Given the higher load of insects in humid subtropical climates, it was necessary to grow vegetables within a structure that can keep out as much as possible insect pests, thereby minimizing pesticide applications. Construction of the 150-m2 screen house in 2011 by an outside contractor took three construction workers 7 days to put up at a cost of ~¥86,000 RMB (Yuan unit of renminbi, Chinese currency) that included materials, construction, and a 15-year warranty, the length of time that the screen house is expected to last. This cost includes two changes of the plastic film as the films need replacing every 5 years. Temperature within was usually the same as ambient temperature except during sunny days when the temperature is generally 2 °C higher with the sunshade nets used. Without the sunshade nets, it was higher by as much as 8.5 °C in the winter and 15 °C in the summer. Light intensity was up to 38 % lower than outside the screen house, but still well above the level needed (20,000 lx) for most vegetables. Since construction, and up to the present time, the roof screen house has been strong enough to weather heavy rain including through four typhoon seasons. The total cost for the 14 hydroponic units, consisting of main tanks, stainless steel support legs, cover plates, solution tanks, water pumps, timers, and other miscellaneous items was ~¥35,000 RMB.
Feasibility of production
Leafy vegetables spoil easily after harvest and are not well suited to long distance transport. If grown within urban areas, the reduced distance to consumers may help preserve freshness as well as lower transportation costs to make them competitive against those grown outside urban centers. We tested 7 different locally grown leafy vegetables listed in Table 1. Yield per hydroponic tank was experimentally obtained for each vegetable in a typical production cycle, defined as the number of days to harvest from transplanting seedlings onto the hydroponic setup. The germination and seedling stage in nursery trays typically adds another 7 to 15 days to the length of time from seed to harvest, but are excluded from the production cycle as they do not require time in the hydroponic tanks. Other data in Table 1 are extrapolated based on using all 14 tanks of the 150 m2 screen house. From our observations, only leaf mustard grew well throughout the year. Hence, the number of production cycles per year from this simple screen house setup is as low as 2 cycles for potherb mustard, but up to 9 cycles for leaf mustard.
Although configuring temperature control for cooling could extend the growing season for many of the vegetables, a more cost-effective solution would be to adjust the types of vegetables to grow. For example, by planting 3 cycles of Italian lettuce from November to March followed by 5 cycles of leaf mustard from April to October, the 150-m2 screen house has the capacity for year-round production of 1800 kg (12.1 kg/m2) of leafy vegetables, or an average of ~5 kg per day. According to the China Nutrition Society’s dietary guidelines, each adult should consume 0.3–0.5 kg vegetables per day (Liu, 2007). If we accept a simple assumption that leafy greens should account for about a third by weight of the total vegetables consumed, 5 kg per day could provide the green leafy vegetable needs for 30–50 individuals.
If the same roof area (252 m2) were on top of a residential building, each floor below would likely comprise of two apartments of ~110 m2. Assuming each floor houses 8 persons, or 2 families of 4, that would mean the 150-m2 screen house has the capacity to provide the green leafy vegetable needs for 4 to 6 floors of inhabitants. Realistically, however, it is unlikely that people would want to be confined to only leaf mustard and Italian lettuce, as suggested for maximum production, or even to the limited list of green leafy vegetables shown in Table 1. Most likely, they would rather supplement their diet with other leafy greens from the market. Given that assumption, roof farming would provide only a portion of green leafy vegetable needs. Nevertheless, even if a 150-m2 screen house provides just a quarter of the green leafy vegetables for 16 to 24 floors of inhabitants, it is still a significant contribution from roof space.
Safety of production
With Italian lettuce, all samples showed nitrate content below the maximum residue limit (Fig. 2). However, 23 of 49 Chinese flowering cabbage samples from the market had nitrate content exceeding the maximum residue limit, even those labeled as pollution-free/green or organic. In contrast, none of the hydroponic vegetables exceeded the maximum residue limit for nitrate. The presence of pesticides has been a concern since the Agriculture Bureau of Guangzhou-Vegetable Office reported that 10.7 % of the vegetables contained excessive pesticide residues when 54 kinds of pesticides were determined by high-pressure liquid chromatography (Guangzhou Agriculture Website 2012). When the less sensitive acetyl cholinesterase inhibition method was used, the rate of detection was only 0.9 %. We had previously conducted a farmer’s survey of the various pesticides used in the Guangzhou area and noted that organophosphate and carbamate insecticides were in common use (Yang et al. 2014). With the acetyl cholinesterase inhibition assay, we found that among 98 market samples, only one common Chinese flowering cabbage sample showed clear contamination, while surprisingly, one pollution-free/green Chinese flowering cabbage showed slight contamination (data not shown). In contrast, these pesticide residues were not found in our hydroponically grown vegetables. During our 2 years of growing vegetables, pest infestation occurred in three instances: aphids in the autumn of 2012, diamondback moths in the spring of 2013, and gray mold in lettuce during the rainy days of spring 2013. However, in each instance, they were effectively controlled by spraying low doses of low toxicity avermectin pesticide immediately after the occurrence, or by removing infected plants and cleaning the affected facility. Overall, there was not much of a need for using pesticides.
