Land reclamation in San Francisco Bay: a primer

Introduction

Many coastal American cities have been built largely on reclaimed land formed by dumping soil into shallow water. Since the slow-growth environmentalist revolution of the 1960s and 1970s, this largely stopped in the United States.

Suggestions for resuming land reclamation crop up with some frequency in urbanist circles, with the San Francisco Bay Area being the most tempting target for reasons that shouldn't require too much explanation. The point of this Q&A is to outline the economic case for land reclamation in the Bay and a few of the difficulties that land reclamation projects might encounter, with the ultimate conclusion that the drawbacks are, for the most part, mitigable.

Disclaimer: this isn't anything close to a feasibility study, which would require specific site studies as well as engineering knowledge beyond what I gleaned from a few textbooks on geology and soil engineering. Still, this essay has more detail than any of the other cursory popular discussions that I've seen. I'd appreciate any comments or corrections, especially from anyone with experience in geology or soil engineering.

How much land has already been reclaimed in the Bay Area, and how much is left?

The best source I've found is a 1959 report by the U.S. Army Corps of Engineers, explaining how the Bay Area would accommodate its predicted explosions in population. These predictions make tragicomic reading today: they estimated a Bay Area population of 14.42 million people by 2020 (actual 2020 Census figure: 7,765,640), including 790,000 persons in Marin County (actual figure: 262,231).

The report's section on land reclamation claims that about 242.8 square miles of land in the Bay Area had been reclaimed in the years 1850 to 1957. This includes 94.8 mi2 in "San Francisco Bay proper" (which is the area south of Point San Pablo, or roughly the location of the Richmond–San Rafael Bridge). Land reclamation projects in the Bay were functionally banned six years after the report's publication, so the figures today should be pretty similar.

The report estimated (p. 88) that there were 80 mi2 of marshlands and 250 of tidal and fully submerged land still "physically susceptible to reclamation". This includes 165 mi2 in San Francisco Bay proper (p. 79), including 20 mi2 of marshland and 145 mi2 of "tide and submerged lands." (These figures don't include 60.1 mi2 of salt ponds in the South Bay, which the report considered already reclaimed but are prime candidates for redevelopment.)

The USACE report ultimately concludes that most of this land would not be economically reclaimable under the economic conditions of the time. But economic conditions today are obviously much different. In its discussion of why reclaimed land could not be profitably sold for industrial use, the USACE mentions that only a handful of businesses would be able to pay land prices of $25,000 per acre (roughly $250,000 today), and these businesses had other reasons to want to locate in or near central business districts. Of course, San Francisco's high-tech industries today are phenomenally more productive than anything foreseeable in 1959, and $250,000 per acre for Bay Area land is almost nothing by today's standards: only out-of-the-way exurbs have land that cheap.

Where are there ongoing land reclamation projects?

For the most part, not in the United States. One of the earliest successes of Bay Area environmentalism was getting the California state legislature to pass the McAteer–Petris Act in 1965, giving jurisdiction over all land reclamation projects to a newly formed commission that prohibited most of them. (More recent state and federal environmental laws, of course, add even more layers of difficulty.)

The biggest current land reclamation projects are in port cities in Asia: several of these are now very high-density developments such as the Songdo special economic zone near Seoul, South Korea. Perhaps the example most comparable to California is Japan, both for economic reasons—Japan is the wealthiest nation in Asia—and because the Bay Area and Japan have similar geographic and seismic challenges. The coastal waters of Japan's core industrial cities, especially Tokyo and Kyoto, have been extensively filled with artificial islands: the most ambitious recent land reclamation project was Kansai Airport, constructed on an artificial island off the coast of Kyoto in 1994. (In 1958, one government official even proposed building an entire "Neo-Tokyo" by filling in two-thirds of Tokyo Bay—the requisite earth to be obtained, in a stunning display of Fifties Mindset, by leveling nearby mountains with nuclear bombs. It's perhaps for the best that this didn't happen.)

Why does land reclamation make sense in San Francisco particularly?

A few reasons. First, the Bay is very shallow: raising the ground above water level would not require much landfill. As these charts from the National Oceanic and Atmospheric Administration show, most of the Bay is much less than 10 feet deep, especially in the southern reaches near Silicon Valley and San Jose (with the exception of a shipping channel dredged as far south as the Dumbarton Bridge).

