View source for Rain Gardens in SF; Capturing Urban Runoff

```wiki
__NOTOC__

== Rain Gardens in SF; Capturing Urban Runoff ==

Along San Francisco streets, small patches of grasses, flowers, shrubs, stones and mulch can easily be mistaken for ordinary landscaping. Many are something more. They are pieces of the city's stormwater system.

Known as rain gardens, bioretention areas, green bulb-outs and other forms of green infrastructure, these installations intercept water flowing from streets, sidewalks, roofs and parking lots. Instead of immediately sending that runoff into pipes, they give it somewhere to slow down, soak into soil and, in some cases, pass through a surprisingly effective biological and physical filtering system.

San Francisco has set an ambitious target: by 2050, the city wants green infrastructure to manage '''1 billion gallons of stormwater every year''', roughly one-tenth of the rain that falls on the city annually. The San Francisco Public Utilities Commission (SFPUC) says 575 projects have already been built or initiated and that, once completed, they will collectively manage more than 333 million gallons annually. [https://www.sfpuc.gov/programs/san-franciscos-urban-watersheds/what-green-infrastructure SFPUC: What Is Green Infrastructure?]

Those numbers make the program significant as municipal infrastructure. But recent research suggests another way to understand what San Francisco is building.

Rain gardens can remove tire particles and chemical contaminants from runoff. Their vegetation can provide habitat. Trees and plants can reduce heat. Soil can replace some of the absorptive capacity lost when a natural landscape is covered by asphalt and concrete. Strategically located projects might eventually form networks connecting larger patches of urban habitat.

Seen this way, San Francisco's rain gardens are not simply an alternative kind of storm drain. They are part of an experiment in whether a densely developed city can restore pieces of the ecological system it displaced.

__TOC__

== 1. The Billion-Gallon Experiment ==

San Francisco's rain gardens make more sense when viewed against the enormous underground system they are intended to assist.

For most of the city, rainwater and sewage travel through the '''same pipes'''.

San Francisco is the only major coastal city in California that operates a combined sewer system on this scale. During dry weather, the system primarily carries wastewater from homes and businesses. When it rains, runoff from roofs, streets, sidewalks and parking lots enters those same pipes. [https://www.sfpuc.gov/construction-contracts/construction-projects/outfallprojects SFPUC: Combined Sewer Discharge Outfall Projects]

That arrangement has an important environmental advantage. In cities with separate storm drains, runoff can flow directly to rivers, bays or the ocean with little or no treatment. Much of San Francisco's stormwater instead enters its wastewater-treatment system along with sewage.

The disadvantage appears when it rains heavily.

San Francisco's wastewater facilities normally handle roughly 60 million gallons per day during dry weather but have hundreds of millions of gallons of wet-weather treatment capacity. Around the city's perimeter, enormous underground transport-and-storage structures can temporarily hold approximately '''200 million gallons''' of combined sewage and stormwater for later treatment. [https://www.sfpuc.gov/sites/default/files/about-us/policies-reports/FY25-34_Capital_Plan_Report_AttB.pdf SFPUC FY25-34 Capital Plan]

Even that system has limits.

During the longest and most intense storms, the storage structures can fill. Rather than allowing combined sewage and stormwater to back up into streets and buildings, the system can discharge partially treated wastewater through one of '''36 combined-sewer discharge points''' around the Bay and Pacific shoreline. SFPUC says these discharges are about '''94 percent stormwater''' and now occur an average of roughly ten times per year, compared with more than 80 annual discharges before the transport-and-storage system was constructed. [https://www.sfpuc.gov/construction-contracts/construction-projects/outfallprojects SFPUC: Combined Sewer Discharge Outfall Projects]

Seen from that perspective, a gallon of rainwater prevented from entering the sewer during a storm is not merely a gallon of water saved.

It is a gallon that does not occupy pipe capacity, storage capacity, pumping capacity or treatment capacity at the moment when those resources are most valuable.

That is the engineering logic behind San Francisco's green-infrastructure program.

=== From Bigger Pipes to a More Absorbent City ===

For most of the twentieth century, the obvious response to increasing urban runoff was to build more conventional infrastructure.

Larger pipes.

More pumping.

More storage.

More treatment capacity.

San Francisco is still doing all of those things. Green infrastructure is not replacing the conventional sewer system. SFPUC is simultaneously investing hundreds of millions of dollars in major capacity improvements in low-lying areas including Wawona, Folsom and Lower Alemany. [https://www.sfpuc.gov/programs/san-franciscos-urban-watersheds/what-green-infrastructure SFPUC: What Is Green Infrastructure?]

But another strategy has developed alongside those projects:

'''Reduce the amount of rainwater that reaches the pipes in the first place.'''

Instead of making only the underground system larger, make the surface of the city more absorbent.

That is the idea behind San Francisco's goal of using green infrastructure to manage '''1 billion gallons of stormwater annually by 2050''', approximately one-tenth of all the rain falling on the city in a typical year. [https://www.sfpuc.gov/programs/san-franciscos-urban-watersheds/what-green-infrastructure SFPUC: What Is Green Infrastructure?]

As of 2026, SFPUC reports that San Francisco has '''built or initiated 575 green-infrastructure projects''' expected to capture more than '''333 million gallons annually once completed'''.

That puts the city roughly one-third of the way toward the billion-gallon target. [https://www.sfpuc.gov/programs/san-franciscos-urban-watersheds/what-green-infrastructure SFPUC: What Is Green Infrastructure?]

But the numbers require careful interpretation.

=== "Captured," "Managed" and "Kept Out of the Sewer" Are Not the Same Thing ===

Stormwater statistics can become confusing because several different outcomes are often described with words such as ''captured'', ''managed'', ''retained'' and ''diverted''.

A rain garden might temporarily hold water and release some of it later.

Another installation might infiltrate almost all of its water into native soil.

A cistern might collect rain for irrigation.

A green roof might retain some rainfall and release the remainder slowly.

A flow-through planter might filter water but ultimately send much of it back into the sewer.

All of these can be useful, but they do not have the same hydrological effect.

The city's experience with the '''Stormwater Management Ordinance''' illustrates the distinction particularly well.

Passed in 2010, the ordinance requires qualifying new development and redevelopment projects to incorporate onsite stormwater management. Large projects creating or replacing at least 5,000 square feet of impervious surface generally must meet defined runoff-performance requirements. [https://www.sfpuc.gov/sites/default/files/construction-and-contracts/design-guidelines/SMR_Brochure_flyer_June2021.pdf SFPUC Stormwater Management Requirements]

By July 2026, more than '''475 projects developed under the ordinance were actively managing more than 280 million gallons of stormwater annually'''.

But only about '''95 million gallons''' were being prevented from entering the sewer entirely through infiltration into the ground or harvesting for reuse. [https://www.sfpuc.gov/about-us/news/sf-green-sponge-stormwater-here-ordinance-made-it-possible SFPUC: SF Is a Green Sponge for Stormwater]

That distinction matters.

The 280-million-gallon figure measures water whose behavior has been changed.

The 95-million-gallon figure represents water actually removed from the sewer system.

Both are valuable.

A rain garden that delays a large volume of runoff for several hours may help prevent a sewer from becoming overloaded even if some of that water eventually enters the system.

This is why evaluating green infrastructure purely by gallons can obscure what it is actually accomplishing.

=== One Program Became Three ===

San Francisco's billion-gallon effort is also noteworthy because it is not a single capital project.

It has gradually developed through at least three different mechanisms.

First, '''private development''' must install stormwater controls when projects trigger the Stormwater Management Ordinance.

Second, '''SFPUC and other city agencies''' construct rain gardens, green streets, creek-restoration projects and other infrastructure on public land.

Third, the city provides '''grants''' for green infrastructure on schools, churches, institutional properties and other large sites. SFPUC says its Green Infrastructure Grant Program can now provide as much as $2.5 million for an individual project and has helped bring large privately owned properties into the stormwater network. [https://www.sfpuc.gov/programs/grants/green-infrastructure-grant-program SFPUC Green Infrastructure Grant Program]

The result is an unusual form of infrastructure expansion.

A traditional sewer program builds infrastructure largely through government capital projects.

San Francisco is also using building regulations, redevelopment, schools, street reconstruction, private properties and community projects to gradually alter the hydrology of the city.

Every qualifying redevelopment can become another small piece of the stormwater system.

=== The Early Experiments ===

Before attempting to reach one billion gallons, SFPUC used a series of demonstration projects to learn whether green infrastructure actually performed as expected under San Francisco conditions.

Projects were constructed in different urban watersheds and included rain gardens, permeable pavement, infiltration galleries and other approaches. [https://www.sfpuc.gov/sites/default/files/programs/GI_Holloway-Monitoring_2018-19.pdf SFPUC: Holloway Green Street Monitoring]

The results varied considerably.

The '''Wiggle Neighborhood Green Corridor''' reduced stormwater entering the sewer from its project area by an estimated '''47 percent, or 870,000 gallons''', during the 2015–16 wet-weather season. [https://www.sfpuc.gov/programs/san-franciscos-urban-watersheds/what-green-infrastructure SFPUC: What Is Green Infrastructure?]

The '''Mission & Valencia Green Gateway''' later achieved an estimated '''86 percent reduction—about 1.5 million gallons—in the 2017–18 rainy season'''. [https://www.sfpuc.gov/sites/default/files/programs/GI_MissionValencia-Monitoring_2017-18.pdf SFPUC: Mission & Valencia Green Gateway Monitoring]

A model block of the '''Sunset Boulevard Greenway''' achieved an estimated '''95 percent reduction''' during its first monitored rainy season. [https://www.sfpuc.gov/sites/default/files/programs/GI_SunsetModelBlock-Monitoring_2016-17.pdf SFPUC: Sunset Boulevard Greenway Monitoring]

The '''Holloway Green Street''' project, combining rain gardens and permeable pavement, reduced runoff entering the combined sewer by roughly '''78 percent, or 764,000 gallons''', during 2018–19. [https://www.sfpuc.gov/sites/default/files/programs/GI_Holloway-Monitoring_2018-19.pdf SFPUC: Holloway Green Street Monitoring]

These differences were not failures of the concept.

