The roof as farmland

Seen from above, a city is startlingly flat. Slopes, rivers and parks are minor exceptions in the aerial view; most of what fills the frame is the roofs of buildings. A considerable share of those roofs are flat roofs without pitch — that is, flat land. It is firm enough to walk on, better lit than anywhere at ground level, and has air moving across it. And yet what actually sits there is usually a waterproof membrane, some air-conditioning units and an inspection ladder. Urban planning has long declined to count this area as area at all. Floor-area ratios and building-coverage ratios count floors and sites, but the plane on top of the roof has not even been given a name.

This series begins with that act of naming. When we look at a roof not as leftover space but as farmland, what becomes possible, and what stands in the way as a constraint? To give the conclusion first: what decides whether rooftop farming is feasible is neither enthusiasm nor technology, but the weight a building can bear and where the rain that falls on it goes. And most of the roofs that have actually been greened despite those two constraints came about not because an owner decided so of their own accord, but because an ordinance demanded it. The roof is the city's most demanding farmland — the place where physics and regulation meet head-on.

There are two reasons to discuss roofs now. One is a tailwind. As urban overheating and short, violent downpours both grow more serious, the roof is being reappraised as a device that cuts heat and catches rain. The US Environmental Protection Agency states that the surface temperature of a green roof can be up to 56°F (about 31°C) lower than that of a conventional roof, and that runoff can be reduced by around 60 per cent for extensive roofs and by up to nearly the whole volume for intensive ones. Putting soil and plants on a roof is ceasing to be a matter of gardening and becoming a matter of urban infrastructure investment.

The other is a headwind. What actually spread green roofs was mandates written into ordinances — and those mandates are now being reconsidered in a number of places. Toronto's bylaw, the first in North America to make green roofs compulsory, was reported repealed by a provincial measure in October 2025, returning green roofs to voluntary status. Just as the tailwind strengthens, the institutional floor is being pulled out. That asymmetry is precisely why roofs deserve discussion now. In what follows we first put the words and the areas in order, and then move on to the question of that floor.

Putting plants on a roof divides into four practices with quite different purposes and quite different people behind them. The first is roof greening. Its aims are environmental performance and appearance, and what it grows is sedum, turf and other plants that survive without attention; people generally do not walk on it and nothing is harvested. The second is the rooftop kitchen garden — residents or employees growing vegetables on the roof of an apartment block or an office, somewhere between hobby and practical benefit. The third is the rooftop farm: commercial production for sale. New York's Brooklyn Grange began in 2010 and now cultivates roughly 5.6 acres (about 2.3 hectares) across three roofs, shipping some 80,000 pounds (about 36 tonnes) of organic vegetables a year.

The fourth is the rooftop plant factory: a greenhouse or closed cultivation system placed on the roof, producing year-round under controlled conditions, mostly hydroponically without soil. By this point the fact that it is a roof has arguably lost its meaning — the criticism that it is merely using cheap surplus floor space is a fair one. What this series mainly treats is the second and third categories, the rooftop kitchen garden and the rooftop farm, where soil and people are involved. The essential thing is not to argue about all four as if they were one, and yet statistics and regulations frequently press all four into the same figure without distinguishing them.

This conflation does real damage. Most of the statistics showing that green roofs have advanced, for example, refer to thin extensive roofs planted with sedum, where no food is being produced at all. Conversely, if we ask only about yield, the original benefits of greening — catching rain, lowering heat — vanish from view. Separating the four words is not mere tidiness; it is a question of the axis on which we measure success. The distinction will keep doing work in the map we set out at the end of this article.

Japan has a statistic that counts the area of green roofs every year: the Ministry of Land, Infrastructure, Transport and Tourism's national survey of rooftop and wall greening work, running since 2000. According to the latest results, about 15.7 hectares of new green roof and about 5.0 hectares of green wall were installed in 2023, bringing the cumulative totals since 2000 to roughly 615 hectares of roof and 125 hectares of wall. The ministry headlined the 2023 figures as the creation of green roofing equal to three Tokyo Domes. Note that the survey is a postal questionnaire sent to 560 firms nationwide, of which 284 responded (a 50.7 per cent response rate) — it is not a complete enumeration.