Land in the Bay is extremely valuable—several million dollars (or even more) per acre, if housing prices are anything to go by (though exact values depend on location and zoning ordinances). New land not subject to existing zoning restrictions would be doubtless even more valuable.

Land reclamation would be especially valuable for industrial uses. High land prices have squeezed much industry out of the Bay Area, to the point that San Francisco has specific provisions in its zoning ordinances to protect remaining industrial land from competition. I criticized these ordinances in San Francisco and several other cities in a report for the Manhattan Institute in 2021: they are often tailored to protect whatever low-productivity industries happened to exist at the time, and some of the land parcels are in highly transit-accessible areas that would be better put to higher-density uses in any case. But the point remains: more industrial land somewhere would be useful, especially if it came in large parcels that could become viable warehouses and factories, and would also benefit the growing number of startups that (in Peter Thiel's phrase) are working with atoms, not bits.

Land along a shore would be especially useful for factories because they could obtain and send shipments by water, a much more labor- and fuel-efficient form of transport than truck. Barges can operate in quite shallow water: the usual draught for a fully laden barge on the Rhine, for example, is only 1.5 m or approximately 5 feet, and special shallow-water barges can operate down to one foot.

How much does land reclamation cost?

It's surprisingly difficult to find unit costs. The only recent estimate that I've been able to find is a remark from a government ministry in Singapore: "The cost of land reclamation projects completed in the last 10 years ranges from around S$270 to S$850 per square metre." One Singaporean dollar is roughly US$0.74, and there are 4,047 m2 per acre, so this works out, in round numbers, to about $800,000 to $2.5 million per acre in reclaimed land.

Even the higher end of this range would be economical at Bay Area land prices. I don't have any good notion of where San Francisco Bay landfill would fall within the range: on the one hand, the waters in the Bay are unusually shallow; on the other hand, seismic precautions (discussed more below) would require more advanced engineering, and I don't have a decent gauge on what this would do to prices.

The 1959 Army Corps of Engineers report gives another frame of reference when it quotes (on p. 85) a wide range of costs for historic projects, giving an average cost of $8,000 per acre for projects over the prior two decades and estimating that a comparable average project at 1959 price levels would cost about $14,000. Many of these projects were filling marshland rather than truly submerged land: on p. 92, it estimates that filling submerged land (as opposed to marshland) is likely to cost closer to $30,000. These figures are of course 1950s prices: inflation since then has been roughly a factor of ten. And for various reasons—more expensive labor and stricter environmental regulations on the one hand, but better labor-saving technology on the other—public works prices may vary in real terms quite a bit from the levels of seven decades ago. But if land reclamation projects are possible for $300,000 per acre today, the economic case becomes even more clear-cut.

The main cost driver for land reclamation recognized in the USACE report, furthermore, was the cost of sand to use as fill, quoted as $1.50 to $2.00 per cubic yard on p. 87. This is quite similar (after inflation adjustment) to the cost today, which a recent USGS report estimates at $12.20 per metric ton (the density of sand is roughly 100 pounds per cubic foot, or 1.25 metric tons per cubic yard). And there's another possible source of fill: using treated consumer waste, as is still done in Tokyo Bay. (The USACE mentioned this possibility but dismissed it on sanitary grounds, but modern waste treatment technology should address these problems well.)

Reclaiming land by building a dike around it and pumping the water out in the manner of Dutch polders, as the USACE acknowledged, is an order of magnitude cheaper. (A proposal to do this in in the Bay in the 1940s, the Reber Plan, did in fact acquire some level of local support.) But diking the Bay would have much more severe ecologica impacts, and failure of the dike would cause a massive flood, so it's understandable if most people end up leery of this possibility.

Would there be costs from impinging on maritime industries in the Bay?