They became part of the experiment.

They demonstrated that performance depends on soil, drainage area, facility size, storm intensity, construction details and the particular technology being used.

The city was beginning to learn an important lesson:

'''A rain garden is not a standardized appliance. It is an engineered ecological system whose performance depends on the landscape around and beneath it.'''

That lesson leads directly to the second part of the story.


== 2. Rebuilding the Urban Water Cycle ==

The basic problem green infrastructure is trying to solve began long before San Francisco constructed its first rain garden.

Cities change what rain does.

On an undeveloped landscape, rainfall encounters leaves, grasses, soil, roots, wetlands, sand and other permeable surfaces.

Some water evaporates.

Some is intercepted by vegetation.

Some infiltrates into soil.

Some is taken up by plants and returned to the atmosphere through transpiration.

Some slowly moves through soil toward creeks or groundwater.

Only a portion becomes immediate surface runoff.

Pavement reverses those proportions.

A raindrop striking asphalt has few places to go. It joins other drops, flows toward the gutter and can reach a catch basin within minutes.

Urban drainage engineering became extremely good at making that happen.

Green infrastructure deliberately slows the process down.

=== What Actually Happens Inside a Rain Garden ===

A San Francisco rain garden usually receives runoff through an opening in the curb or another inlet.

Water spreads across a shallow depressed planting area rather than immediately disappearing into a pipe.

For a short period, the garden is designed to flood.

Sediment settles.

Water begins passing through mulch and engineered soil.

Roots create pathways through the media.

Some water is held within pore spaces.

Some may eventually be used by plants.

If the underlying native soil is sufficiently permeable, water continues downward into the ground.

When the garden receives more water than it can store or infiltrate, an overflow structure returns the excess to the conventional drainage system.

The '''Sunset Boulevard Greenway''' provides an unusually clear example.

Its rain gardens were designed so runoff passes through approximately '''18 inches of bioretention soil'''. Water infiltrates downward until the system becomes saturated. Only then does excess water overflow back toward the street and sewer. [https://www.sfpuc.gov/sites/default/files/programs/GI_SunsetModelBlock-Monitoring_2016-17.pdf SFPUC: Sunset Boulevard Greenway Monitoring]

The concept sounds simple.

Its real-world behavior is not.

=== San Francisco Has Very Different Soils ===

San Francisco contains sandy former dunes, clay-rich areas, artificial fill, bedrock and heavily modified urban soils.

A rain garden on one side of the city can therefore behave very differently from an apparently identical garden somewhere else.

The Sunset Boulevard monitoring site sits on exceptionally permeable sandy soils.

Field testing found native-soil infiltration sufficiently high that SFPUC used a conservative '''design infiltration rate of 10 inches per hour''' in its hydrological model. [https://www.sfpuc.gov/sites/default/files/programs/GI_SunsetModelBlock-Monitoring_2016-17.pdf SFPUC: Sunset Boulevard Greenway Monitoring]

During the 2016–17 rainy season, researchers monitored 43 storms.

The rain gardens completely absorbed runoff from '''33 of them'''.

Across the entire season, the model block reduced runoff volume to the combined sewer by approximately '''95 percent''' and reduced peak flow by an average of about '''94 percent'''. [https://www.sfpuc.gov/sites/default/files/programs/GI_SunsetModelBlock-Monitoring_2016-17.pdf SFPUC: Sunset Boulevard Greenway Monitoring]

Move across the city to Holloway Avenue and the underground conditions change.

Tests along the Holloway Green Street found design infiltration rates ranging from roughly '''0.10 to 1.22 inches per hour'''—far below the Sunset site. Some blocks could infiltrate runoff directly; others required underdrains to carry water away after it passed through the soil. [https://www.sfpuc.gov/sites/default/files/programs/GI_Holloway-Monitoring_2018-19.pdf SFPUC: Holloway Green Street Monitoring]

Monitoring there showed annual volume reductions ranging from '''56 to 97 percent among individual blocks''', strongly associated with differences in soil infiltration. [https://www.sfpuc.gov/sites/default/files/programs/GI_Holloway-Monitoring_2018-19.pdf SFPUC: Holloway Green Street Monitoring]

That is a useful reminder that "nature-based" infrastructure still requires substantial engineering.

Designers cannot simply dig a hole and add plants.

They must know what lies beneath it.

=== The Shape of a Storm Matters as Much as the Amount of Rain ===

The Sunset monitoring produced another revealing result.

One storm dropped '''1.66 inches of rain over 37 hours'''.

The rain gardens absorbed the entire runoff volume.

Another storm delivered a very similar '''1.69 inches over 27 hours''', but it followed an earlier storm and included periods of intense rainfall.

The soil was already close to saturated.

That time, the gardens reduced runoff volume by about '''78 percent''' rather than 100 percent. [https://www.sfpuc.gov/sites/default/files/programs/GI_SunsetModelBlock-Monitoring_2016-17.pdf SFPUC: Sunset Boulevard Greenway Monitoring]

The total rainfall was nearly identical.

The hydrological outcome was not.

Rain gardens need time.

During a long, gentle storm, water can infiltrate while additional rain continues falling.

During an intense cloudburst, runoff can arrive faster than soil can absorb it.

Once temporary storage is full, excess water begins flowing through the overflow system.

This distinction will become increasingly important if climate change changes not simply how much rain San Francisco receives but '''how intensely that rain falls'''.

Green infrastructure can dramatically reduce runoff without making urban flooding physically impossible.

=== Size Matters Too ===

The '''Cesar Chavez Streetscape Improvement Project''' produced another useful lesson.

Its 18 rain gardens were not all the same size relative to the amount of pavement draining into them.

One relatively undersized planter had green-infrastructure area equal to only about '''0.5 percent of its drainage area'''. Modeling estimated that it reduced seasonal runoff by about '''31 percent'''.

Another planter was much closer to recommended sizing, with green infrastructure equal to about '''3.8 percent of its drainage area'''.

That installation reduced estimated seasonal runoff by approximately '''89 percent'''. [https://www.sfpuc.gov/sites/default/files/programs/GI_CesarChavezStreetScapeReport.pdf SFPUC: Cesar Chavez Streetscape Monitoring Report]

Across all 18 planters, the project was estimated to reduce stormwater entering the combined sewer by about '''53 percent—more than 1.5 million gallons during an average rainfall year'''. [https://www.sfpuc.gov/sites/default/files/programs/GI_CesarChavezStreetScapeReport.pdf SFPUC: Cesar Chavez Streetscape Monitoring Report]

The implication is straightforward:

A small rain garden connected to an enormous area of pavement may still help, but it cannot perform the same job as a properly sized facility.

The important ratio is therefore not simply how many rain gardens a neighborhood has.

It is how much '''impervious drainage area those gardens are capable of managing'''.

=== Many Small Gardens Can Behave Like One Large Facility ===

Mission and Valencia provides yet another variation.

The project contains a series of rain gardens that are '''hydraulically connected''', allowing multiple installations to function in aggregate rather than as completely independent planters.

SFPUC specifically monitored the project to understand whether this distributed chain of facilities could behave like a larger stormwater system.

During 2017–18 it reduced runoff entering the sewer from its project area by approximately '''86 percent, or 1.5 million gallons'''. Of 41 storms producing measurable runoff, 25 were fully managed; the average reduction in peak flow was approximately 90 percent. [https://www.sfpuc.gov/sites/default/files/programs/GI_MissionValencia-Monitoring_2017-18.pdf SFPUC: Mission & Valencia Green Gateway Monitoring]

This points toward one of the larger ideas behind the billion-gallon program.

The unit of analysis does not always have to be one garden.

Several installations can operate together along a street.

Several streets can operate together within a neighborhood.

Neighborhood projects can eventually become part of a watershed-scale system.

=== Slowing Water Can Matter Even When It Is Not Eliminated ===

Complete infiltration attracts attention because it produces an easily understandable result:

The water never entered the sewer.

But delay can also be valuable.

Imagine that an intense storm sends ten units of water toward a sewer pipe during one hour.

If a rain garden temporarily holds four units and releases them several hours later, it may have prevented the pipe from becoming overloaded even though those four units eventually entered the sewer.

This is why engineers also measure '''peak-flow reduction'''.

At Sunset, peak flows across the monitored season fell by approximately 94 percent. At Holloway, monitored peak-flow reductions averaged about 72 percent, although performance varied substantially among blocks. [https://www.sfpuc.gov/sites/default/files/programs/GI_SunsetModelBlock-Monitoring_2016-17.pdf SFPUC: Sunset Boulevard Greenway Monitoring]

The ecological analogy is useful.

Natural watersheds rarely make rainfall disappear.

They '''slow it down'''.

Green infrastructure attempts to restore some of that temporal buffering inside a landscape designed to move water away as rapidly as possible.

That may be one of its most important functions.


== 3. Rain Gardens as Pollution Filters ==

The water flowing down a city street during a rainstorm is not chemically equivalent to the rain that fell from the sky.

Between those two moments, it encounters the city.

Runoff passes across automobile tires, brake dust, motor oil, building materials, pesticides, trash, atmospheric deposition, animal waste, plastics and thousands of other substances associated with urban life.