How to read these figures is the question. A cumulative 615 hectares is about thirty-eight times the area of Tokyo's Hibiya Park — hardly trivial. Look at single years, however, and the increase is essentially flat: about 15.6 hectares of roof in 2022 and about 15.7 in 2023. The momentum of the era around the turn of the millennium, when ordinances were put in place and green roofs got their institutional start, has already passed. Cumulative totals only ever go up, so every year is a record; but the flow of new installation is not growing. Looking at the cumulative number alone hides that stagnation.

More important still is the denominator. Whether 615 hectares is a lot or a little cannot be judged without comparing it with the total roof area of Japan — and here lies this field's weak point: there is virtually no continuous official statistic capturing the total flat-roof area of Japanese cities. For reference, about one in ten of all roofs in Germany is said to be greened. It is hard to believe Japan's rate is anywhere near that level, but no accurate comparison is possible because the denominator is missing. The intuition that there is enormous room on our roofs has, in fact, never been properly quantified. This series starts with area precisely in order to make that blank space explicit.

What can be placed on a roof is decided, in the end, by structural calculation. The weight of greening is generally put at roughly 50–120 kilograms per square metre for thin, light extensive systems and 390–730 kilograms per square metre for intensive systems with deep soil. Japanese sources offer rules of thumb of about 300 kilograms per square metre for turf and about 1,000 for trees, assuming saturated soil. The figures vary so widely because weight changes completely with soil depth, water content and the plants chosen — and none of these rules of thumb means anything unless it is assessed in the saturated state, when the roof is at its heaviest after rain, rather than dry.

This single point binds the design of rooftop farming at its root. On an existing building, one cannot exceed the imposed load assumed when it was designed. That is why real rooftop farms pile up workarounds instead of piling up soil: lightweight artificial growing media, heavy elements placed directly over beams, drainage layers that refuse to hold standing water. The ground-level common sense that vegetables grow if you dig deep does not carry over unchanged. The roof is farmland stripped, at the outset, of the most basic freedom of all — depth of soil. Tomorrow's second instalment takes that constraint head-on.

The other physics is water. A roof has no groundwater, and no water rising from below by capillary action; the only sources are the rain that falls and the water people carry up. The thinner the soil the faster it dries, and strong wind and sun speed evaporation further. A rooftop farm without irrigation is destroyed within days in midsummer. Yet that same water is a threat to the building: a single tear in the waterproof layer becomes a leak into the floor below, which is why drainage and root barriers — layers that stop roots piercing the waterproofing — matter more, as equipment, than the greenery itself. The roof is farmland built on a contradictory condition: never let the water run out, and never let a drop get inside.

Green-roof area rose sharply in the early 2000s not because technology suddenly advanced, but because regulation moved. Tokyo's Ordinance on the Protection and Restoration of Nature in Tokyo came into force in April 2001; under Article 14, anyone constructing or extending a building on a site of 1,000 square metres or more (250 square metres or more for sites held by the national or local government) must file a greening plan covering both the ground level and the building itself. For roofs and similar surfaces, greening of roughly 20 per cent of the usable area is said to be required. It is no coincidence that the ministry's statistics begin in 2000: that is exactly when this institutionalisation happened.

The same thing happened abroad. In May 2009 the City of Toronto passed the first bylaw in North America to make green roofs compulsory. It applied to new buildings with a gross floor area of 2,000 square metres or more, on a sliding scale requiring 20 to 60 per cent of available roof area to be greened. More than 1,200 roofs are said to have been greened under the bylaw, and the approach was copied by cities in other countries. Regulation created demand, demand grew contractors and material manufacturers, and the resulting industry brought costs down — the spread of green roofs proceeded in that order.

