Welcome to Purple-Roof's Green Roof Retention Modeler (v0.2.1, beta).
Select a location, then click "Let's model”. Then the Modeler will compare the performance of sixty-four (64) common green roof profiles, using stochastic weather for the selected location. For more information, check click "How To Use the Retention Modeler" above.
Purple-Roof's Green Roof Retention Modeler uses stochastic weather, or weather data that is statistically likely for any given area, but which does not match any specific year. We start historical weather data at an hourly timestep from Open Weather and process with our own algorithms that generally:
As a final step, we review each data set before deploying to the modeler to ensure that the data reasonably represents weather for that location. Just below each model is a table summarizing statistics for the weather set shown, along with sources of information.
We do this because stochastic weather most closely resembles weather for any given year, whereas raw historical data is often incomplete and might not represent norms.
Because past weather might not accurately predict future weather, we will soon incorporate options to allow you to adjust weather based on climate predictions.
Contact us to request addition of a location. Meanwhile, modeling for a nearby location may be sufficient.
We are currently in beta release of a stormwater detention modeler that models the Purple-Roof and other detention roof concepts. The retention modeler on this page came first. In so many ways, retention is far easier to model, and prior to 2018, almost no-one was researching detention within green roofs.
Retention modeling lends itself to a long time scale, such as the one-year scale shown on this page. This is because retention is predominantly seasonal (check out the model, and you'll see that retention is low when weather is cold and wet), and because retention requires evapotranspiration over the course of several days or weeks.
Detention roof modeling lends itself to a much shorter timespan, such as multiple hours of a few days. This is because detention causes runoff delay of multiple hours to about 2 days, which is not apparent on a larger graph, such as one-year.
For rooftop detention modeling, contact us or one of our partners directly, as our detention modeler is not availble to the general public yet.
The diagram below illustrates the primary functions within the modeler. The modeler constructs a series of continuous flow vegetative and hydrologic models, using a daily timestep. Generally there are six functions, as outlined below.
Purple-Roof's Green Roof Retention Modeler is a joint effort between Green Roof Diagnostics (GRD) and the consortion of companies who bring you the Purple-Roof concept. GRD selects equations, determines parameters, and is generally responsible for ensuring accuracy, applicability, and objectivity of the Green Roof Retention Modeler. The Purple-Roof consortium makes the Modeler accessible to the public via graphic user interface..
We are developing this project together to build upon our collective strengths and shared interests. GRD seeks to improve performance standards and for practicioners, such as civil engineers, to utilize improved green roof performance standards. Purple-Roof wants to help architects and engineers use higher performing green roof assemblies on more of their projects.
We are actively developing this project and will continually release new versions. Each version will include full documentation of calculations, per below.
The following sections cover each of the steps above in greater detail. Please note that these are equations for version 0.2.1. This page will be updated with current equations for each version of the Modeler.
Version 0.2.1 utilizes a series of static, empirical equations in a continuous flow pattern.
As antecedent conditions are not known before the first increment, the assumed antecedent condition is 0% plant stress and profile water retention of 20% of its ASTM maximum water storage capacity. These are reasonable assumptions that neither benefit nor penalize any particular assembly.
Regardless of the accuracy of this initial assumption, the effects of this initial assumption are minimal, as they have no bearing on the model after the first rain event.
The Modeler assumes a uniform roof slope of 2 percent. Version 0.2.1 currently does not accept user input for these variables, though input will be allowed in later versions.
Note: slope is currently not considered in any equations other than water inputs. Future versions will utilize slope as a variable affecting the water balance equation.
Version 0.2.1 utilizes stochastic weather data from a specific location conforming to 20-year historical averages. Weather statistics are obtained various reputable sources, and actual historical weather is obtained from Open Weather, though different sources may be used in the future. Open Weather aggregates weather data from a variety of pubic and private sources. We use a custom program to create stochastic weather that conforms to statistical averages as well as hourly patterning from historical data.
Note: later versions may consider other water input sources, such as cisterns or contributing drainage areas.
Rainfall that is observed falling from the sky onto the ground is not necessarily the same volume of rainfall that lands onto a rooftop, due to wind, roof slope, and orientation of roof slope. Observed rainfall is adjusted per the following equation.
Win = Robs * Ω * ϑ ((X × π) / 180 ) * ( 1 + arain * X )
Win = rainfall (mm day-1) adjusted to roof slope and orientation
Robs = the observed rainfall (mm day-1);
Ω (Omega) = surface area (m2) of the roof
arain = dimensionless program parameter; The value of arain depends on the orientation of the roof and local wind and rainfall characteristics since roofs oriented to a dominant wind direction receive more rainfall than flat roofs or those oriented towards another direction
X = slope of the roof (degrees).
Note that version 0.2.1 currently sets the value of arain to a constant value of 1, as local wind and rainfall characteristics are not currently available.
"Runoff and vegetation stress of green roofs under different climate change scenarios".Landscape And Urban PlanningVolume 122, Februrary 2014.Equation #1
Find it at: Elsevier
Differences between green roof profiles account for some of the more significant differences in performance between green roof applications. Green Roof Diagnostics and others are currently researching these conditions. Green Roof Diagnostics is parameterizing values to be used in modeling.