The first rainfall after a dry period can mobilize material that has accumulated on pavement for days or weeks.

Stormwater therefore becomes a transportation system for urban pollution.

For years, engineers already knew that bioretention could capture sediment and common pollutants such as metals.

The more difficult question was what happens to the growing class of substances researchers call '''contaminants of emerging concern'''—chemicals that may be environmentally important but are incompletely monitored or regulated.

In 2026, San Francisco became the site of an unusually broad field test.

=== Four Rain Gardens, 21 Chemicals and a Year and a Half of Storms ===

Scientists led by the San Francisco Estuary Institute sampled stormwater flowing into and out of '''four San Francisco bioretention systems over approximately a year and a half'''.

They looked for tire-wear particles and 21 chemical contaminants.

The list included:

6PPD-quinone;

1,3-diphenylguanidine;

HMMM;

benzothiazoles;

benzotriazoles;

pesticides;

pharmaceuticals;

personal-care-product chemicals;

and other vehicle-related compounds. [https://www.sfei.org/news/rain-gardens-CECs-study-26 SFEI: San Francisco Rain Gardens Filter Microplastic Tire Particles and Dozens of Chemicals from Stormwater]

Many of these substances were not pollutants the rain gardens had specifically been designed to remove.

That made the results particularly interesting.

Researchers found that concentrations of vehicle-derived contaminants entering the gardens were generally at least '''twice the levels previously reported in regional stormwater''', consistent with the fact that the installations primarily received runoff from streets and parking areas. [https://www.niva.no/publikasjoner/publikasjon?cristinid=10480156 NIVA: Load and Concentration Based Performance Outcomes for Emerging Organic Contaminants]

The gardens were therefore being tested against a realistic and relatively concentrated mixture of urban pollution.

=== A Small Garden Treating a Much Larger Landscape ===

At the Visitacion Valley study location, runoff from an area '''larger than a football field''' drained toward a rain garden whose footprint was roughly comparable to that of a small house.

Despite that difference in scale, the garden reduced every contaminant tested there by at least three-quarters, and most were reduced by '''90 percent or more'''. [https://www.sfei.org/news/rain-gardens-CECs-study-26 SFEI: Rain Gardens and Emerging Contaminants]

This is an important way to understand the physical leverage of green infrastructure.

The ecological footprint does not have to equal the paved footprint.

A strategically placed patch of soil can intercept runoff generated across an area many times larger than itself.

The street effectively delivers its pollution to the garden.

=== Concentration and Load Tell Different Stories ===

The 2026 study also highlights an important distinction in water-quality research.

Suppose one liter of runoff contains 100 units of a chemical.

If treatment reduces that concentration to 20 units per liter, concentration has fallen by 80 percent.

But suppose the rain garden also infiltrates three-quarters of the water.

Far less water now leaves the site carrying the remaining contaminant.

The reduction in '''total pollutant mass''', or load, can therefore be considerably greater than the reduction in concentration alone.

Across the four San Francisco systems, median concentration reductions exceeded '''76 percent for nine of the 19 contaminants''' for which that calculation could be made.

At the site where researchers also modeled stormwater volumes, median reductions in the '''total mass''' of most contaminants exceeded '''90 percent'''. [https://www.niva.no/publikasjoner/publikasjon?cristinid=10480156 NIVA: Load and Concentration Based Performance Outcomes]

Three processes were occurring simultaneously:

the outgoing water could contain a lower concentration of pollution;

the system could release less water;

and solid particles could remain trapped in the soil.

That combination explains why bioretention can sometimes achieve very high load reductions.

=== Soil Is Doing More Than Acting Like a Sieve ===

It is tempting to imagine a rain garden as an ordinary filter in which dirty water enters the top and physical particles get caught between grains of soil.

That certainly happens.

But bioretention is chemically and biologically more complicated.

Particles can become physically strained and trapped.

Dissolved compounds can '''sorb''' onto soil minerals, organic matter and other surfaces.

Plants alter water movement and soil structure.

Roots support microbial communities.

Microorganisms can transform some organic compounds.

Plants themselves may take up certain contaminants.

Evapotranspiration reduces the amount of water leaving the system.

Research reviews describe plants and microbes as active components of bioretention rather than merely decoration around an engineered soil filter. [https://www.sciencedirect.com/science/article/pii/S0301479721011701 ScienceDirect: Bioretention Review]

This explains why the word '''bioretention''' is useful.

The system combines hydrology, soil chemistry and biology.

=== Not Every Pollutant Behaves the Same Way ===

The same complexity creates limitations.

A chemical that binds strongly to organic matter may be retained efficiently.

A highly soluble compound may move through the soil much more readily.

Some pollutants degrade.

Others persist.

Some nutrients can even be released by certain bioretention soils under particular conditions.

A major review of conventional bioretention media concluded that standard soil mixtures often perform well for particulate pollutants but can struggle with some '''dissolved pollutants''', and that some media can themselves leach nutrients or other constituents. Researchers have therefore experimented with additions such as biochar, iron-containing materials and other amendments designed to target particular pollutants. [https://www.sciencedirect.com/science/article/pii/S0043135420311830 ScienceDirect: Bioretention Media and Dissolved Pollutants]

This raises an intriguing possibility for the next generation of urban green infrastructure.

Today, a city might design a rain garden primarily according to the amount of runoff it must manage.

In the future, soil mixtures might increasingly be selected according to '''what is present in that runoff'''.

A garden beside a heavily traveled roadway might be optimized for tire chemicals and metals.

A garden receiving roof runoff might need a different design.

A project near an ecologically sensitive creek could include treatment media selected for contaminants most dangerous to that ecosystem.

Green infrastructure would then move closer to becoming deliberately engineered '''ecological water treatment'''.

=== Where Does the Pollution Go? ===

High removal percentages also create a question that should not be ignored.

If a rain garden removes a contaminant from stormwater, the contaminant has not necessarily ceased to exist.

Some chemicals degrade biologically or chemically.

Others become bound to soil.

Solid particles accumulate.

The 2026 San Francisco study found tire particles and other microplastics retained in the bioretention media, with many of the other microplastics concentrated near the surface. [https://www.niva.no/publikasjoner/publikasjon?cristinid=10480156 NIVA: Load and Concentration Based Performance Outcomes]

That is good news if the alternative is their immediate transport into the Bay.

But it creates a long-term management question:

'''What happens after years or decades of accumulation?'''

At what concentration should soil be replaced?

How should removed material be disposed of?

Could contaminants eventually migrate deeper?

Does treatment efficiency change as a garden ages?

These are precisely the kinds of questions that become important when rain gardens stop being experimental landscaping and become permanent municipal infrastructure.

San Francisco's research is therefore demonstrating two things simultaneously.

Rain gardens can be remarkably effective filters.

And understanding their entire environmental life cycle requires more than measuring the water flowing out of them.


== 4. Microplastics, Tires and the Chemistry of the Street ==

Few pollutants illustrate the hidden chemistry of urban runoff better than automobile tires.

A tire looks solid.

It is not permanent.

Every time a car accelerates, brakes or turns, friction removes microscopic pieces of rubber and road material.

Those particles accumulate along streets.

Then it rains.

=== Trillions of Particles ===

Research by the San Francisco Estuary Institute and partners found microplastic pollution throughout San Francisco Bay.

Stormwater emerged as an unexpectedly important pathway.

Using measurements from Bay Area tributaries and regional modeling, researchers estimated that urban stormwater transports approximately '''7 trillion microplastic particles into San Francisco Bay each year'''.

That estimated load was roughly '''300 times greater than the load from wastewater-treatment-plant discharges''' in the pathways studied. Nearly half of the stormwater particles were black rubber-like fragments associated with tire wear. [https://www.sfei.org/projects/san-francisco-bay-microplastics-project SFEI: San Francisco Bay Microplastics Project]

Later research suggests the original estimate may even have undercounted smaller tire-wear particles because early sampling methods could not capture the smallest size fractions. [https://www.sfei.org/sites/default/files/events/20240613%20Bay%20RMP%20TRC%20Agenda%20Package%20Final.pdf SFEI Regional Monitoring Program Technical Review Committee Materials]

SFEI estimates that vehicles on Bay Area roads collectively generate approximately '''15,000 to 18,000 metric tons of tire-wear particles every year'''. [https://www.sfei.org/programs/areas/tons-tire-particles SFEI: Tons of Tire Particles]

Most people never see this pollution.

Unlike a discarded plastic bottle, tire wear disperses gradually.

The pollution is generated by the ordinary act of driving.

=== Can Soil Stop the Particles? ===

This is one place where rain gardens appear particularly effective.

The 2026 San Francisco study found the monitored bioretention systems could remove '''up to 99 percent of tire-wear particles''' from incoming stormwater. [https://www.sfei.org/news/rain-gardens-CECs-study-26 SFEI: Rain Gardens and Emerging Contaminants]

Soil cores confirmed that tire particles and other microplastics were being retained within the treatment media rather than simply disappearing from the measurements. Other microplastics were found particularly in the upper soil layers. [https://www.niva.no/publikasjoner/publikasjon?cristinid=10480156 NIVA: Load and Concentration Based Performance Outcomes]

That result suggests a surprisingly simple intervention between the roadway and the Bay:

'''make polluted street water pass through soil before it reaches the receiving environment.'''

But tires pose another problem that is not visible even under a microscope.

They contain chemicals.

=== The Mystery of the Dying Salmon ===

For years, researchers in the Pacific Northwest had observed a strange phenomenon.

