Published by City Farmer, Canada's Office of Urban Agriculture


Urban Agriculture:
The Potential of Rooftop Gardening

Chapter 2: The Site and the Plan

By Joseph St. Lawrence
Joseph St.Lawrence

Report of a Major Project submitted to the Faculty of Environmental Studies in partial fulfilment of the requirements for the degree of Master in Environmental Studies.

York University, North York, Ontario, Canada.
July 29, 1996 [Images are not included here.]

Chapter 2 (The Site and the Plan)
Chapter 3 (Observations)
Chapter 4 (Conclusion), and References

The Site and The Plan

To be sustainable, an agricultural endeavour must conform to the environment of which it is a part. This environment is typically seen as being ecological in scope, but it also has social and economic dimensions. The site of an agricultural intervention must dictate the nature of that intervention, and my rooftop garden is no exception in this regard. This chapter describes the site where the rooftop garden was constructed, and outlines the cultivation plan I developed based on the opportunities it presented.

The Site

Field to Table is a non-profit organisation. The Good Food Box programme is one of its key initiatives. There are over 1,500 people in the programme who purchase a box of food at the beginning of the month, and receive fresh produce three weeks later when their funds are typically running low. There is a number of different boxes to cater to the needs of city residents, including the Good Food Box, the Organic Box, the Mom-to-Be Box, and a Caribbean Box.

Field to Table also operates the Focus on Food programme. Job skills, nutrition, and gardening are all part of the programme which is designed to provide women on social assistance with the skills they need to get back into the workforce.

Both of these programmes are related to food security, and there is considerable potential to integrate the rooftop garden with these initiatives. The garden is compatible with the food-focus of the organisation, and the manager and staff are very supportive of my efforts there.

Field to Table operates out of a warehouse in downtown Toronto, at 200 Eastern Ave. The building has a number of useful peculiarities that I was able to exploit for the garden project. It also presented a few problems. Both sets of issues are discussed below. (I should note here that market gardening is a permitted land use in the industrial zones I1 to I3 and IC in Toronto (City of Toronto, 1995). The warehouse is in zone I3.)

Figure 2.1: The warehouse roof -- shown from above (top) and from the west (bottom).


Snow load

There are two types of roof loading: dead load and live load. Dead load refers to the weight of the roof structure itself and any permanent fixtures situated on the roof. To accommodate weights beyond the dead load (from snow, rain water, or people, for example) buildings are designed with a live loading capacity. The minimum live load permitted by the Toronto Building Code is 30 pounds per square foot (p.s.f.) for flat roofs (RGRG 1993). This is meant to bear the weight of snow in the winter, but can also be used by rooftop gardeners during the summer.

Snow will drift against walls, and building codes therefore require designers to accommodate extra loading on roofs that abut a wall. The Field to Table building has two such areas. Both the loading dock and the office sections of the warehouse are lower than the main warehouse, and hence have high live load capacities against the main warehouse wall (see Figure 2.1). For the office rooftop (where the garden was built) the live load ranges from 120 pounds per square foot (p.s.f.) at the wall to 54 p.s.f. at mid-span, and down to 32 p.s.f. at the southern extremity. The loading dock has a similar pattern of loading capacity. The live loads were used to calculate the maximum depth of soil the roof can support, and were used to determine the appropriate size and distribution of the garden's planters (discussed under The Plan, below).

Water supply

I was fortunate to have access to the municipal water supply if it proved necessary. I connected a hose to a faucet in the warehouse, and ran it up to the work area (Photo 2.2), where it could be taken out the window to the garden. Two 50' lengths of hose proved adequate for this purpose. A back-flow device was affixed to the faucet to avoid pressure-related plumbing problems when the elevated hose was turned off.

Light and wind

The garden was built at the base of a south-facing wall, and there are no overhanging trees or tall buildings in the area, so it enjoys maximum exposure to sunlight. I decided not to use the loading dock roof because it faces east and therefore has the double disadvantage of being shaded all morning, and being exposed to the hot afternoon sun.

