 1 minute for all 5 sensors, yielding 8 separate data points every minute including the time/date stamp. The total collected for the relatively short 12-day period was 414,720 unique data points. Only 8 of those days were used for comparisons.

Upon collection and analysis of the data from all three weather stations, an observation was made about the lack of rainfall change; specifically, all three gauges read 0.00" of precipitation over a 12-day period in February. Simple intuition indicated that this was a near impossibility and a check of the USGS HYDRA data for Portland, Oregon indicated a total of 1.40" of precipitation for that time period. To supplement the lack of accurate precipitation data at the individual weather station sites, HYDRA precipitation data was taken from a site approximately equidistant from all three weather station locations. HYDRA precipitation data is only available in 1-hour records and the graphed results reflect this difference in available data.

Soil Moisture

Fig. 3: Soil Moisture, Soil Temperature and Ambient Temperature relative to Precipitation.

5.2 Indigo Building: Shading

Data from the two weather stations on the Indigo building shows a clear differentiation between the two conditions of 100% shade and 100% solar radiation exposure. The relative temperatures show a consistent trend and are nearly identical (Fig. 3), with only minor variations.

The difference in soil temperatures for the two conditions of shading is telling when looked at in relationship to the ambient temperature (Fig. 4). While the ambient temperatures are almost identical for the period, the soil temperatures show difference in trend. The soil temperature in the solar shading condition shows a trend similar to the ambient temperature but in most cases at a lower absolute temperature. The soil temperature in the solar radiation condition shows a propensity for more rapid changes, higher absolute temperature, and closer trending to the ambient temperature numbers. This is likely due to the impact of direct solar radiation on the surface of the roof. The addition of the shortwave IR radiation from the sun causes a more rapid increase in the temperature of the soil, even at a depth of approximately 1" (temperature sensor location).

The soil moisture content for the solar radiation condition shows much higher soil moisture content than the solar shading condition. This difference is likely due to the effect that the physical barriers providing the shade have on the available precipitation (Fig. 1). However, not all shading can be responsible for the lower levels of soil moisture. When the curve showing soil moisture content for the solar shading condition is corrected to meet the solar radiation condition curve (Fig. 5), thus eliminating most of the physical discrepancy, the solar radiation condition still shows higher soil moisture content. The corrected data shows that while the moisture content differs in the two conditions, the trends of inflow and outflow from the roof are similar. Both show an increase in moisture during and after rainfall events and then a decrease after the event. The curves briefly meet as the moisture content begins to stabilize (ex Feb 15th). This trend can be seen extended beyond the stabilization point during Feb 17th and Feb 18th. Beyond the stabilization point, the solar shading condition retains more moisture for a longer period of time than does the roof in the solar radiation condition.

Soil Moisture

Fig. 4: Soil Temperatures relative to the Ambient Temperature. In the Solar Shade Condition the roof is slower to change temperature and maintains a lower absolute temperature than the Solar Radiation Condition.

Soil Moisture

Fig. 5: Soil Moisture Content relative to Precipitation and corrected to show similar trend points. In the Solar Shading Condition the moisture content is lower than the Solar Radiation Condition, even when corrected to show trend.

It is likely that the addition of the solar radiation begins to dry out the roof more quickly after a more extended period with no precipitation.

5.3 Cyan Building: Weather Influence

The Cyan green roof is located three stories above street level and is unshielded from all environmental impacts. The roof receives full solar radiation exposure (save the building shadow in afternoon), full wind gust exposure, and full precipitation influx. As a result of the unshielded nature of the roof, many weather variables have a potential hydrological impact on the performance of the roof. Most notably among these is the potential for wind to affect the evapotranspiration rate of the vegetation and precipitation to influence the potential peak discharge and drainage of the roof assembly.