Adult coho salmon returning to spawn in urban streams sometimes died rapidly following storms.

The pattern was so consistent that scientists referred to it as urban runoff mortality syndrome.

In 2020, researchers identified a major culprit: '''6PPD-quinone''', usually abbreviated 6PPD-q or 6PPDQ.

The story begins with '''6PPD''', a chemical manufacturers add to tires to protect rubber from ozone and extend tire life.

That function is important for vehicle safety.

But when 6PPD reacts with ozone in the environment, it transforms into 6PPD-quinone.

The transformation product turned out to be extraordinarily toxic to coho salmon and harmful to some other aquatic organisms. [https://dtsc.ca.gov/2023/07/26/news-release_t-06-23/ California DTSC: 6PPD in Motor Vehicle Tires]

This was not a pollutant intentionally poured into a river.

It was a chemical created indirectly from an additive in an everyday consumer product after particles from that product entered the environment.

That makes 6PPD-q a useful example of why scientists use the term '''emerging contaminant'''.

The pollutant existed before researchers understood the risk.

=== The Problem Has Reached San Francisco Bay ===

Coho salmon have been absent from San Francisco Bay for decades.

That does not make the contaminant irrelevant locally.

SFEI analyzed Bay samples collected from '''2021 through 2024''' and detected 6PPD-q in the estuary. Concentrations were highest near shore after winter storms—the same season when migratory fish are moving between ocean, Bay and freshwater streams. [https://www.sfei.org/news/new-data-salmon-killing-chemical-san-francisco-bay SFEI: New Data on Salmon-Killing Chemical in San Francisco Bay]

Threatened steelhead trout use the Bay and surrounding watersheds and are also sensitive to 6PPD-q, although sensitivity varies substantially among salmonid species.

SFEI reports that concentrations measured in Bay water itself were not expected to cause acute mortality in steelhead, but researchers still know relatively little about possible chronic or sublethal exposure in an estuarine environment. [https://www.sfei.org/news/new-data-salmon-killing-chemical-san-francisco-bay SFEI: New Data on Salmon-Killing Chemical in San Francisco Bay]

This is an important distinction.

Detection does not automatically equal ecological catastrophe.

But it demonstrates that a chemical shed from tires on roads has moved through the urban watershed and into one of North America's largest estuaries.

=== A 95 Percent Reduction That May Still Not Be Enough ===

The San Francisco rain-garden study produced one of its most striking findings for 6PPD-q.

Bioretention reduced its concentration by approximately '''95 percent'''.

Yet the '''median concentration leaving the studied systems was about 12 nanograms per liter'''. [https://www.niva.no/publikasjoner/publikasjon?cristinid=10480156 NIVA: Load and Concentration Based Performance Outcomes]

For comparison, EPA's current acute freshwater screening value is '''11 nanograms per liter'''.

EPA describes that value as a concentration expected to be generally protective of freshwater aquatic communities, including sensitive salmonids, during short-term exposure. [https://www.epa.gov/system/files/documents/2024-05/6ppd-q-screening-value-2024.pdf EPA: 6PPD-Quinone Freshwater Screening Value]

The comparison requires an important qualification.

EPA has '''not yet established equivalent acute or chronic screening values for estuarine or marine water''' because the necessary toxicity data are insufficient. The 11-ng/L figure is specifically a freshwater value. [https://www.epa.gov/system/files/documents/2024-05/6ppd-q-screening-value-2024.pdf EPA: 6PPD-Quinone Freshwater Screening Value]

So the San Francisco outlet concentration should not be interpreted as proof that water leaving the rain gardens is toxic to organisms in San Francisco Bay.

It demonstrates something subtler and perhaps more important.

A treatment system can remove '''95 percent of a pollutant''' and still leave a concentration high enough to concern scientists.

Removal efficiency and ecological safety are not the same measurement.

=== Treatment Downstream, Prevention Upstream ===

This finding points to a fundamental limit of green infrastructure.

Suppose a pollutant enters stormwater at a concentration 20 times higher than a biologically protective level.

A rain garden removing 90 percent might still leave twice as much as the desired concentration.

Making the filter somewhat better may help.

But eventually it becomes more effective to reduce the amount entering the environment in the first place.

Environmental scientists call that '''source control'''.

In the case of 6PPD, source control means changing the tire itself.

California has already begun that process.

Since '''October 1, 2023''', the California Department of Toxic Substances Control has regulated motor-vehicle tires containing 6PPD as a Priority Product under its Safer Consumer Products program.

The regulation does not currently ban 6PPD.

Instead, tire manufacturers must investigate alternatives capable of maintaining tire safety while reducing environmental harm. Manufacturers have selected '''21 potential alternative antidegradants''' for further analysis, and final alternatives analyses are currently due in 2029. [https://dtsc.ca.gov/scp/motor_vehicle_tires_containing_6ppd/ California DTSC: Motor Vehicle Tires Containing 6PPD]

This distinction matters because replacing 6PPD carelessly could create another hazardous chemical with similar properties. California is therefore requiring manufacturers to evaluate the potential tradeoffs of replacements rather than simply substituting another related compound. [https://dtsc.ca.gov/scp/motor_vehicle_tires_containing_6ppd/ California DTSC: Motor Vehicle Tires Containing 6PPD]

=== Rain Gardens as the Second Line of Defense ===

The emerging tire-pollution story suggests a useful hierarchy.

The best outcome is to prevent a dangerous chemical from being released.

But source control is rarely immediate.

Millions of vehicles already travel on existing tires.

Other contaminants will continue accumulating on streets.

Storms will continue carrying them toward waterways.

Rain gardens therefore occupy an important middle position between '''pollution prevention and downstream cleanup'''.

They cannot eliminate the source.

They can interrupt the pathway.

That distinction may ultimately be one of the strongest arguments for urban green infrastructure.

A city cannot control every chemical used in every automobile, building material or consumer product.

It can influence what happens when those chemicals wash across its streets.

Instead of allowing that mixture to move immediately toward a pipe or waterway, the city can place soil, plants and biological processes in its path.

San Francisco's rain gardens are showing that this relatively modest intervention can remove surprisingly large quantities of particles and chemicals.

But the 6PPD-q results also establish the boundary of the approach.

'''A rain garden can be an extraordinarily effective filter without being permission to keep polluting upstream.'''

That makes green infrastructure not an alternative to pollution prevention, but a second line of environmental defense.


== 5. Creating Habitat in the Streets ==

The ecological importance of a rain garden begins with a simple fact: unlike a storm drain, it is alive.

A conventional drain provides almost no habitat. A planted bioretention area contains soil, roots, flowers, insects, moisture, organic matter and sometimes shrubs or trees. Even a relatively small installation can therefore become a patch of habitat within a landscape otherwise dominated by buildings and pavement.

But the distinction between '''green infrastructure''' and '''ecologically valuable green infrastructure''' is important.

Putting plants around a drainage structure does not automatically create significant wildlife habitat. The species planted, the size and shape of the garden, how often it is disturbed, the availability of water, its relationship to nearby habitat and the way it is maintained all influence what organisms can actually use it.

San Francisco has increasingly begun to recognize this distinction.

The city's Biodiversity Guidelines state that projects throughout San Francisco should be viewed as opportunities for creating wildlife habitat and improving ecological connectivity. They specifically encourage the intentional use of native and pollinator-friendly plants to benefit birds, bees, butterflies, bats and other wildlife. [https://www.sfenvironment.org/city-san-francisco-biodiversity-guidelines San Francisco Biodiversity Guidelines]

That creates the possibility of thinking about rain gardens not as isolated landscaping projects but as pieces of a larger ecological network.

=== Visitacion Valley: A Rain Garden Becomes Part of a Corridor ===

One of the clearest examples is the '''Visitacion Valley Green Nodes project'''.

Along Leland and Sunnydale avenues, SFPUC constructed several rain gardens that combine stormwater management with pedestrian improvements, community space and vegetation.

The tiered gardens near McLaren Park capture, treat and absorb about '''600,000 gallons of stormwater annually from 1.5 acres of impervious surface'''. A nearby plaza manages another '''180,000 gallons annually'''. [https://www.sfpuc.gov/sites/default/files/construction-and-contracts/Visitacion%20Valley%20Green%20Nodea%20Walking%20Tour%20Map.pdf SFPUC: Visitacion Valley Green Nodes Walking Tour]

Those numbers alone make them functioning pieces of sewer infrastructure.

But their location adds another dimension.

SFPUC planted the gardens with native and drought-tolerant vegetation and describes them as improving natural habitat. The project also lies along San Francisco's '''Green Connections''' network, a system of greener walking and cycling routes intended to improve connections among parks, natural areas and the waterfront. At Visitacion Valley, the route helps connect the larger landscape containing Lake Merced, McLaren Park and Candlestick Point. [https://www.sfpuc.gov/sites/default/files/construction-and-contracts/Visitacion%20Valley%20Green%20Nodea%20Walking%20Tour%20Map.pdf SFPUC: Visitacion Valley Green Nodes Walking Tour]

The distinction is subtle but important.

A single rain garden may provide only a few thousand square feet of vegetation. Its ecological importance may increase when it sits between other vegetated spaces.

For a person, the distance between two parks may be a short walk.

For a bee, butterfly or small bird moving through streets and buildings, intermediate patches of vegetation can matter considerably more.

Ecologists often call such patches '''stepping-stone habitat'''.

