You have probably read that people who live near green space are healthier. Thousands of studies say so. Very few of them tell you what “green space” was in the study.
Here is the short version. In most of this research, nobody counted trees. A satellite measured how much red light and how much near-infrared light bounced off the ground, and a computer turned that ratio into a greenness score for a square of land roughly 100 feet on a side.
That is not a complaint. Satellites are the only practical way to measure the surroundings of millions of people at once. But it changes what the results can mean, and it is not even the hardest problem in this field.
The hardest problem is that greener neighborhoods differ from less green ones in a long list of other ways, and pulling the greenery apart from everything that travels with it is the question this entire body of research keeps running into.
How a satellite decides your street is green
The measure behind most of these studies is the normalized difference vegetation index, usually shortened to NDVI. It works on a quirk of leaves. Chlorophyll absorbs red light to run photosynthesis, roughly in the 0.4 to 0.7 micrometer range, while the internal cell structure of a leaf throws back near-infrared light strongly, roughly 0.7 to 1.1 micrometers. Compare those two bands and you get a number.
NASA gives the formula as NDVI equals near-infrared minus visible, divided by near-infrared plus visible. The US Geological Survey publishes the same calculation as a Landsat product, with the specific band numbers changing between satellite generations: bands 4 and 3 on Landsat 4 through 7, bands 5 and 4 on Landsat 8 and 9.
The output runs from minus 1 to plus 1. According to NASA, values at 0.1 and below indicate barren rock, sand or snow. Shrubland and grassland sit around 0.2 to 0.3. Temperate and tropical rainforest reaches 0.6 to 0.8, and 0.8 to 0.9 is about as dense as green leaf cover gets.
Now the part that matters for reading a study. NDVI does not know what kind of green it is looking at. A lawn, a cemetery, a golf course, a soybean field, a highway median and a mature oak canopy are all green to a satellite. It does not know whether the green is fenced, private, mowed to the dirt, behind an interstate, locked at dusk, or across a county line.
It also has technical limits that NASA states plainly. Clouds and airborne particles can block the view entirely, so researchers often average over 10 or 30 days, which reduces cloud errors and loses detail. Unusually cold weather can lower the reading. And the index saturates where chlorophyll is abundant.
Resolution is the last piece. The Environmental Protection Agency’s EnviroAtlas, a federal mapping tool for ecosystem data, uses the National Land Cover Database at 30-meter resolution for its national layers. Higher-resolution 1-meter land cover exists, but only for about 30 urbanized community areas covering more than 1,400 cities and towns. For most of the country, the fine-grained version of your neighborhood has never been mapped.
Three other ways to measure the same neighborhood
Satellite greenness is the most common exposure in this literature, not the only one. Three others show up constantly, and they do not measure the same thing.
Straight-line distance to the nearest park. Cheap to calculate from an address and a parks layer. It ignores the highway in between, the fence, the entrance on the far side, the hours, and whether the park is a ball field with no shade.
Land-cover share inside a buffer. A circle is drawn around a home and the percentage of green inside it is calculated. A 2009 study in the journal Health and Place by Jolanda Maas and colleagues at the Netherlands Institute for Health Services Research used exactly this, computing percent green within 1 kilometer and 3 kilometers for 10,089 Dutch residents. Note what the circle is centered on in many studies: not a front door, but the coordinates of a postal code.
Tree canopy cover. A narrower measure that separates a street lined with sycamores from a field of weeds.
These measures can disagree about the same block, which is why the choice is not a technicality. A 2017 expert workshop report published in Environmental Research by Iana Markevych and 18 co-authors, including researchers at Oregon State University and the University of British Columbia, recommends that studies pick greenspace measures relevant to the pathway they are testing, and that they assess presence, access and use separately. That recommendation exists because most studies have not done it.
Six proposed mechanisms, weighed one at a time
The report in Environmental Research sorts the possible pathways into three functions: reducing harm, restoring capacities and building capacities. Here is each proposed mechanism with its actual weight.
Air quality
This one has the best numbers and the most surprising conclusion. A 2014 study in Environmental Pollution by David Nowak and colleagues at the USDA Forest Service modeled pollution removal by trees across the contiguous United States. Trees and forests removed an estimated 17.4 million tonnes of air pollution in 2010, with a plausible range of 9.0 to 23.2 million tonnes. The authors valued the associated health effects at $6.8 billion and estimated that removal avoided more than 850 deaths and about 670,000 incidences of acute respiratory symptoms.