Wind is generally more of an issue on a rooftop than at ground level due to increased exposure. There is a stand of trees to the west of the warehouse, an access ramp to the Don Valley Parkway to the north, and the garden itself sits at the base of a south-facing wall. These features combine to limit the exposure of the garden to the drying and sometimes mechanically damaging effects of wind.



When I first undertook this project, it was my hope to design an irrigation system based on captured run-off from the roof of the main warehouse. This proved impractical. The drain for the main warehouse roof is connected directly to the building's plumbing, and the pipes are all internal. Short of blasting a hole in the wall and redirecting the outlet pipes, there was no means of diverting this water from the storm drain.

Collecting the water would have posed another problem as there is no suitable place to store a large vessel. To provide adequate pressure (approximately 10 pounds per square inch) in the drip irrigation system I envisioned, a gravity-fed hose would need a supply tank 23 feet higher than the hose outlets. The maximum drop I could provide at the site is 10 feet (the distance from the garden rooftop to the main warehouse rooftop). This would only provide 5 p.s.i., and would therefore be inadequate for drip irrigation. The second problem would be getting water into this container if it were situated on same roof from which the water would be collected. Finally, the upper roof is the weakest of the three roof surfaces. A 55 gallon drum filled with water weighs approximately 460 lb. The weight of this tank would have to be distributed evenly over a fourteen square-foot area to reach the 32 p.s.f. live-load limit of the rooftop, and a number of these tanks would be needed to adequately supply the garden. These factors combined to make the irrigation system and water collection impractical.


The office roof is reached through a window in the main warehouse. There is a section of floor about 15' above the main warehouse floor space. This area is above the doorway that leads to the building's basement (Photo 2.2). It was there that I carried out my work making soil blocks and seeding. There is a bank of windows adjacent to this area, and it is there that I set up the capillary mats and tables for the seedlings (the windows face south). One of these windows opens, and provides access to the garden area.

The main problem with this arrangement is that it is awkward getting materials (e.g. blocking mix components and garden bed lumber) from the warehouse floor to the work area. I was fortunate to have access to a hand-operated lift for this purpose (also shown in Photo 2.2). It would have been quite difficult to manage the project without this device.

Indoor lighting

To provide seedlings with sufficient light, two tables were set up in front of the south-facing windows of the work area. These windows were a great advantage because seedlings could be kept in the work area, and hardened-off easily because of their proximity to the garden area. It was unfortunate however, that only a small portion of this window space is accessible, and the main watering table became very crowded in the early season. A grow room was constructed in the warehouse basement for the Focus on Food programme. Part of this space will be needed for the production of seedlings for the rooftop in future. (Spring is the only time this space will be needed, as seedlings for the main crop, tomatoes, eggplants, flowers, and herbs are all competing for space at that time.)

The Plan

Most of my time in January and February of 1996 was consumed by planning. It was necessary to choose crops, outline planting schedules, and design the garden's containers. Basically, I needed to know what to grow, what to grow it in, and how to grow it. The challenge was compounded by the discovery that the roof is much weaker than I had originally been told, and this necessitated a radical change in the site plan shortly after it had been completed. I would thus recommend that all engineering consultations be completed (if possible) before any effort is expended on planning.


Field to Table moved to the Eastern Avenue site in late June, 1995. One of the staff members was interested in rooftop gardening, and attempted to construct a garden on the site now occupied by my efforts. The late start prevented much success, but the legacy he left behind (i.e. the soil), was fundamental to the success of this project.

The soil used for the garden beds is municipal compost donated by the city. I expected it to have fairly high nutrient concentrations, but an early soil test revealed that it was deficient in both nitrogen and potassium. These deficiencies were remedied with the addition of kelp meal (for potassium and micro-nutrients), and blood meal (for nitrogen).

It would have been difficult for me to complete this project in one growing season if the compost were not already on the roof. The city's compost operation does not begin spring operations until May, and it would have taken a few weeks (if not longer, considering the heavy rain this spring) to move the compost and fill the growing beds. This would have delayed the initial plantings considerably.