Research outside San Francisco has demonstrated that this idea is more than theoretical. A study of bioretention basins planted with native wildflowers found pollinators moving between installations separated by distances of up to 86 meters, suggesting that disconnected rain gardens can sometimes function collectively as a habitat network. [https://www.sciencedirect.com/science/article/pii/S1618866723003047 Urban Forestry & Urban Greening: Multifunctional Bioretention Basins and Pollinators]

San Francisco already has another local example of this principle. Its '''Green Hairstreak Corridor''', although not a stormwater project, uses a series of small planted habitat patches to connect populations of the native green hairstreak butterfly in the city's western neighborhoods. San Francisco Environment now maps hundreds of pollinator and native-plant gardens throughout the city. [https://www.sfenvironment.org/pollinator-and-native-plant-gardens-san-francisco San Francisco Environment: Pollinator and Native Plant Gardens]

Rain gardens could potentially become another layer in that network.

=== Bringing a Buried Creek Back to the Surface ===

An even more ambitious experiment is now taking shape in McLaren Park.

For generations, much of San Francisco treated urban creeks the same way cities treated stormwater: put them underground.

The '''Upper Yosemite Creek Daylighting Project''' partially reverses that process.

Yosemite Creek historically flowed through what is now McLaren Park toward San Francisco Bay. The city plans to bring approximately '''1,700 feet of the creek back to the surface''', with an additional underground segment producing a new creek system approximately 2,100 feet long. [https://sfrecpark.org/1614/McLaren-Park---Yosemite-Creek-Daylightin San Francisco Recreation and Parks: Yosemite Creek Daylighting]

Instead of immediately moving stormwater underground, the project will use an open seasonal creek, rain gardens, bioretention areas and underground storage.

It is designed to manage runoff from approximately '''106 acres of McLaren Park''', including flows from Yosemite Marsh and McNab Lake. [https://www.sfpuc.gov/construction-contracts/construction-projects/upper-yosemite SFPUC: Upper Yosemite Creek Daylighting]

But the city is not designing the project solely for water.

The creek corridor will use California native plants, include '''87 new trees''', and create habitat intended for birds, butterflies, insects and other wildlife. During the rainy season the channel will carry water. During dry periods, vegetation around the creek will remain as habitat. [https://www.sfpuc.gov/construction-contracts/construction-projects/upper-yosemite SFPUC: Upper Yosemite Creek Daylighting]

This is a very different conception of drainage infrastructure.

Instead of:

'''park → storm drain → pipe'''

the system becomes something closer to:

'''park → marsh → creek → vegetation → soil → storage → reuse or sewer'''

Some of the captured water will also be stored beneath the Louis Sutter athletic field and used for irrigation, reducing potable-water demand. [https://www.sfpuc.gov/construction-contracts/construction-projects/upper-yosemite SFPUC: Upper Yosemite Creek Daylighting]

The creek is therefore being asked to perform several functions simultaneously:

manage stormwater;

reduce pressure on the combined sewer;

provide habitat;

support native vegetation;

irrigate a playing field;

create public open space;

and teach visitors that a watershed still exists beneath the city.

Construction is expected to proceed during the current 2026–2028 period. [https://sfrecpark.org/1614/McLaren-Park---Yosemite-Creek-Daylightin San Francisco Recreation and Parks: Yosemite Creek Daylighting]

=== Reassembling an Urban Watershed ===

Yosemite Creek becomes even more interesting when viewed alongside other ecological projects occurring around McLaren Park.

Near '''Grey Fox Creek''', Recreation and Parks has been removing invasive vegetation and restoring riparian habitat. Social trails through sensitive areas are being closed to reduce trampling and soil compaction.

Near the Upper Reservoir, the city has reshaped drainage to create a '''seasonal wetland'''. The goal is simultaneously to slow and absorb stormwater, feed Yosemite Creek and improve wildlife habitat. Native plants are being installed around the restored areas. [https://www.sfrecpark.org/631/McLaren-Park---Trails-Improvement-Projec San Francisco Recreation and Parks: McLaren Park Trails Improvement Project]

These projects begin to resemble something larger than a collection of landscaping improvements.

A wetland feeds a creek.

The creek connects to rain gardens.

Rain gardens connect to planted streets.

Planted streets connect neighborhoods with parks and other open spaces.

The city is not reconstructing the original watershed exactly as it existed before development. Buildings and roads make that impossible.

Instead, it is attempting to '''reassemble some of the functions of a watershed inside the existing city'''.

=== From Individual Projects to Ecological Planning ===

The San Francisco Estuary Institute's Next Generation Urban Greening program takes this concept another step.

Rather than asking where one rain garden could fit, researchers have been examining the '''Islais Creek watershed as a whole'''.

Their planning model first identifies places where green infrastructure can reduce runoff and pollutants such as PCBs. Researchers then evaluate those same locations for additional benefits including biodiversity, increased tree-canopy access and urban heat reduction. [https://www.sfei.org/documents/watershed-plan-case-study-approach-incorporating-multiple-benefits-green-stormwater SFEI: Watershed Plan]

The objective is to find locations where a single investment can accomplish several things at once.

A location that is only moderately useful for stormwater capture might become highly valuable if it also fills a gap between habitat patches, shades an overheated block and serves residents with little existing access to greenery.

SFEI describes this approach as '''multi-objective, watershed-scale planning'''. [https://www.sfei.org/projects/next-generation-urban-greening SFEI: Next Generation Urban Greening]

That could represent an important evolution in how cities build green infrastructure.

The first generation asked:

'''Where can we put a rain garden?'''

The second asked:

'''Where will a rain garden capture the most stormwater?'''

The next generation can ask:

'''Where will a rain garden capture stormwater, remove pollution, cool the city and strengthen an ecological network at the same time?'''

=== Habitat Is a Claim That Can Be Tested ===

There is an important caution.

A project designed to provide wildlife habitat does not necessarily succeed in doing so.

Native plants may die. Aggressive species can dominate. Small gardens can remain too isolated. Heavy pedestrian disturbance can reduce wildlife use. Maintenance designed primarily to keep a streetscape neat can remove flowers, seeds, dead wood and other features useful to animals.

Research on birds around bioretention systems has found ecological differences from ordinary lawns but also cautions that biodiversity improvements can be modest. Simply calling something green infrastructure is not evidence that it has become a functioning ecosystem. [https://pubmed.ncbi.nlm.nih.gov/39382737/ PubMed: Urban Bird Community and Bioretention Study]

That suggests another logical step for San Francisco.

The city already measures '''gallons of stormwater'''.

Researchers are measuring '''chemical contaminants'''.

Future monitoring could increasingly measure:

pollinator abundance;

bird use;

native plant survival;

soil organisms;

vegetation structure;

habitat connectivity;

and changes in species diversity over time.

If San Francisco wants rain gardens to become ecological infrastructure, biodiversity can ultimately be measured just as stormwater is.

The central question is no longer whether plants can be added to the sewer system.

It is whether hundreds of living pieces of infrastructure can begin functioning together as an urban ecosystem.


== 6. Cooling Streets, Improving Neighborhoods—and Who Gets the Benefits? ==

The ecological benefits of rain gardens do not disappear when the storm ends.

Vegetation intercepts sunlight. Soil stores water. Plants release moisture through evapotranspiration. Trees create shade. Replacing dark pavement with planted surfaces can change how much heat an urban block absorbs and retains.

These effects become increasingly important as San Francisco confronts a climate for which much of the city was never designed.

San Francisco's reputation for cool summers can disguise its vulnerability to extreme heat. Many homes lack air conditioning, and neighborhoods differ dramatically in the amount of vegetation available to moderate temperature.

Those differences are striking.

=== A City With Very Unequal Canopy ===

A 2022 tree-canopy analysis produced for the USDA Forest Service and CAL FIRE and released in 2025 estimated tree canopy at about '''12.8 percent citywide'''.

But that citywide number hides enormous differences.

The analysis estimated canopy at:

'''2.6 percent in the Tenderloin;'''

'''2.7 percent in South of Market;'''

'''2.8 percent in Chinatown;'''

'''4.2 percent in Bayview Hunters Point;'''

and '''4.7 percent in Visitacion Valley.'''

By contrast, the figure for McLaren Park itself was about '''40.9 percent'''. [https://www.sfenvironment.org/files/2025-10/From%20Concrete%20To%20Canopy_%20A%20Data-Driven%20Path%20to%20Urban%20Forest%20Equity.pdf San Francisco Environment: From Concrete to Canopy]

The relevance to rain gardens is not that every bioretention planter can grow a large tree. Many cannot.

It is that green-stormwater infrastructure creates opportunities to replace portions of impervious streetscape with living surfaces and, where conditions allow, integrate trees and larger vegetation.

Those opportunities have greater value in some places than in others.

A new tree planted beside an already heavily wooded park and one planted on an almost treeless block do not necessarily provide the same marginal public benefit.

The same is true of a rain garden.

=== Heat Vulnerability Follows Geography ===

San Francisco public-health data identify '''Chinatown, the Tenderloin, South of Market, Japantown and the Mission''' among the neighborhoods with the greatest vulnerability to extreme heat.

South of Market is also particularly affected by air pollution, while South of Market, Bayview and the Tenderloin appear among neighborhoods with elevated vulnerability to flooding. [https://sfhip.org/chna/community-health-data/climate-and-the-natural-environment/ SFHIP: Climate and the Natural Environment]

This creates precisely the sort of overlap that multifunctional infrastructure is designed to address.

A conventional stormwater analysis might prioritize a location because water collects there.

A public-health analysis might prioritize another because residents face high heat exposure.

A biodiversity analysis might identify a third because vegetation is sparse.

An equity analysis might identify communities carrying unusually high cumulative environmental burdens.

Increasingly, San Francisco is trying to put those maps on top of one another.