Evapotranspiration can be affected by many different factors including relative humidity, soil moisture, temperature, and wind speed. The effects of wind speed on an unshielded roof manifest themselves in both the temperature of the roof and the soil moisture content. As evapotranspiration increases, the rate at which the soil loses moisture will increase, especially near the surface. Fig. 6 shows the weather data for Cyan station and indicates a change in soil moisture rate when the wind gust speed increases markedly (Feb 17th and Feb 18th). The soil temperature also behaves differently when the wind gust speed increases. Typically, the soil temperature trends with the ambient temperature showing an increased lag time of change. When the wind gust speed increases with consistency during Feb 17th and Feb 18th (Fig. 6) the soil temperature decreases below the ambient temperature but continues to follow the trend of the ambient temperature.

Soil Moisture

Fig. 6: Weather Station Measurements for the Cyan Building; note the decrease in soil temperature relative to ambient temperature and soil moisture content when the wind gust speed increases.

5.4 Cyan Building: Hydrology Calculations

Basic calculations to determine the peak flow of a green roof can be done by using the Water Balance Equation (Fig. 7).

Q = W + Gin – (ET + Gout) ± ΔS
Fig 7: Water Balance Equation (Q = Discharge, W = green roof can be done by using the Water Balance Precipitation, Gin = Groundwater Inflow, Gout = Equation Groundwater Discharge, ET = Evapotranspiration, and ΔS = Change in Storage)

In the case of a typical green roof, Gin and Gout can be disregarded because there is no groundwater flow in a closed system (in the case of green roofs where independent irrigation is applied, Gin should be substituted with net influx from irrigation). ΔS can be calculated from soil moisture content. Soil moisture content is measured in volume: water per volume: soil and therefore can be expanded to represent a volume: water content for the entire roof. The change in these values over time represents the ΔS of the roof system.

ΔS Roof (m^3) W (m) ET Roof (m/day)
2.32920 0.005080 3.3444
2.03872 0.008636 3.3444
‐0.65490 0.004826 3.3444
0.00000 0.001270 3.3444
0.36315 0.009144 3.3444
0.50968 0.005334 3.3444
‐1.01930 0.001270 3.3444
‐1.38251 0.000000 3.3444
‐1.52900 0.000000 3.3444

Table 3 illustrates the proper format for the variables in order to calculate discharge for the entire surface of the roof. In this case, the roof is 557.4 m2 (6,000 ft2).

Date Rain(mm) Q W ET + ΔS (m^3/day)
Feb 10th 5.08 ‐1.0100824
Feb 11th 8.64 ‐1.297044
Feb 12th 4.83 ‐3.9945128
Feb 13th 1.27 ‐3.34313
Feb 14th 9.14 ‐2.972109
Feb 15th 5.33 ‐2.829386
Feb 16th 1.27 ‐4.36249
Feb 17th 0.00 ‐4.726907
Feb 19th 0.00 ‐4.87344

Table 4 shows a clear correlation between discharge lag time and the precipitation. Lag time for the Cyan roof is between one and two days. In this case the roof is performing the primary function of stormwater mitigation effectively.

Because of the complication with measuring ET rates in systems where no physical alteration to the environment is possible the ET rate for the equation can be estimated and used as a constant for time t. In this study the rate of .6mm/m2/day (.0022"/ft2/day) was used as the ET constant(6).


6.1 Indigo Building Conclusions

The Indigo Building, completed in the fall of 2009, has a new green roof. The vegetation has yet to come to full maturity and as such only covers a portion of the actual green roof surface. This fact makes some hydrological conclusions difficult to make when the full hydrological system is not yet fully in place. The impacts of solar radiation on the roof can still be determined, however, and as the measured data applied only to the soil itself, this is doubly true.

That the solar radiation has an effect on the roof performance is clear from the data presented, but the impacts are of varying consequence, depending on the primary role the green roof plays in the building system. Solar radiation impacts the thermal properties of the roof as the temperature fluctuates more over time than does the full shade.

Data on the precipitation absorption of the green roof in shaded and unshaded conditions is inconclusive because of the lack of actual precipitation data for the specific locations during the time period we collected the data. The substitution of HYDRA data showed the overall trends of the green roof in the two conditions relative to the different environmental factors but failed to show ultra-specific data for precipitation affecting the different locations of shade and radiation. Ultimately, this would have proved semi-useful as the precipitation absorption of a shaded green roof depends fully on the type of shading. Equipment and buildings will have little to no effect on vertical precipitation while horizontally mounted PV arrays would have a near 100% reduction on vertical precipitation.