Version 0.2.1 currently utilizes the following parameters:
|0.35||ASTM E-2399 maximum retention value of green roof soil / green roof media, i.e. 35% per volume (35% volumetric water capacity)|
|0.93||ASTM E-2399 maximum retention value of mineral wool in green roofs, i.e. 93% per volume (93% volumetric water capacity)|
|0.75||Efficiency factor without detention layer: actual retention is assumed to peak at 75% of ASTM values|
|1.05||Efficiency factor detention layer: actual retention is assumed to peak at 105% of ASTM values|
|Mixed Sedum||Plant palette assumed to be mixed sedum|
|0.56||Crop coefficient of evapotranspiration|
|12L||Stress threshold of available water. I.e. the selected plants are assumed to begin to exhibit stress when available water drops below 12 liters per square meter of soil surface area. For a 100mm (4-inch) thick green roof profile, this equates to 12% volumetric water capacity, or approximately 1/3 of total maximum water-holding capacity of a soil-only green roof.|
The crop coefficient of 0.56 for Sedums is an estimated average for mixed Sedum, utilizing multiple publications that document a range of crop coefficients for a single species of Sedum. The ASTM values listed are based on values commonly observed within the industry, and documented by multiple ASTM tests. The remaining parameters are derived from soon-to-be-published research performed by Green Roof Diagnostics, based on research from 2016-2019.
Currently the Green Roof Retention Modeler is using a very simple water balance equation. Conceptually a water balance equation assumes that the green roof is a "reservoir" that receives water (with 100% efficiency) until it reaches capacity, and after reaching capacity, runoff begins and occurs instantaneously.
We chose this simple equation for version 0.2.1, but future versions will more accurately predict efficiency and detention.
Wro = Win − (WSmax − WSact)
Wro = amount of water lost by runoff (l day-1) if Win exceeds the actual storage capacity of the roof
Win = rainfall (mm day-1) adjusted to roof slope and orientation (output of formula 1)
WSmax = maximum amount of water retained on the roof (l day-1). WSmax depends on the roof type and surface area.
WSact = actual amount of water retained on the roof (l day-1). WSact depends on incoming rainfall and outgoing green roof evapotranspiration and is updated on a daily basis. WSact varies between zero when the roof is completely dry and WSmax when the roof is saturated with rainfall.
"Runoff and vegetation stress of green roofs under different climate change scenarios".Landscape And Urban PlanningVolume 122, Februrary 2014.Equation #2
Find it at: Elsevier
Plant stress is an important factor to consider when designing and selecting a green roof. We think it is important to include plant stress in any modeling efforts. Zero plant stress might not be an attainable goal, but plant stress should be minimized and/or reduced to a few manageable occurrences per year.
The primary contributor to plant stress is lack of available water, and the equation below only considers lack of available water as a factor causing stress. Later versions may consider other, minor, factors.
S = 100 * (1 - (WSact / (p * WSmax) ) )
S = stress level
p is the threshold for vegetation water stress
WSmax = maximum amount of water retained on the roof. Result of water balance equation.
WSact = actual amount of water retained on the roof. Result of water balance equation.
"Runoff and vegetation stress of green roofs under different climate change scenarios".Landscape And Urban PlanningVolume 122, Februrary 2014.Equation #4
Find it at: Elsevier
Evapotranspiration is the combined loss of water via evaporation and transpiration, i.e. all water that is lost from the green roof in the form of water vapor.
The definition of retained water is water that is absorbed by the green roof and which only leaves the roof through evaporative processes. Therefore evapotranspiration is the single most important process in the continued function of a retention-oriented green roof.
The Modeler uses the global standard FAO Penman-Monteith evapotranspiration equation. Specifically, FAO Penman-Monteith estimates evapotranspiration for a reference crop of grass. This value is then multiplied by a stress factor and a crop coefficient factor (see above for both) to determine evapotranspiration on a given date.
Δ + γ (1 + 0.34U2)
ET = ETo * Ks * Kc
ETo = reference evapotranspiration [mm day-1],
Rn = net radiation at the crop surface [MJ m-2 day-1],
G = soil heat flux density [MJ m-2 day-1],
T = mean daily air temperature at 2 m height [°C],
u2 = wind speed at 2 m height [m s-1],
es = saturation vapour pressure [kPa],
ea = actual vapour pressure [kPa],
es - ea = saturation vapour pressure deficit [kPa],
Δ = slope vapour pressure curve [kPa °C-1],
γ = psychrometric constant [kPa °C-1].
Ks = water stress coefficient
Kc = crop coefficient
"Chapter 2 - FAO Penman-Monteith equation".Crop evapotranspiration - Guidelines for computing crop water requirements - FAO Irrigation and drainage paper 56Rome, 1998.
Find it at: FAO
The state of the green roof after evapotranspiration occurs is saved as an "increment" or a final snapshot of that day. The next day is then modeled, using the prior day's conditions as antecedent. E.g. when the model completes May 15, that data is saved, and the final water retained within the green roof and the final plant stress on May 15 becomes the antecedent conditions used by the Modeler to begin May 16.
This process is repeated for each day of the weather series. Currently version 0.2.1 is generating 365-day models, per calendar year.