Garden structures

One needs some form of container to hold the soil for a rooftop garden. Almost anything can serve -- plastic buckets are a common sight in many neighbourhoods, as are bushel baskets, but I've also seen old bathtubs, buckets, and garbage bags performing the duty.

The reasons for containing the soil go beyond the obvious need to prevent it from washing away. It is vital that the roof's drainage system is not compromised, and efforts must be undertaken to ensure that the drains do not become clogged with garden soil, compost, mulch, &c.

Beds and containers

My search for appropriate containers provided me with a great opportunity to meet one of my project objectives: viz. the use of neglected urban resources. All of the garden containers were constructed from, if you will excuse the indelicate term, garbage.

The main cropping area contains 15 beds, each 4' by 6' and 8" deep (Figure 2.2). The lumber for these beds was salvaged from discarded packing skids (Figure 2.3) at the Field to Table warehouse.

Figure 2.2: The garden. Figure 2.3: A typical packing skid.

I encountered my first hurdle when trying to separate the packing skid lumber by removing the nails in the skid. This is almost impossible because the nails often have barbs to prevent them from being pulled. I decided instead to cut the sides off the skid, and then use the leverage provided by the loosened planks to pry the planks free from the central support. This was slow work, even with the use of a power saw for the cutting, because the prying operation is manual, and fairly difficult.

Figure 2.4: An assembled side for a garden bed.

This procedure produced planks of varying widths (usually 3" and 5"), and approximately three feet in length. To form the long sides of the garden beds, four planks were screwed together, producing a side 6' long and approximately 8-10" wide. To form the ends, the remaining 3' planks were cut down to 2' lengths. Four of these were attached, yielding a side 4' long and 8-10" wide. In each case, the four planks were screwed to a 1' length of wood placed in the centre of each side (Figure 2.4).

Once the sides were assembled, they were moved to the work area above the warehouse floor. The beds were assembled outdoors in the garden area by screwing the sides to blocks placed in the beds' corners (Photo 2.3). These blocks were cut from the central supports of the packing skids from which the planks were separated.

This only provided sides for the garden beds. To protect the roof membrane from root penetration, some form of bottom is required. This was provided by placing a layer of waxed cardboard beneath the frame, and a layer of polyethylene plastic sheeting on top of the waxed cardboard (both materials were salvaged from warehouse waste). According to ECHO (1993), a moisture barrier is not necessary, but Yang (1990) recommends that roots be prevented from reaching a roof's bitumen, as they can penetrate and damage it.

To prevent soil from draining out of the frame, straw was placed around its inside perimeter, in the corners, and over any gaps. A layer of Styrofoam chips was then placed at the bottom of each bed to provide drainage. These chips were made from broken grape boxes. Finally, as recommended by Yang (1990), a layer of newspaper four sheets thick was placed on the chips and straw to keep soil from filling the drainage layer.

The beds were filled to a depth of three inches with compost donated by the Municipality of Toronto. I chose this depth because of weight limitations. Four inches of saturated compost weighed in at 31 p.s.f, slightly higher than the flat roof minimum live load in Toronto. Although the Field to Table roof has much higher live loads in some areas, I decided to limit my investigation to soils of this depth so that the research would be applicable to any flat roof in the city. Most gardeners will balk at this depth (I did), but according to ECHO's (1993) experiments in Haiti and St. Petersburg, this is sufficient for many plants, including corn (!), beans, and tomatoes. This depth also left a few inches of lumber above the soil to contain a layer of mulch. (The bed building procedure is shown in Photos 2.4-2.8.)

A few other containers were collected for some experimental plantings. Bushel baskets were used to plant squash and cucumbers, and discarded plastic buckets housed a few tomatoes, tomatillos, and eggplants. Drainage was provided by drilling holes in the bottoms of the buckets (the baskets are leaky enough) and adding a layer of Styrofoam chips, followed by a thin covering of straw. The buckets and baskets were placed on used pieces of plywood to distribute the wet-container weight over enough area to reach the 30 p.s.f. limit.