It is clear from the data that the shaded location did experience less precipitation. The soil moisture content indicates as much. The soil moisture data also indicates a higher lag time for the discharge of water in the green roof; the roof exposed to solar radiation sheds water faster. This is likely the result of increased evapotranspiration as well as higher soil temperatures. For the time period tested, if the primary purpose of the roof is to mitigate stormwater, shading may not benefit that mission and may in fact hinder the performance in that respect.

6.2 Cyan Building Conclusions

The weather data gathered for the Cyan Building green roof seems to show several trends relative to the wind speed and the moisture content of the soil. These trends could be the result of increased evapotranspiration due the increased wind speed or they could simply be the natural regulation of the soil moisture due to lack of rain and increased discharge. Further data, over a longer period of time is needed to fully understand what the implications of wind are on the soil moisture.

The wind seems to have a more direct impact on the soil temperature and the data supports the conclusion that the increase in wind speed is decreasing the soil temperature of the green roof. This has impacts for unshielded roofs and their potential for thermal protection. In the hotter summer months, when thermal insulation is needed, this may be beneficial. Yet, in the winter months, when heat loss needs to be negated, the cooling of the roof could have negative impacts.

The hydrological data for the Cyan roof shows a green roof which is performing the primary role of stormwater mitigation effectively. The discharge data (Table 4) shows a delay of approx 1.5 to 2 days between large rain events and the increase in discharge associated with rain events. On a normal roof this discharge would have been nearly immediate. Add to the equation the fact that the Cyan building captures and filters nearly all the discharge from the green roof and the net result is promising for the potential of green roofs for stormwater mitigation.

Again, the addition of more accurate precipitation data would help to ensure more accuracy in the hydrological calculations.


Special thanks to Gerding Edlen Developers' Mark Johnson, Bobby Lemmon (Cyan Building Manager), and Rob Hitchcock (Indigo Building Manager) for providing access and encouraging the development of the study on their buildings. Thanks to ZGF Architects, GBD Architects, and THA Architecture for aiding in the understanding of the buildings. Additional thanks go to Christine Theodoropoulos, Head, UO Department of Architecture, and Cindy Lundeen, Associate Director of Development for providing travel support from the Louis C. Rosenberg Scholarship; the UO Case Study Lab for the weather stations and equipment loan, and Suzanne Walther for hydrological guidance. Last but not least Dr. Alison Kwok who provided guidance, expertise, understanding, and patience and without whom, this study would not have been possible.


  • (1) Bureau of Environmental Services, Portland, Oregon. "2005 Stormwater Management Facility Monitoring Report," September, 2006. <>
  • (2) Kwok, Alison, and W. Grondzik. "Roddy/Bale Green Roof Study," Proceedings of the 33rd National Passive Solar Conference-Solar 2008, San Diego, CA, May 4-8, 2008
  • (3) Liu, Karen, and B. Baskaran. "Thermal Performance of Green Roofs Through Field Evaluation," Proceedings for the First North American Green Roof Infrastructure Conference, Awards and Trade Show, Chicago, IL, May 2930, 2003
  • (4) VanWoert, Nicholaus D, D Bradley Rowe, "Green Roof Stormwater Retention: Effects of Roof Surface, Slope, and Media Depth." Journal of Environmental Quality 34, 1036-1044 (2005)
  • (5) Bureau of Environmental Services, Portland, Oregon. "2009 EcoRoof Handbook," 2009 <>
  • (6) Kasmin H., Stovin V.R., "Towards a generic rainfall-runoff model for green roofs."
    Department of Civil and Structural Engineering,
    The University of Sheffield, Sheffield, UK.
  • (7) Dingman, S. Lawrence. "Physical Hydrology," Waveland Press Inc,. Long Grove, IL: 2002

©ASES 2010 reprinted with permission from the American Solar Energy Society