Standard keys to success in bioretention construction apply to trees for bioretention, including the following.
See the bioretention section for other construction guidelines and specifications.
In addition to general bioretention guidelines and specifications, the following guidelines and specifications apply specifically to trees for stormwater design.
Each of the above techniques is described and compared below. Links to construction guidelines are also provided.
In areas that do not have enough open space to grow large trees, techniques like suspended pavement can be used to extend tree rooting volume under HS-20 load bearing surfaces and create favorable conditions to grow large trees in urban areas. This rooting volume can also be used for bioretention. While suspended pavement has been built in several different ways, all suspended pavement is held slightly above the soil by a structure that “suspends” the pavement above the soil so that the soil is protected from the weight of the pavement and the compaction generated from its traffic.
One of the earliest examples of trees grown in suspended pavement is in Charlotte, North Carolina, where a reinforced concrete sidewalk was installed over the top of poured in place concrete columns. While this is an effective way to grow large trees, it is labor intensive and requires intensive surveying to ensure that column heights are precise.
A more recently developed, and less labor intensive, technique to build suspended pavement is through the use of soil cells. An example is Silva Cells, modular proprietary pre-engineered structural cells manufactured by Deeproot Green Infrastructure. Other examples are RootSpace systems manufactured by GreenBlue Urban and Stratavault systems manufactured by Citygreen. The modular design allows flexibility to size the rooting/bioretention volume as needed for each site. Underground utilities can be accommodated within these systems. Because soil in a suspended pavement system is protected from compaction from loads on pavement above the cells, a wide range of soils can be used in these systems, so soil can be tailored to desired functions (e.g. tree growth and stormwater management). Construction documents and specifications for a wide range of soil cell applications can be found on the manufacturers' websites (Deeproot Green Infrastructure, GreenBlue Urban, and Citygreen).
Photo galleries below illustrate the Silva Cell and Stratacell/Stratavault technologies.
Rock based structural soils are engineered to be able to be compacted to 95 percent Proctor density  without impeding root growth. Rock based structural soils are typically gap graded engineered soils with
Desired characteristics for the stone base used in rock based structural soils include the following.
According to University of California-Davis (1994), 2 inch stones would be able to support most tree roots. Because limestone has been found to crush on some projects, granite stone is recommended. If limestone is used, it should meet specifications described in the permeable pavement section of this manual.
Soil needs to have adequate nutrient and water holding capacity to provide for the tree’s needs.
The tackifier, if used, should be non-toxic and non-phototoxic.
Construction drawings are available on Cornell University’s website, including
Several types of rock based structural soils have been developed, including
Note: University of California (UC) Davis structural soil is not designed to be load-bearing and therefore should not be compacted. This structural soil is 75 percent lava rock (0.75 inches) and 25 percent loam soil (by volume) (Xiao and McPherson, 2008). Because of the lava rock, this soil stores more stormwater than other structural soils and has a very high surface area to facilitate pollutant trapping.
Wenz provides a discussion of some structural soils, including case studies.
Day and Dickinson (2008) provide information on use of trees in structural soils, including design specifications. The following considerations should be made in using structural soils.
Based on the above studies, Urban (2008) recommends: “Given the extreme inefficiency of the ratio of excavated volume to soil usable by the tree, strips of structural soil less than 20 feet wide might be better constructed as soil trenches or structural cells, where more soil can be included for less cost. A 5-foot wide soil trench set of structural cells…will provide more soil usable by the tree than a 20 foot wide trench of soil/aggregate structural soil. Soil/aggregate structural soils may have applications as a transition to other options, and to add soil in places where other options may not be practical. These might include tight, contorted spaces and fills around utility lines and against foundations where full compaction is required.”
More information about rock based structural soils is available online at
Sand based structural soils were first developed in Amsterdam when some trees were in poor condition because of an “unfavorable rooting environment” (Couenberg, 1993). Because the natural soils in Amsterdam, bog-peat, was non-load bearing, the top 2 meters of soil had been replaced with a medium coarse sand, which had insufficient nutritional value. Amsterdam soils were developed in an effort to grow better trees but still provide adequate bearing capacity for pavement bearing light loads, such as sidewalks. The Dutch studied various mixes for tree growth, soil settlement, and several other parameters. The resulting Amsterdam Tree Soil contains medium coarse sand with 4 to 5 percent organic matter and 2 to 4 percent clay by weight and also meets other criteria, including, for example, (1) the medium coarse sand must meet specific gradation requirements, (2) soil mix must be free of salt, (3) mix must contain less than 2 percent particles below 2 micrometers, and (4) amount of particles below 2 micrometers must be considerably less than the amount of organic materials (Couenberg 1993).
All medium coarse sand (the layer above the Amsterdam Tree Soil) is compacted to 95 percent to 100 percent Proctor density. Amsterdam tree soil is not compacted to 100 percent density, but “is compacted until the soil has a penetration resistance between 1.5 and 2 MegaPascal (187 to 250 pounds per square inch (PSI)) ... Comparison of soil density values after filling with soil density at 100 percent Proctor Density has shown that soil density of Amsterdam Tree Soil after filling amounts to 70 to 80 percent Proctor Density” (Couenberg 1993).