Shade and mulch

I expected to have some trouble with heat during the project, and searched for means of keeping heat-sensitive crops cool (since I planned to grow primarily greens, just about everything is heat-sensitive). I thought some form of shading would be necessary, as other rooftop gardeners I had spoken with mentioned problems of over-exposure to sunlight. I considered purchasing sheets of lattice work, but luckily I came across a free alternative: corn crates. Field to Table orders large quantities of sweet corn, and it is shipped in crates made of thin wooden slats bound with wire. The sides of these crates can be separated and arranged to yield an easy to use plant shade. Nine of these open crates will completely cover a 4'x6' garden bed, providing enough shade to keep plants cool and moist during the hotter parts of the day (Photo 2.9).

Mulch was provided from three bales of straw that had been left behind from the previous garden effort at the site. It was to be used in all beds, both to conserve water, and keep the soil cool.

Cropping plan

One of my goals in this project was to maximise the yield from the garden in an effort to prove it an economical endeavour. It was therefore necessary to devise a detailed plan to squeeze as much from the earth as possible.

Table 2.1: Garden crop data.


The first decision was what to grow. To help with this, I sifted through a number of gardening books and compiled Table 2.1, which lists time to maturity, spacing, and basic planting instructions for a variety of crops. It suggested that I concentrate my efforts on salad greens -- they are fast maturing, require little space, and are relatively high-value. By planting a mesclun-type mix I could take advantage of the recent popularity of the product and the spatial and temporal efficiencies of the greens.

I ordered seeds from The Cook's Garden, a company in Vermont that specialises in salad greens. I chose non-hybrid varieties, both to allow me to save my own seeds, and to address my concerns about genetic diversity. Some may question the decision to patronise an American company, but they offer the best service in the salad green market that I was able to find.


Once the crop was chosen, the mechanics of cultivation had to be worked out. I decided to devote 10 beds to the lettuce and greens mix, and the remaining five beds to edible flowers, herbs, and areas for testing the practicality of the 3" soil depth for other vegetables.

Most of the greens (lettuce and mustard, primarily) have a maturation period of approximately 50 days, or seven weeks from seed. I reasoned that I could start the plants in soil blocks (explained below), keep them in the blocks for three weeks, and transplant outside to spend the remaining four weeks in the garden. A given crop would therefore only require outdoor space for four weeks. This meant I could plant two beds to salad mix each week (Figure 2.5). After four weeks, the first pair of beds would be emptied, and could rest fallow for one week before the space was needed by another crop. In essence, each of the five pairs of beds would go through a five week rotation (four weeks in crop, and one week fallow) before being needed again.

I chose not to broadcast all seeds into the beds (this is the usual practice with mesclun) for two reasons. The first was my effort to minimise the time each bed was occupied -- most of the crops grown from seed would require a full seven weeks outdoors, as opposed to the four weeks needed by transplants. The second reason was that I was concerned about the growth medium. The same compost had been used the year previously, but I was told that plants did not do very well in it. I figured that transplants would have more resilience than sprouts, especially if damping-off fungus was present in the soil. The transplant seeding schedule is shown in Table 2.2.

Table 2.2: Transplant seeding schedule.

I used Bartholemew's (1981) square foot gardening method for spacing the transplants. For the lettuces, I bent a wire coat hanger into an 8" square, and used this to mark the location of each transplant. The other greens were transplanted at 6" spacing. To mark this out, I took a 12" piece of board, tamped its edge into the soil four times to make a square, and then used the board to make a cross in the middle of the square. This forms four smaller squares, each 6" apart. (This sounds like an involved process, but one soon becomes quite proficient at it, and I believe that for small beds it is faster than using a line to mark the rows.)