=== San Francisco Has Now Mapped Where Green Infrastructure Could Matter Most ===

The city's '''Heat and Air Quality Resilience Green Infrastructure Priority Zones''' represent a particularly important development.

Rather than relying on a single measure, San Francisco Public Works and partner agencies combined:

satellite temperature measurements from a heat wave;

fine-particle air-pollution exposure;

existing tree canopy;

asthma hospitalizations;

and diabetes hospitalizations.

The resulting maps identify locations where green infrastructure could provide particularly large public-health benefits. [https://data.sfgov.org/Energy-and-Environment/HAQR-Priority-Green-Infrastructure-Zones/nn26-kuy2 San Francisco Open Data: HAQR Priority Green Infrastructure Zones]

That is a significant change in the logic of infrastructure planning.

Instead of deciding where to plant first and describing the benefits afterward, the city can identify where the need is greatest '''before choosing the project location'''.

The city's Environmental Justice Framework points in the same direction.

San Francisco defines its Environmental Justice Communities as areas facing the top one-third of combined environmental and socioeconomic burdens. They are concentrated substantially along the city's southern and eastern sides and include parts of Bayview Hunters Point, Visitacion Valley, the Mission, South of Market, Tenderloin, Potrero Hill, Chinatown and other neighborhoods. [https://generalplan.sfplanning.org/Environmental_Justice_Framework.htm San Francisco General Plan: Environmental Justice Framework]

The General Plan goes further than simply identifying the problem.

Its environmental-justice policies specifically call for expanding '''nature-based solutions, green infrastructure and urban greening''', developing neighborhood-specific targets for tree canopy, watersheds and biodiversity, and aligning those investments with areas of high climate vulnerability. [https://generalplan.sfplanning.org/Environmental_Justice_Framework.htm San Francisco General Plan: Environmental Justice Framework]

In other words, equity is no longer merely a possible side benefit of San Francisco's green-infrastructure program.

It has become an explicit planning objective.

=== Southeast San Francisco as a Test ===

Some of the most interesting current experiments are therefore occurring in southeastern San Francisco.

The Next Generation Urban Greening initiative was intentionally centered in part on the Yosemite and Sunnydale watersheds and other southeastern neighborhoods. Its projects include Yosemite Creek, McLaren Park, Visitacion Valley, Crocker Amazon and planning connected with India Basin and Candlestick Point. [https://www.sfei.org/projects/next-generation-urban-greening SFEI: Next Generation Urban Greening]

This geography matters.

Many southeastern neighborhoods combine substantial impervious surfaces, historic industrial activity, transportation pollution, limited tree canopy and populations identified by the city as carrying disproportionate environmental burdens.

Visitacion Valley illustrates how the pieces can overlap.

Its rain gardens manage stormwater, incorporate native vegetation, provide pedestrian improvements and form part of a Green Connections route. The neighborhood simultaneously appears on San Francisco's Environmental Justice Communities map and has considerably less tree canopy than the citywide average. [https://www.sfpuc.gov/sites/default/files/construction-and-contracts/Visitacion%20Valley%20Green%20Nodea%20Walking%20Tour%20Map.pdf SFPUC: Visitacion Valley Green Nodes]

That makes the project more than a drainage intervention.

It is a small example of what '''environmental infrastructure targeted by cumulative need''' might look like.

But individual examples do not answer the larger equity question.

=== Are the 575 Projects Actually Distributed Equitably? ===

San Francisco now has enough information to ask a much harder question.

SFPUC can map hundreds of green-infrastructure projects.

Planning can map Environmental Justice Communities.

Public Works can map heat and air-quality priority zones.

Public-health agencies can map heat vulnerability.

Tree-canopy data can identify neighborhoods with the least vegetation.

Flooding and impervious surfaces can also be mapped.

Putting those datasets together could show whether San Francisco's green infrastructure is actually being distributed according to need.

The relevant question is not merely:

'''Does every neighborhood have a rain garden?'''

It is:

'''Are the greatest investments occurring where the combined environmental benefits are greatest?'''

That analysis should ideally measure more than the number of projects.

A large project managing several million gallons should not necessarily count the same as a tiny planter.

A meaningful citywide equity accounting could compare:

green-infrastructure spending per neighborhood;

acres of impervious surface treated;

stormwater volume managed;

tree canopy added;

existing canopy deficiency;

heat vulnerability;

flood exposure;

pollution burden;

access to parks;

and population served.

Such an analysis would allow the billion-gallon program to be evaluated not only for '''efficiency''', but for '''distributional fairness'''.

=== The Maintenance Question Is Also an Equity Question ===

Green infrastructure has another characteristic that distinguishes it from an underground pipe.

People see it.

And people often help take care of it.

San Francisco's '''Rain Guardians''' program allows residents to adopt individual rain gardens and help keep curb openings clear of leaves, trash and debris. In one part of the Mission, residents have adopted all 11 rain gardens belonging to the Mission & Valencia Green Gateway project. [https://www.sfpuc.gov/about-us/news/meet-rain-guardians-who-adopted-mission-green-infrastructure SFPUC: Meet the Rain Guardians]

Community stewardship can be a genuine advantage. Residents become familiar with how local infrastructure works and may develop a sense of ownership over public space.

But relying excessively on voluntary stewardship introduces another equity issue.

Some neighborhoods have more residents with spare time, strong neighborhood organizations and experience navigating city government. Others do not.

Academic researchers studying green-stormwater infrastructure have cautioned that maintenance requirements and cost-sharing programs can unintentionally favor communities with greater financial or institutional capacity. [https://pubs.acs.org/doi/10.1021/acs.est.3c07062 Environmental Science & Technology: Green Stormwater Infrastructure and Equity]

A system designed for environmental equity therefore cannot assume that communities receiving green infrastructure should supply unpaid labor to keep it functioning.

The city ultimately remains responsible for infrastructure performance.

=== The Green-Gentrification Problem ===

There is an even more difficult question.

What happens if green infrastructure succeeds?

Greener streets can be more attractive. Trees provide shade. Traffic-calming improvements can make walking safer. Cleaner and more pleasant public spaces can make neighborhoods more desirable.

Those are precisely the outcomes cities want.

But in expensive housing markets, environmental improvements can sometimes increase property values and contribute to pressures already affecting lower-income residents.

Researchers call this phenomenon '''green gentrification'''.

Research in Washington, D.C., found an association between concentrations of installed stormwater-control measures and subsequent demographic displacement, although such observational research cannot establish a simple one-step causal relationship. [https://pmc.ncbi.nlm.nih.gov/articles/PMC8508036/ PMC: Green Stormwater Infrastructure and Displacement]

A Minneapolis study likewise found disproportionately high green-stormwater-infrastructure investment in gentrifying areas and faster increases in rents near some installations. [https://www.tandfonline.com/doi/abs/10.1080/1523908X.2021.1945917 Journal of Environmental Policy & Planning: Green Stormwater Infrastructure and Gentrification]

These studies are '''not evidence that San Francisco's rain gardens have displaced residents'''.

We should be careful not to make that claim without local evidence.

They demonstrate instead that environmental justice has two sides.

Historically underserved communities should not be denied greenery because improvements might increase property values.

But environmental investment should ideally be accompanied by policies that help existing residents remain long enough to enjoy the improvements.

The objective should be:

'''green neighborhoods without displacement.'''

=== Measuring More Than Gallons ===

This returns us to the billion-gallon goal.

One billion gallons is an easy metric to understand.

It is also incomplete.

If San Francisco reaches that target but concentrates the benefits in already green and affluent areas, something important would have been missed.

If instead the same investment reduces sewer flows while adding shade to overheated blocks, filtering pollution near vulnerable waterways, connecting habitat and increasing access to vegetation in neighborhoods with the least greenery, the ecological and social return could be much larger.

The future scoreboard for green infrastructure might therefore need more than one number.

Instead of measuring only:

'''Gallons of stormwater managed'''

San Francisco could increasingly measure:

'''Gallons managed'''

'''Pollutants removed'''

'''Impervious surface converted'''

'''Tree canopy created'''

'''Habitat connected'''

'''Heat exposure reduced'''

'''Environmental-justice communities served'''

'''Maintenance resources provided'''

and perhaps, eventually,

'''Whether the people who lived in those neighborhoods before the improvements were still there to benefit from them afterward.'''

That would transform green infrastructure from a stormwater program into something much more ambitious:

'''a method for deciding where ecological restoration can produce the greatest public benefit inside the city.'''


== 7. Can Cities Rebuild Nature One Block at a Time? ==

At first glance, calling a rain garden a form of ecological restoration may seem excessive.

San Francisco's streets are still streets.

Buildings remain where dunes, grasslands, wetlands and creeks once existed. Most buried waterways will remain buried. A rectangular planter receiving runoff from an asphalt roadway is obviously not equivalent to the ecosystem that occupied the site before urbanization.

That distinction matters.

The '''Society for Ecological Restoration''', in the third edition of its international standards published in 2026, defines ecological restoration more narrowly around recovery of native ecosystems. Restoration is evaluated against characteristics of a reference ecosystem and includes such attributes as species composition, ecosystem structure, physical conditions, ecological functions, connectivity and the reduction of threats. The standards also recognize a broader '''restorative continuum''' in which projects can recover some ecological attributes even when complete restoration is neither possible nor intended. [https://www.ser.org/page/serstandards/international-standards-for-the-practice-of-ecological-restoration.htm Society for Ecological Restoration: International Standards]

That continuum provides a useful way to understand San Francisco.

A rain garden at a street corner is generally not restoring the historical ecosystem that once occupied that exact location.

But it may restore '''some of the processes that ecosystem once performed'''.