Then comes the qualifier the paper states directly: all of that equated to an average air quality improvement of less than one percent. Most of the removal happened in rural areas, while most of the health value landed in cities, because that is where the people are.
So the air-quality pathway is real, measurable and small as a share of what you breathe. Trees also emit volatile organic compounds and pollen, which cuts the other way for some people.
Noise
Noise appears in the reducing-harm column of the Markevych framework, and the logic is intuitive. The trouble is that green space studies rarely measure noise and greenness at the same time, so within this literature the step from trees to quieter remains a hypothesis rather than a measured link. Article 721 covers noise and health directly.
Physical activity
A 2018 systematic review in the International Journal of Environmental Research and Public Health, led by Michelle Kondo of the USDA Forest Service with co-authors at the University of Pennsylvania and Columbia, looked specifically at urban green space. Physical activity was among the outcomes with a consistent positive association, along with attention and mood.
Direction of causation is the open question. People who already want to walk may choose greener places to live, and the review’s authors call for more rigorous designs before anyone generalizes.
Social contact
The Maas study above measured this. After adjusting for socioeconomic and demographic characteristics, less green nearby went along with reported loneliness and a perceived shortage of social support, and those two factors partly explained the link between green space and health.
Read it carefully. It is cross-sectional, self-reported, and conducted in the Netherlands, and a mediation analysis on observational data describes a pattern rather than proving a chain. Article 675 owns loneliness and 734 owns group outdoor activity.
Stress recovery
This is the mechanism most people find obvious, and it is among the least consistent in the urban epidemiology. The Kondo review found mixed results or no association for stress measured through cortisol concentration, and the same for general health and for depression. Article 732 covers the psychological theories.
Microbial exposure
The most speculative and the most interesting. A 2012 paper in the Proceedings of the National Academy of Sciences by Ilkka Hanski and colleagues at the University of Helsinki studied adolescents across a 100 by 150 kilometer region of Finland. Environmental biodiversity around their homes was related to the bacterial classes living on their skin. Adolescents with allergic sensitization had less biodiversity near home and lower diversity of one bacterial class, gammaproteobacteria, on their skin.
That is a cross-sectional association in one region, about skin bacteria and allergic disposition. It is a long way from showing that adding greenery changes anyone’s immune development. Call it preliminary research, which is what the authors do.
Why greener neighborhoods differ in every other way too
Here is the center of the subject. In most countries, greener addresses are wealthier addresses. They tend to have quieter streets, less traffic, newer housing, better-funded services and residents who were already healthier when they moved in.
Researchers handle this by adjusting statistically for income, education and area deprivation. Adjustment helps. It does not finish the job, because the variables available in administrative data are crude substitutes for the thing that differs between two streets.
There is also the problem of who moved where. Healthier, more active people may select into leafier neighborhoods, which would produce the same correlation with no causal arrow at all.
One research design handles this better than adjustment does, and it is worth understanding because it shows up rarely. It is the negative control. A 2008 study in The Lancet by Richard Mitchell and Frank Popham at the University of Glasgow classified the entire population of England below retirement age, 40,813,236 people, by income deprivation and green space exposure, then tracked 366,348 deaths between 2001 and 2005. Income-related inequality in all-cause mortality was smaller in the greenest areas: an incidence rate ratio of 1.93 comparing the most and least deprived quartiles in the least green areas, against 1.43 in the most green. For circulatory disease the figures were 2.19 and 1.54.
The design detail is the clever part. The authors also tested two causes of death they did not expect green space to influence, lung cancer and intentional self-harm, and found no such pattern. If greenery were purely standing in for being better off, you would expect those to move too.
Then there is a finding that complicates the confounding story from the other direction. When the emerald ash borer, an invasive beetle, killed roughly 100 million ash trees, Geoffrey Donovan and colleagues at the USDA Forest Service treated it as a natural experiment. Their 2013 paper in the American Journal of Preventive Medicine analyzed county-level mortality from 1990 to 2007 across 15 states and found increases in cardiovascular and lower-respiratory deaths in infested counties, an estimated 15,080 additional cardiovascular deaths and 6,113 additional lower-respiratory deaths. A 2015 follow-up in Health and Place using Women’s Health Initiative data reported a hazard ratio of 1.25 for cardiovascular disease among women living in infested counties.