Not really. The main maritime industry (loosely speaking) in the parts of the Bay best located for land reclamation is salt production: Cargill produces 500,000 tons of salt per year from salt ponds in the south part of the Bay. This may sound like a lot, but it's barely 1 percent of America's total production of salt, which currently stands at about 42 million tons per year. The going price for "vacuum and open pan" salt is $220 per ton according to a report by the US Geological Survey, so the Bay Area salt industry is producing about $100 million annually. Compared to the whole economy, this is a rounding error, especially when compared to the amount of space required—the salt ponds, according to Wikipedia, occupy about 16,500 acres, so we're talking about about $6000 per acre per year. In a region as productive as the Bay Area, almost any other use would be an improvement. As an indicator of the lack of productivity from the salt ponds, in 2003, Cargill sold many of its ponds to various governmental and nonprofit organizations to restore them to a wild state.

The Bay also has a small herring and salmon fishing industry (and, in fact, several fishing seasons have been canceled by order of the state government in order to stop the fishing industry from driving the fish stock completely to extinction), but as best as I can tell—which isn't very well—these are mostly in the estuary in the north of the Bay, far from the most plausible candidates for land reclamation; and in any case, I don't think that filling in shallow waters on the margins of the Bay should affect fish populations much one way or the other.

Would reclaiming land in the Bay cause environmental harms?

According to the relevant environmental protection statutes' de facto definition of environmental harms (i.e. making something look different from the early 1970s), yes.

Less facetiously: possibly also yes. Although the Bay is far from idyllic (and, for that matter, it's heavily polluted by mercury used in Gold Rush mining operations), much of the existing marshlands in the South Bay, including some active salt ponds, are part of the Don Edwards National Wildlife Refuge. Annoyingly, the refuge occupies most of the waters closest to Silicon Valley.

The refuge does provide habitat for a few threatened bird and rodent species, and I'm not sure if the impact of de-accessioning part of the refuge could be mitigated by habitat restoration elsewhere. But even with the wildlife refuge off limits, there are still many miles of Bay shoreline that would be available to redevelopment. Some of this would be subject to height restrictions due to proximity to San Francisco and Oakland Airports, but these aren't terribly onerous: about 1 foot in height per 200 feet from the runway endpoint, if I'm interpreting the relevant regulations correctly, so there should be fine.

And this whole discussion presumes the hyper-local tunnel vision approach to evaluating environmental harms that has caused so much trouble. The alternative to developing the Bay Area is not development nowhere: it's development in untrammeled nature further out—or, for that matter, in Houston or Shenzhen or any number of other places not known for sensitivity to ecological matters. Ultimately, a couple of endangered species might have to take one for the team.

Does reclaimed land pose risks in an earthquake?

Yes, but they're fixable.

Reclaimed land in an earthquake is especially prone to a risk called soil liquefaction. In essence: when soil is shaken, the soil particles can become separated from one another. But though particles in dry soil would simply resettle into a denser configuration, the water in moist soil can prevent soil particles from re-adhering to one another. This turns the soil into a viscous liquid, incapable of withstanding the shear forces imposed by the edges of building foundations.

Soil liquefaction has been responsible for much of the worst damage in eqrthquakes. For instance, in the 1989 Loma Prieta earthquake, areas of San Francisco that were built on historic landfill, such as the Marina district, suffered especially severe damage. This would be a concern for any Bay Area landfill as well: a map prepared by the Association of Bay Area Governments does, indeed, identify most of the existing coastline in San Francisco Bay as having moderate to high soil liquefaction risk in an earthquake.

The hazards, though, can be fought with modern soil engineering. A 2008 report by the California state government mentions several techniques for mitigating soil liquefaction hazards, such as building retaining walls at the edges of areas of reclaimed land and densifying the soil by "vibro-compaction"—inserting a vibrating column into the landfill to encourage soils to settle more densely. We have a good test of the effectiveness of these measures: according to a report from researchers at UC Berkeley on the 1989 Loma Prieta earthquake, areas of San Francisco built on landfill that had been compacted with modern techniques withstood the earthquake well and showed no signs of liquefaction, even as severe damage on adjacent areas of older, untreated fill.

Japan, also notoriously prone to earthquakes, is a worthwhile comparison. Kansai International Airport, the main airport for Kyoto and Osaka, was completed in 1994 on an artificial island in Kyoto Bay and suffered minimal damage in the Great Hanshin earthquake in 1995.