Amsterdam Tree Soil was found to settle 19 millimeters [0.75 inches] in 3 years compared to the surrounding pavement, which was acceptable according to Dutch standards (Couenberg 1993), but may not be acceptable to many communities in the US to minimize risk of litigation related to trip and fall hazards.
The standard design in Amsterdam includes the following from bottom to top.
While Amsterdam sand based soils work well in Amsterdam, trees grown in Amsterdam soils in Minnesota would likely need significant irrigation, as they have low water holding capacity. Amsterdam receives an average of 36 inches of rain per year, which is higher than the range of average yearly rainfall in Minnesota. Rainfall events are also more frequent and lower intensity in Amsterdam than in Minnesota. Summer temperatures are also higher in Minnesota than in Amsterdam, resulting in higher water needs by trees. According to Couenberg (1993), who helped develop Amsterdam soils, “the Amsterdam tree soil has been developed in an area where there is sufficient rainfall. If this soil is used in other climatic areas, adaptations to the tree pit will have to be made after the local situation has been monitored.”
Most of Amsterdam also has a high water table which is controlled to have almost no seasonal fluctuation, and groundwater wicks up into tree rooting zones by capillary action, so trees always have access to water from the ground water table. According to Urban (2008), “tests of Amsterdam planting soil in other locations in Europe without high water tables showed less promising results with overly dry soil, which required significant supplemental water.” So, in summary, sand based structural soils may be a viable in Minnesota if some settlement is acceptable with light structural loads and trees are irrigated.
For more information on Amsterdam Tree Sand see 
Soil boxes are concrete boxes designed for bioretention. They are typically proprietary products, such as, for example, the boxes made by Filterra and Contech. Rooting volume capacity of these boxes is typically limited to large shrubs. Soil volumes provided by these boxes are typically not sufficient to grow healthy large trees. A standard 6 feet by 6 feet Filterra box, for example, provides 72 cubic feet of soil, assuming a 2 foot depth of soil. Given that the recommended soil volume for trees is 2 cubic feet of soil per 1 square foot of tree canopy, this is only enough to support a tree with a 5 foot radius canopy.
Comparison of soil volumes, open space, and underground space needed for open grown tree vs. tree in suspended pavement, rock based structural soil, sand based structural soil, and soil boxes.
Link to this table
|Technique||Soil volume provided per cubic foot rooting zone (not including pavement profile where applicable)||Open space needed to grow a 30’ diameter canopy tree assuming 34” soil depth1||Total volume recommended per 30’ diameter canopy tree1||Structural Capacity (Traffic load supported)|
|Open tree pit (no pavement or foot traffic above rooting space)||1 ft3||500 2||1413 ft3||Not even foot traffic should be allowed on open tree pits|
|Structural cells||0.92 ft3||5 ft by 5 ft||1536 ft3||Can support vehicle loading up to AASHTO H-20 rating of 32,000 lbs. per axle (U.S. Federal Highway Bridge Standard). This rating refers to the ability of a roadway to safely accommodate 3-4 axle vehicles, such as a large semi-truck and trailer (Deeproot website)|
|Rock based structural soils3||0.5 to 0.75 ft2||5 ft by 5 ft||2826 to 2120 ft3||Can be used under pedestrian mall paving, sidewalks, parking lots, and low-use access roads|
|Sand based structural soils||1 ft3||5 ft by 5 ft||1413 ft3||No standard test data available; Amsterdam sand settled 19 mm in 3 years compared to the surrounding pavement (Couenberg 1993), which is generally not acceptable in the U.S.|
|Soil boxes||1 ft3||Not large enough to grow 30 ft diameter tree||Not large enough to grow 30 ft diameter tree|
1 Based on 2 ft3 of soil volume per 1 ft3 of canopy area, assumes Silva Cells are used for structural cells, assumes 92% void space in Silva Cells; assumes CU Structural Soil is used for rock based structural soil; assumes soil component of rock based structural soil is 20%.
2 Although a typical rock-based soil includes 80% rock and 20% soil, the effective volume of the soil is greater than 20%. The recommended ratios in the table reflect information provided in Grabowsky et al., 2009, and by Dr. Nina Bassuk, Cornell University (personal communication).
Recommended number of soil borings, pits or permeameter tests for bioretention design. Designers select one of these methods.
Link to this table
|Surface area of stormwater control measure (BMP)(ft2)||Borings||Pits||Permeameter tests|
|1000 to 5000||2||2||10|
|5000 to 10000||3||3||15|
1an additional soil boring or pit should be completed for each additional 2,500 ft2 above 12,500 ft2
2an additional five permeameter tests should be completed for each additional 5,000 ft2 above 15,000 ft2
The permanent stormwater management system must meet all requirements in sections 15, 16, and 17 of the CSW permit and must operate as designed. Temporary or permanent sedimentation basins that are to be used as permanent water quality management basins have been cleaned of any accumulated sediment. All sediment has been removed from conveyance systems and ditches are stabilized with permanent cover.
The following pages address incorporation of trees into stormwater management under paved surfaces