That is a considerably more defensible—and perhaps more interesting—claim.

=== Restoring Functions Without Recreating the Past ===

Consider what happened when San Francisco urbanized its landscape.

Soil was covered.

Creeks were confined or buried.

Wetlands disappeared.

Vegetation was removed.

Rainfall that once moved slowly through plants and soil was redirected into gutters and pipes.

Habitats became fragmented.

Pollution accumulated on impervious surfaces.

The physical landscape changed so thoroughly that recreating its former condition across most of the city is impossible.

But ecological functions are not necessarily all-or-nothing.

A city cannot remove a neighborhood to reconstruct the original watershed beneath it.

It can nevertheless ask whether some water can encounter soil again before entering a pipe.

It can ask whether vegetation can intercept rainfall.

Whether roots can create infiltration pathways.

Whether microorganisms can process pollutants.

Whether native flowers can feed insects.

Whether planted areas can help connect larger habitat patches.

Whether trees can cool overheated streets.

Whether a buried creek can be brought to the surface where sufficient space remains.

These are not replicas of the original ecosystem.

They are pieces of ecological function inserted back into an altered landscape.

San Francisco's own Biodiversity Guidelines increasingly embrace this idea. They instruct municipal projects to treat the built environment as an opportunity for wildlife habitat and ecological connectivity and explicitly describe development and landscaping projects as opportunities to restore local nature. [https://www.sfenvironment.org/city-san-francisco-biodiversity-guidelines San Francisco Biodiversity Guidelines]

The distinction can therefore be stated simply:

'''San Francisco is unlikely to restore the original landscape across most of the city. It may be able to restore portions of what that landscape used to do.'''

=== A Rain Garden Is More Interesting as Part of a Network ===

The ecological significance of the program also changes depending on the scale at which it is viewed.

One rain garden is easy to dismiss.

It may occupy only a few hundred square feet.

Its contribution to citywide biodiversity is tiny.

The amount of water it manages may represent an insignificant fraction of the city's annual rainfall.

But San Francisco is no longer dealing with one rain garden.

It has built or initiated hundreds of green-infrastructure projects.

Private developments continue adding others.

Schools and institutions are participating.

Green streets are being connected with parks.

Wetlands are being restored.

Yosemite Creek is being daylighted.

Native-plant gardens and habitat corridors are appearing in neighborhoods.

The relevant ecological unit may therefore eventually become not the '''individual planter''', but the '''network'''.

This is precisely the direction taken by the San Francisco Estuary Institute's Next Generation Urban Greening work. Instead of evaluating projects solely as isolated installations, the program considers green infrastructure at watershed scale and attempts to optimize several benefits simultaneously: runoff reduction, pollution removal, biodiversity, habitat connectivity, heat resilience and community benefit. [https://www.sfei.org/projects/next-generation-urban-greening SFEI: Next Generation Urban Greening]

Its 2025 '''Regional Framework for the Next Generation of Urban Greening''' goes further, organizing planning across landscape, individual-site and regional scales. The underlying premise is that the location and design of individual projects should be influenced by the larger ecological and social system in which they sit. [https://www.sfei.org/documents/regional-framework-next-generation-urban-greening SFEI: Regional Framework for the Next Generation of Urban Greening]

That changes the basic planning question.

Instead of:

'''Where can we fit another rain garden?'''

the question becomes:

'''Where will the next rain garden strengthen the larger system?'''

A planter could be located where it intercepts a particularly polluted roadway.

Another might fill a gap between habitat areas.

Another might be placed on a block with little vegetation and high heat vulnerability.

Another might reduce runoff entering a sewer bottleneck.

Some locations might accomplish several of those things simultaneously.

The result is not simply a collection of projects.

It begins to resemble an '''urban ecological network assembled one intervention at a time'''.

=== Ecology Can Improve the Infrastructure Itself ===

There is another reason the distinction between "infrastructure" and "nature" becomes increasingly difficult to maintain.

The living components of green infrastructure do not merely receive benefits from the engineering system.

They can help make the engineering system work.

A 2024 review in ''npj Urban Sustainability'' found growing evidence that biodiversity within blue-green infrastructure can influence stormwater retention, water purification, cooling and resilience. Different plant forms create different root structures, water-use patterns and responses to drought and flooding. Functionally diverse vegetation can therefore sometimes perform multiple engineering tasks better than biologically simple plantings. [https://www.nature.com/articles/s42949-024-00163-y npj Urban Sustainability: Engineering Blue-Green Infrastructure for Biodiversity]

The implication is significant.

Engineers traditionally design infrastructure and then try to keep biological growth from interfering with it.

A pipe works best when roots, animals and sediment stay out.

A rain garden can operate according to almost the opposite principle.

Roots can maintain pathways for water.

Plants remove moisture from soil through transpiration.

Soil organisms alter pore structure.

Microbial communities participate in nutrient and contaminant transformations.

Invertebrates can change soil structure.

Vegetation itself evolves as plants establish, die, reproduce and compete.

The biology is not merely decoration surrounding the infrastructure.

'''The biology is partly the infrastructure.'''

That creates an unusual engineering challenge.

A concrete pipe is generally expected to remain physically similar from one year to the next.

A rain garden is expected to change.

=== Infrastructure That Gets Older—and Becomes More Alive ===

One of the persistent concerns about bioretention has been that sediment will gradually clog soil pores until infiltration declines and the system stops functioning properly.

Laboratory experiments have sometimes reinforced that fear.

Longer-term field evidence is more complicated.

A 2026 review in the ''Journal of Environmental Management'' synthesized more than '''150 studies''' of bioretention performance. Among field systems roughly five to 22 years old, most well-designed and adequately maintained installations continued to meet or exceed typical infiltration thresholds. [https://www.sciencedirect.com/science/article/pii/S0301479726013496 Journal of Environmental Management: Long-Term Bioretention Performance Review]

The researchers proposed thinking about rain gardens in life stages: establishment during roughly the first three years, maturation over approximately years three through ten, and a later period of relative maturity.

Most interestingly, they suggest that biological processes may help explain why real rain gardens sometimes age better than laboratory columns predict.

Roots grow through soil.

Earthworms and other organisms create channels.

Wet and dry cycles restructure soil.

Organic material accumulates.

New soil develops.

These processes can partially counter the physical clogging produced by accumulated sediment. [https://www.sciencedirect.com/science/article/pii/S0301479726013496 Journal of Environmental Management: Long-Term Bioretention Performance Review]

That does not mean rain gardens are maintenance-free.

Poorly designed or neglected installations can fail.

Vegetation dies.

Inlets clog with trash.

Sediment accumulates.

Weeds dominate.

Standing water may persist where drainage was supposed to occur.

San Francisco itself requires ongoing inspection for green infrastructure installed under its stormwater requirements. Property owners responsible for qualifying installations must conduct annual inspections and certify that the systems remain functional. [https://www.sfpuc.gov/construction-contracts/new-developments/stormwater-management SFPUC: Stormwater Management Requirements]

But the long-term research suggests a fascinating possibility:

'''A mature rain garden may not simply be an old piece of infrastructure. It may become a more developed ecosystem.'''

That is very different from the aging process of a pipe.

=== The Same Biology Creates Tradeoffs ===

Living infrastructure is not automatically superior infrastructure.

Ecological complexity creates benefits, but it also creates conflicts.

Leaves provide habitat and organic matter but can clog drainage structures.

Dense vegetation can improve habitat while making inspection more difficult.

Plants useful for pollution removal may accumulate contaminants.

A garden that traps microplastics protects downstream waters but concentrates those particles within its soil.

Vegetation desirable to wildlife may look untidy according to conventional streetscape expectations.

A plant species exceptionally good at absorbing water may not provide much habitat for local insects.

An installation designed to maximize infiltration might not maximize cooling.

Research into blue-green infrastructure increasingly emphasizes exactly these tradeoffs. The goal cannot simply be to attach as many claimed benefits as possible to every green project. The ecological and engineering objectives have to be intentionally designed together. [https://www.nature.com/articles/s42949-024-00163-y npj Urban Sustainability: Engineering Blue-Green Infrastructure for Biodiversity]

San Francisco therefore should not assume that a rain garden producing one benefit automatically produces all the others.

The benefits have to be measured.

=== The Billion-Gallon Number Is Necessary—but Not Sufficient ===

This brings the article back to San Francisco's most memorable target:

'''1 billion gallons of stormwater managed annually by 2050.'''

The number is useful.

It gives the city a concrete objective.

It allows progress to be measured.

It communicates the enormous cumulative effect that many small projects could eventually have.

But if San Francisco genuinely wants to understand whether green infrastructure is becoming ecological infrastructure, the billion-gallon figure cannot be the only score.

Imagine two hypothetical systems.

Both manage exactly 1 billion gallons.

The first consists largely of installations with minimal vegetation, little wildlife value, limited pollutant removal and no particular relationship to neighborhoods facing heat or environmental burdens.

The second manages the same quantity of water while removing substantial amounts of pollution, expanding native habitat, connecting ecological corridors, adding canopy to overheated neighborhoods and providing vegetation in communities historically underserved by green space.

Hydrologically, the programs might look identical.

Ecologically and socially, they would be very different.

The logical next step for San Francisco is therefore to expand its definition of performance.

=== What Should San Francisco Measure? ===

The city already has many of the pieces required to create a much broader accounting system.

For '''water''', it can measure:

stormwater volume managed;

volume prevented from entering the sewer;

peak-flow reduction;

infiltration rates;

and changes in performance as installations age.