Soil blocks

I chose soil blocks for starting seedlings because they have a number of advantages over the traditional use of pots and flats. According to Coleman (1989), they are ideal because they avoid completely the transplant shock associated with most other methods of preparing transplants. Normally, transplant shock slows plant growth for a period while the plant repairs root damage and adjusts to its new environment. Soil blocks allow uninterrupted growth after transplanting because there is no root disturbance -- this is how I planned to grow a seven week crop in seven weeks with only four weeks spent outside.

Soil blocks are made with a block maker, or blocker -- a small form with a spring loaded, manual ejection mechanism (Photo 2.10). The blocker is pressed into moistened blocking mix, and the blocks are ejected onto a surface for planting. Each block is a cube of soil (2" per side with my blocker), and has a small indentation on the top for holding a seed. Only one seed is planted per block, and the seed is not covered. According to Coleman (1989) germination is excellent, as the seeds have access to all the moisture and oxygen they require.

My blocking mix recipe was based on the one given by Coleman. It contains:

To make the mix, the two peats are passed through a 1/4" screen and then mixed with the lime to balance their acidity. Sand or vermiculite and the fertilisers are then added and mixed, followed by screened potting soil. The peat provides the fibre necessary for the initial stability of each block (the blocks become very durable once penetrated by plant roots). The sand or vermiculite encourages drainage in the block, and the fertilisers provide all of the nutrients required by the seedlings, and ensure healthy, sturdy transplants. Coleman uses a soil and compost mix as the last component, but I had no source of compost or soil when the first blocking was required, and chose potting soil as a reasonable substitute. Its purpose is to provide more nutrients and bulk to the blocking mix. One part water is added to three parts blocking mix to produce the mortar-like consistency required by the blocker.

Figure 2.5: Typical plantings.

Watering the seeded blocks can be difficult, as they are prone to erosion before the plant's roots develop. To avoid this problem, I purchased a pair of capillary mats (Photo 2.10). These mats are designed to wick water from a reservoir, and spread it evenly over their surface. Blocks resting on the mat will draw as much water as they can hold through similar capillary action. The mats would take care of all transplant watering.


I decided to concentrate my marketing efforts on local health food stores and restaurants. To assist in this process, I produced a small pamphlet, outlining the idea behind the garden and the produce I expected to harvest (Figure 2.6), and I registered the garden as a business with the Ontario Ministry of Consumer and Commercial Relations. I ruled-out the possibility of selling at a farmers' market, as I did not expect to have enough produce to attract many customers. According to Lee (1993), farmers' market sellers must offer a number of different produce items if they expect to be successful, and my small garden could clearly not meet this criterion. I hoped instead to find a small, environmentally conscious restaurant or store proprietor interested in supporting urban agriculture.


The Ontario Farm Products Grades and Sales Act (1980) has no mention of mesclun, and hence it is not regulated with regard to product grades. Packaging, however must conform to certain standards. It must identify the grower, the product if the package is not clear, the weight, and the price. I decided to follow the example of other growers and package the produce loose in boxes for restaurant sales, or in individual zip-closure bags for retail.

Health Canada regulations regarding standards for cleanliness and packaging are concerned primarily with the cleanliness of processed foods, dairy products, and meat products, and the use of food-grade plastics for packaging (Government of Canada, 1994). My impression from reading the overview of the standards is that food would have to be fairly unclean to warrant a complaint. In any event, I planned to sterilise all containers used for washing the salad mix with a weak bleach solution, and to package the mix in food-grade zip-closure bags. The mix would be washed with water from the municipal supply. These precautions would ensure that the packaged salad mix fell within the regulations.

Field to Table has a small institutional-type kitchen on the premises for catering and training. I planned to use this for washing and packaging the greens.

Figure 2.6: Promotional pamphlet (both sides).

Chapter 2 (The Site and the Plan)
Chapter 3 (Observations)
Chapter 4 (Conclusion), and References

pointer Return to Contents' Page pointer

Revised February 24, 1997

Published by City Farmer
Canada's Office of Urban Agriculture