With respect to heavy metals, routine quality and safety monitoring by the Guangzhou Agricultural Product Quality Safety Supervision and Inspection Center showed that in 2011, 5.3 % of the market vegetables contained lead, cadmium, chromium, or mercury above the maximum residue limit (Guangzhou Agriculture Website 2012). In our analysis of 109 samples, excessive amounts of Cd, Cr, or Hg were not found (data not shown). However, one Italian lettuce sample and four Chinese flowering cabbage samples exceeded the maximum residue limit for Pb (Fig. 2), and two Italian lettuce samples exceeded the maximum residue limit for As (Fig. 2). The contamination was not confined to common vegetables, but included pollution-free/green and organic samples. In contrast, all five heavy metals were low in the hydroponic vegetables. This is likely due to the ability to control the content of the hydroponic solution, which aside from plant nutrients was otherwise municipal grade tap water. This agrees with the report by Antisari et al. (2015) that soil-less planting systems can reduce the accumulation of heavy metals, such as up to 71 % in rosemary leaves.
Quality of production
As for minerals, hydroponic Italian lettuce contains the highest Ca, K, Mg, and Fe but the lowest Zn content, while hydroponic Chinese flowering cabbage contains the highest Ca and K but lowest Fe and Zn (Fig. 3). Common and organic vegetables were found to have higher crude fiber than pollution-free/green and hydroponic vegetables (Fig. 3), and this may be due to their outdoor growth, as opposed to the hydroponic and pollution-free/green vegetables that thrive in relatively protected environments. Less fiber can be desirable as the vegetables have a more tender texture. As shown in Fig. 3, the vitamin C content of the hydroponic vegetables was higher than their market counterparts.
This agrees with other studies that also concluded high-quality vegetables from soil-less cultivation (Gruda 2009).
Cost of production
The cost in year 2011 for erecting the roof screen house and its interior was ~¥120,000 RMB. Spreading the cost over the life expectancy of the facility and equipment, the cost for the facility is ¥62.7 RMB/m2/year. Because differences exist among vegetables with respect to planting density, nutrient solution, workload, production cycle, and unit yield, the costs were calculated individually for each type of vegetable (Table 1). Cost of facility and equipment range from ¥15.5 to ¥62.7, consumables ¥2.8 to ¥12.5, and labor ¥14 to ¥63 RMB/m2/year. The cost of consumables (less than 10 % of the total) includes utilities which were rather inexpensive. Water was used for the solution and the occasional cleaning of the tanks and floors, and electricity was mainly for small water pumps that operate only 5 min every 2 h. This works out to a total (facility, consumables, labor) cost/kg of ¥6.8 RMB for Italian lettuce to ¥50 RMB for caraway. Due to the clean production method, we expect that the hydroponic vegetables can be comparable in quality to pollution-free/green vegetables. Therefore, the typical market prices of pollution-free/green vegetables were used for reference (Fig. 4). For all vegetables except for caraway, the cost of production is less (27 to 52 %) than market price (Fig. 4), with highest potential profit for leaf mustard at ¥296 RMB/m2/year (¥44,400 RMB/150 m2 screen house). This is even higher than the potential profit for maximum kilogram yield from 3 cycles of Italian lettuce and 5 cycles of leaf mustard (¥282 RMB/m2/year) due to the lower market value of Italian lettuce, although these figures do not include the costs for rent, distribution, packaging, and advertising.
Substantial distribution cost can be averted if vegetables from the roof are sold to building residents. In such a scenario, vegetables would only need an elevator ride for distribution to individual households within a residential building. As for packaging, instead of the polystyrene foam trays and plastic wrappings of premium vegetables in supermarkets, freshly harvested ones can be delivered with reusable bags or baskets. Such a distribution system not only reduces labor cost, but also ensures freshness. Rent, however, is difficult to estimate for the cost of space that customarily is not used. As a reference, land for vegetable production in local suburbs is ¥5 RMB/m2/year according to our survey. Even if it is tenfold higher, at ¥50 RMB/m2/year, it would only amount to ¥12,600 RMB per year for the entire 252-m2 roof, reducing the estimated leaf mustard profit from ¥44,400 (¥296/m2) to ¥31,800 RMB.
We specifically chose to test only leafy vegetables because they are readily perishable and would benefit the most from local production. They also have a relatively shorter production cycle that reduces cost. Most of the biomass is consumable, whereas growing crops with a high amount of plant waste, such as rice stalks that must be transported from the roof for disposal, would increase the cost of production. Our study also benefitted from the geographical advantage of the warm South China climate that precludes the need for a glass or polycarbonate encased greenhouse. A relatively inexpensive screen house is sufficient to keep insects out and protect crops from heavy rain. Eliminating glass or polycarbonate not only reduces construction costs, but enhances safety by reducing the possible shattering and falling of dangerous materials from the roof. Hence, we do not view our experience in roof farming in a warm climate as universally applicable. However, even if it is confined to Guangzhou with its 13 million residents, the Guangdong Province with 104 million residents, or to South China with an even a larger population, it is a sizable geographical area with ample potential for this type of roof farming practice.
As for alternate uses of roof space, rooftop greening has received popular attention with many local governments as a way to combat urban air pollution (for review, see Li and Babcock, 2014), but if it is merely gardening, it most likely does not generate direct profits. As for solar energy, current technology should be able to generate ~546 Wh/day/m2 using the most efficient solar panels (Li et al. 2013). In the case of our 252-m2 rooftop, it could potentially generate some ~138 kWh/day of electricity with some losses in conversion efficiency, although the smog in most Chinese cities would likely reduce this production potential. That might be sufficient to power eight households without air conditioning. However, solar generated electricity is still rather expensive and currently requires government subsidies.