For '''pollution''', it can measure:

tire-wear particles;

6PPD-quinone;

metals;

PCBs;

microplastics;

sediment;

and emerging contaminants.

For '''ecology''', it could increasingly measure:

native plant survival;

vegetation cover;

pollinator abundance;

bird use;

soil biological activity;

species richness;

functional diversity;

and the degree to which individual installations improve habitat connectivity.

For '''climate resilience''', it could measure:

surface temperature;

shade;

tree-canopy growth;

soil moisture;

and performance during droughts and extreme rainfall.

And for '''environmental equity''', it could examine:

which neighborhoods receive investments;

who receives additional canopy;

where heat is reduced;

which environmental-justice communities benefit;

whether maintenance resources are distributed equitably;

and whether existing residents remain able to enjoy improvements to their neighborhoods.

This would create a very different kind of infrastructure dashboard.

Instead of asking only:

'''How much water did we manage?'''

San Francisco could ask:

'''How many ecological functions did we restore, where did we restore them, and who benefited?'''

That would be a far more difficult measurement problem.

It would also tell the city far more about what it was actually building.

=== Success Would Have to Be Measured Over Decades ===

There is another reason monitoring matters.

A rain garden that performs well during its first year has not necessarily succeeded.

Trees take years to produce substantial shade.

Habitat value may increase as vegetation matures.

Pollutants may gradually accumulate.

Soils evolve.

Plant communities change.

Maintenance practices change.

Climate conditions change.

The 2026 long-term review of bioretention systems found encouraging evidence for mature installations, but also emphasized the scarcity of performance data beyond roughly two decades. [https://www.sciencedirect.com/science/article/pii/S0301479726013496 Journal of Environmental Management: Long-Term Bioretention Performance Review]

That is especially relevant to San Francisco.

The billion-gallon target is set for '''2050'''.

Projects installed today will need to function through decades of storms, droughts, changing temperatures and continued urban development.

The correct question is therefore not simply:

'''Did this rain garden work when it was constructed?'''

It is:

'''Will the network still work in 2040, 2050 and beyond?'''

That turns maintenance into part of ecological management rather than a housekeeping expense.

Dead plants must be replaced.

Blocked inlets must be cleared.

Sediment must be managed.

Pollutant accumulation must be monitored.

Plant communities may need to be adjusted as climate conditions change.

And monitoring results should feed back into future designs.

In ecology this is often called '''adaptive management''': observe what actually happens, compare it with what was intended, and change management accordingly.

The city's hundreds of existing installations could therefore become more than infrastructure.

They could become hundreds of small experiments.

=== Climate Change Raises the Standard ===

Climate change adds another layer of uncertainty.

San Francisco's monitoring has already demonstrated that storm intensity matters.

A rain garden can absorb an entire long, moderate storm and overflow during another storm producing nearly the same rainfall over a shorter period.

Future infrastructure therefore has to work under conditions that may differ from those under which it was originally designed.

Vegetation must survive longer dry periods.

Soils must remain functional after drought.

Systems must accommodate increasingly intense precipitation events.

Trees must withstand heat and water stress.

Species selected today must remain viable several decades from now.

This means resilience cannot be measured by asking whether green infrastructure performs perfectly.

No realistic stormwater system can guarantee that.

A more meaningful definition of resilience is whether the system '''continues providing useful functions when conditions depart from the ordinary'''.

Does it still reduce peak runoff during an extreme storm even if it overflows?

Does vegetation recover after drought?

Does a diverse planting retain some function when one species fails?

Can one part of a distributed network compensate when another part performs poorly?

These are ecosystem questions as much as engineering questions.

=== Distributed Infrastructure Has an Unusual Advantage ===

San Francisco's approach may possess another form of resilience simply because it is distributed.

A conventional treatment plant is extraordinarily efficient, but much of its capacity is concentrated in one place.

A distributed green-infrastructure system consists of hundreds or thousands of smaller components.

One rain garden can clog without the entire system failing.

One tree can die without eliminating the city's canopy.

One project can perform poorly while another performs better than expected.

New projects can be added gradually.

Old designs can be modified as scientists learn more.

That modularity does not make green infrastructure superior to conventional infrastructure. San Francisco still requires pipes, pumping stations, storage structures and treatment plants.

The two systems do different things.

The more interesting possibility is that they can complement one another.

Underground infrastructure provides enormous engineered capacity.

Surface green infrastructure reduces the burden placed on it while producing benefits underground pipes cannot provide.

A sewer pipe does not feed a butterfly.

A storage tunnel does not shade a pedestrian.

A wastewater plant cannot create a habitat corridor along a neighborhood street.

A rain garden cannot replace any of those facilities.

But it can perform several smaller functions at once.

=== From Gray Infrastructure to Hybrid Infrastructure ===

This may ultimately be the most useful way to understand what San Francisco is building.

The future city does not have to choose between '''gray infrastructure''' and '''nature'''.

It can increasingly use hybrid systems.

A creek can carry stormwater while providing habitat.

A park can store floodwater.

A street tree can shade pavement while intercepting rainfall.

A rain garden can reduce sewer flows while trapping tire particles.

A wetland can become infrastructure.

And infrastructure can become habitat.

The 2024 research on blue-green infrastructure argues that cities should deliberately engineer these systems '''for biodiversity and with biodiversity''', rather than designing engineering first and treating ecological benefits as accidental extras. [https://www.nature.com/articles/s42949-024-00163-y npj Urban Sustainability: Engineering Blue-Green Infrastructure for Biodiversity]

San Francisco is beginning to move in that direction.

Its biodiversity policies call for connecting natural habitat through the built environment. [https://www.sfenvironment.org/city-san-francisco-biodiversity-guidelines San Francisco Biodiversity Guidelines]

Its Islais Creek research is identifying locations where stormwater management can simultaneously improve biodiversity, tree-canopy access and heat resilience. [https://www.sfei.org/documents/watershed-plan-case-study-approach-incorporating-multiple-benefits-green-stormwater SFEI: Watershed Plan]

Its Next Generation Urban Greening program explicitly describes its goal as integrating water quality, biodiversity, flood resilience and community benefits at watershed scale. [https://www.sfei.org/projects/next-generation-urban-greening SFEI: Next Generation Urban Greening]

And the citywide biodiversity vision calls for robust, interconnected habitats, watersheds and urban forests integrated throughout the natural and built environment. [https://www.sfenvironment.org/policy/resolution-adopting-citywide-biodiversity-goals San Francisco Environment: Citywide Biodiversity Goals]

These efforts do not prove that San Francisco has created a functioning citywide ecosystem.

They show that the city is beginning to plan infrastructure as though ecological relationships matter.

That is an important difference.

=== Can a Watershed Be Reassembled Without Removing the City? ===

San Francisco cannot restore most of its historical watersheds in a conventional sense.

But perhaps a watershed can exist in more than one form.

Imagine rain falling over a future neighborhood.

Some lands on a green roof.

Some flows toward a street tree.

Some enters a rain garden.

Some infiltrates into soil.

Some is stored and reused.

Some moves through a daylighted creek.

Some passes through a constructed wetland.

Only the remainder enters a conventional pipe.

Along the way, vegetation cools the neighborhood.

Pollutants are filtered.

Insects encounter flowers.

Birds encounter insects.

Plants and soil organisms occupy spaces that previously consisted almost entirely of pavement.

The original watershed has not returned.

But the city behaves '''a little more like a watershed again'''.

That may be the more realistic objective for urban ecological restoration.

Not to erase the city.

Not to pretend a rain garden recreates wilderness.

But to redesign the city so that water, soil, vegetation and wildlife once again participate in processes that urban development had systematically removed.

=== The Real Experiment ===

San Francisco's billion-gallon initiative began with a practical engineering problem.

Too much rainwater was entering a combined sewer system.

The solution was to intercept some of it before it reached the pipes.

That remains the program's essential purpose.

But once soil and plants were inserted between pavement and sewer, other possibilities appeared.

The soil filtered pollutants.

The plants created habitat.

Trees produced shade.

Native vegetation strengthened ecological connections.

Stored water could be reused.

Neighborhood streets became greener.

Researchers began measuring microplastics and tire chemicals.

Ecologists began asking where projects could best support wildlife.

Public-health planners began asking where vegetation could reduce heat.

Environmental-justice planners began asking who was receiving the benefits.

What began as '''stormwater management''' gradually became a much larger question about how a city should function.

That is why San Francisco's rain gardens are worth watching.

The important experiment is not whether a planter can absorb rain.

We already know that it can.

The experiment is whether '''hundreds of small pieces of living infrastructure can be deliberately connected, measured and managed until they begin functioning as part of a larger urban ecological system'''.

If San Francisco eventually reaches its billion-gallon target, that number will be impressive.

But the more consequential achievement would be harder to express in gallons.

It would be demonstrating that one of America's densest cities can restore meaningful ecological processes without ceasing to be a city.

A street would still be a street.

A neighborhood would still be a neighborhood.

A sewer would still be necessary.

But rain would encounter soil before pipes.

Pollution would encounter biological filters before the Bay.

Wildlife would encounter habitat between parks.

Residents would encounter vegetation where there was once only pavement.

And the city itself would begin doing, in small but cumulative ways, some of the work that the landscape once did on its own.

That may be the larger lesson of San Francisco's rain gardens:

'''Urban nature does not have to exist only in the places where the city stops. It can increasingly be built into the way the city works.'''
```