Which is precisely why the reverse can also happen. Toronto's mandate was reported repealed by a provincial measure in October 2025, returning green roofs to voluntary status: the first city in North America to require them became, sixteen years later, the first to lose the requirement. The lesson to draw is that rooftop greenery has not taken root because it is technically established, but exists only for as long as regulation supports it. The reversibility of institutions is a theme the sixth instalment takes on directly, yet it is also a premise worth stating on day one.

This series inspects the roof as farmland from seven angles over seven days. The second instalment is science: the physics that make a roof work — load, waterproofing, drainage, lightweight growing media, wind and heat. The third is history — where the idea of putting plants on a roof came from, and how, in Japan, it connected to the distinctive culture of the department-store rooftop. The fourth surveys cases around the world, tracing how cities such as New York, Chicago, Basel and Singapore reached for their roofs, each from a different motive.

The fifth takes stock of where Japan stands: the stagnation shown in the ministry's statistics, what Tokyo's ordinance actually accumulated, and how far corporate and residential rooftop farms have come. The sixth turns to institutions and economics — mandates, subsidies, tax measures, cost structures and business models — and confronts what Toronto's repeal really meant. The seventh faces the controversies and the limits: whether rooftop farming genuinely contributes to a city's food supply, whether mandates have become an alibi for greening, and, granting all that, where a citizen who wants to start should actually begin.

One question recurs throughout the seven days: by what do we measure a roof? Measured by yield, rooftop farming is too small; by cost, too expensive; by appearance, too subjective. But count runoff reduction, roof surface temperature, corridors for wildlife and the number of places where people can put their hands in soil — count all of those together, and the assessment changes entirely. Not reaching a conclusion from a single indicator: that is this series' method.

Let us retrace, in order, the path this article has taken. We began by confirming that the flat roofs spreading beneath an aerial view of the city are a plane to which the language of urban planning has never given a name. We then separated four words — roof greening, rooftop kitchen garden, rooftop farm and rooftop plant factory — and noted the damage done when statistics and regulations mix them together. From there we counted the area, surveyed the two physics that govern a roof (weight and water), and finally saw that what actually created that area was regulation rather than the market, and that regulation can be withdrawn. Beginning with definitions and landing on institutions: that is the design of day one.

Three points deserve restating by name. First, the stagnation of area. Japan's green roofs have accumulated to about 615 hectares since 2000, but the annual flow is essentially flat — about 15.6 hectares in 2022 and about 15.7 in 2023. A cumulative record conceals a stalled flow. Second, load as a prior condition: a ceiling of roughly 50–120 kilograms per square metre for extensive systems and 390–730 for intensive ones decides soil depth and crop choice before anything else is decided. Third, the reversibility of institutions: Tokyo introduced its requirement in April 2001 and Toronto in May 2009, and that Toronto mandate was reported repealed in October 2025. What regulation builds, regulation can remove.

At the same time, what we do not know must be stated plainly. The largest blank is the denominator. How much flat-roof area exists in Japan's cities, and how much of it could structurally bear greening or cultivation — no continuous official statistic capturing this could be confirmed. The premise of this article, that there is great room on our roofs, is therefore strictly speaking untested. The ministry's survey, too, is an estimate based on a postal questionnaire (sent to 560 firms, answered by 284), not a complete enumeration; nor can the statistics separate out how many greened roofs actually grow food. The likely objections are fair ones: rooftop farming costs a great deal per unit area, contributes only marginally to a city's food supply, and mandates push up development costs. This article's position has been that yield is not the answer to that objection.

As the first instalment of the series The roof as farmland, this article has laid the foundation the remaining six will share: the distinctions between the words, the current state of the area, the two physics, and the dependence on regulation. Tomorrow's second instalment specialises in those two physics. How load is calculated; how waterproof and root-barrier layers are stacked; why lightweight growing media are light and yet still hold water; what heat and wind on a roof do to plants. What today was left at the level of survey will be taken down, tomorrow, to the level of design — an instalment about understanding the area above the city as a piece of equipment.