The strange part: the effect was larger in counties with above-average median household income. That is the opposite of what a simple income-confounding story predicts, and it is hard to route through any of the six mechanisms above. Nobody has a clean explanation.
A natural experiment removes self-selection, since nobody chose to have a beetle kill their street trees. It does not remove everything. Ash trees were not randomly distributed, the infestation spread along highways and nursery routes, and the outcome was measured at county level.
The studies that come closest to an experiment
Two trials in Philadelphia got as close to random assignment as this field has managed, and both are worth knowing in detail.
A 2018 paper in the Proceedings of the National Academy of Sciences by Charles Branas and colleagues at Columbia, Penn, Rutgers and the USDA Forest Service randomly sampled 541 vacant lots and randomly assigned them to treatment or control, then followed police records and 445 sampled residents over 38 months. The greening package was unglamorous: remove trash, grade the land, plant grass and a few trees, install a low wooden fence, and maintain it monthly. Residents near treated lots reported a 36.8 percent drop in perceived crime and a 57.8 percent drop in safety concerns about going outside, plus a 75.7 percent increase in using outdoor space to relax and socialize. In neighborhoods below the poverty line, recorded crime fell 13.3 percent and gun violence 29.1 percent. The paper also notes the scale of the raw material: about 15 percent of land in US cities is classified as vacant or abandoned.
A companion trial published the same year in JAMA Network Open, led by Eugenia South at the University of Pennsylvania, measured self-reported psychological distress. It enrolled 442 adults across 110 clusters containing those 541 lots and split them three ways: full greening, trash cleanup only, or nothing. Compared with nothing, greening was followed by a 41.5 percent decrease in participants reporting feeling depressed and a 50.9 percent decrease in feeling worthless, while the overall measure of poor self-reported mental health fell 62.8 percent but missed statistical significance at p equals .051.
The most informative result in that trial is the arm that did nothing much. Trash cleanup alone produced no significant change. That separates “somebody cleaned up your block” from “somebody planted something,” which almost no other study in this field can do.
The limits are real too. Nobody can be blinded to a new lawn, the mental health outcomes were short questionnaire items rather than diagnoses, it is one city, blighted vacant lots are a specific starting point, and the greening package changed several things at once.
Availability is not use, and the dose is unknown
Every measure in the first half of this article captures what sits near you. None of them captures whether you go.
The clearest evidence on that gap comes from a 2019 study in Scientific Reports by Mathew White and colleagues at the University of Exeter Medical School, with a co-author at the University of Michigan. Drawing on 19,806 respondents to a national English survey, weighted to be representative, the authors asked about recreational nature contact in the previous seven days. People reporting 120 minutes or more were significantly more likely to report good health and high wellbeing. The association peaked between 200 and 300 minutes a week and rose no further. It made no difference whether the 120 minutes came in one visit or several.
Two features of that study matter here. First, the analysis controlled for residential green space and still found the pattern, which tells you availability and use are genuinely different exposures. Second, the authors state plainly that prospective and intervention studies are the critical next step before anyone writes a weekly nature guideline. Nobody has written one.
That is the honest position on dose. The Markevych workshop concluded that much remains unknown about the functional form of these relationships, which in plain words means nobody can tell you that twice as much greenery is twice as good, or where the benefit flattens out, or whether adding trees to an already leafy street does anything at all.
Even the federal agency that built a tool on this literature labels it that way. EPA’s Eco-Health Relationship Browser says many of the studies behind it do not prove cause and effect, and instead highlight “statistically significant, plausible associations in space and time.”
Five questions that sort a green space headline in a minute
- What was measured: satellite greenness, distance to a park, or actual visits?
- How big was the buffer, and was it centered on a front door or a postal code?
- Did anyone assign the exposure, or did people choose where to live?
- What did the comparison group get? “Nothing” and “we removed the trash” are very different comparisons.
- Is the outcome a death, a diagnosis or a questionnaire score, and over what span of time?
What this does not establish, and when to see a clinician
More green is not uniformly better in every direction. Pollen is a cost for some people, ticks are a real exposure that article 727 covers, maintenance budgets decide whether a green space is usable at all, which is article 731’s subject, and access to green space is unevenly distributed in ways a later article takes up directly.
Nothing in this literature has established a dose, a threshold, or that the benefits are the same for everyone. The strongest claim it supports is that the association shows up repeatedly, in large populations, in several countries, and that two Philadelphia trials and one beetle infestation point the same direction.
One practical boundary. If you are thinking about spending more time outdoors because something feels persistently wrong, that is a reason to get evaluated, not a reason to walk farther. New chest pressure, unusual breathlessness or dizziness while walking needs same-day medical attention. Low mood, anxiety or disrupted sleep lasting more than two weeks, or getting worse, deserves an appointment rather than a project. If you are in crisis, call or text 988. None of the studies described here tested green space as a replacement for care, and none of them support using it that way.
A filter for reading this research
The best output of this research for a reader is not a prescription. It is a filter.
You now know to ask what was measured and who chose. The studies that come closest to an answer are the ones where somebody other than the resident decided: a beetle, or a lottery over 541 vacant lots. Those studies are rarer, smaller and messier than the satellite ones. They are also the ones to read first.
Sources
- NASA. “Measuring Vegetation (NDVI & EVI), NASA Earth Observatory.” science.nasa.gov
- US Geological Survey. “Landsat Normalized Difference Vegetation Index.” usgs.gov
- US Environmental Protection Agency. “About EnviroAtlas Data (19 January 2025 web snapshot).” 19january2025snapshot.epa.gov
- US Environmental Protection Agency. “EnviroAtlas Eco-Health Relationship Browser (19 January 2025 web snapshot).” 19january2025snapshot.epa.gov
- Markevych I, Schoierer J, Hartig T, et al.. “Exploring pathways linking greenspace to health: Theoretical and methodological guidance. Environmental Research, 2017. PubMed. DOI: 10.1016/j.envres.2017.06.028.” doi.org
- Kondo MC, Fluehr JM, McKeon T, Branas CC.. “Urban Green Space and Its Impact on Human Health. International Journal of Environmental Research and Public Health, 2018. PubMed. DOI: 10.3390/ijerph15030445.” doi.org
- Branas CC, South E, Kondo MC, et al.. “Citywide cluster randomized trial to restore blighted vacant land and its effects on violence, crime, and fear. Proceedings of the National Academy of Sciences, 2018. PubMed. DOI: 10.1073/pnas.1718503115.” doi.org
- South EC, Hohl BC, Kondo MC, MacDonald JM, Branas CC.. “Effect of Greening Vacant Land on Mental Health of Community-Dwelling Adults: A Cluster Randomized Trial. JAMA Network Open, 2018. PubMed. DOI: 10.1001/jamanetworkopen.2018.0298.” doi.org
- Donovan GH, Butry DT, Michael YL, et al.. “The relationship between trees and human health: evidence from the spread of the emerald ash borer. American Journal of Preventive Medicine, 2013. PubMed. DOI: 10.1016/j.amepre.2012.09.066.” doi.org
- Donovan GH, Michael YL, Gatziolis D, Prestemon JP, Whitsel EA.. “Is tree loss associated with cardiovascular-disease risk in the Women's Health Initiative? A natural experiment. Health & Place, 2015. PubMed. DOI: 10.1016/j.healthplace.2015.08.007.” doi.org
- Nowak DJ, Hirabayashi S, Bodine A, Greenfield E.. “Tree and forest effects on air quality and human health in the United States. Environmental Pollution, 2014. PubMed. DOI: 10.1016/j.envpol.2014.05.028.” doi.org
- Hanski I, von Hertzen L, Fyhrquist N, et al.. “Environmental biodiversity, human microbiota, and allergy are interrelated. Proceedings of the National Academy of Sciences, 2012. PubMed. DOI: 10.1073/pnas.1205624109.” doi.org
- Maas J, van Dillen SME, Verheij RA, Groenewegen PP.. “Social contacts as a possible mechanism behind the relation between green space and health. Health & Place, 2009. PubMed. DOI: 10.1016/j.healthplace.2008.09.006.” doi.org
- Mitchell R, Popham F.. “Effect of exposure to natural environment on health inequalities: an observational population study. The Lancet, 2008. PubMed. DOI: 10.1016/S0140-6736(08)61689-X.” doi.org
- White MP, Alcock I, Grellier J, et al.. “Spending at least 120 minutes a week in nature is associated with good health and wellbeing. Scientific Reports, 2019. PubMed. DOI: 10.1038/s41598-019-44097-3.” doi.org