Where should Tallinn plant new nature first?
We scored every square of the Tallinn area on four simple questions, to find where new, biodiverse green space would reach the most people — above all children.
Damiano Cerrone · Siim Tuksam · Heli Telgmaa — the EKA EDITUA team, Faculty of Architecture, Estonian Academy of Arts
In one minute
Green means: new nature would help most here.
Every square on this map has a score from 0 to 100. Deep green squares are where new, biodiverse nature would reach the most residents and children every day. Orange squares are where few people live, so new nature there would reach few residents.
The score comes from four questions: Do many people live here? Are there places for children nearby? Is nature missing? Are everyday services nearby?
Whether there is free land to plant is a separate check (chapter 08).
How to read this page
Scroll, and the map follows the story.
First, why nature near home matters. Then how the four questions build the score, what the two workshop teams chose instead, and how different the three views of the city really are.
At the end you can explore every square yourself — in the Tallinn area and across all of Estonia — and try your own weights.
On a computer, point at any square on the map to see its score.
Biodiversity and child health
Living variety, not just “green”, is what helps.
When daycare yards in Finland were given forest floor, turf and planters, the children’s immune systems showed healthier signs within 28 days.
It is contact with living biodiversity that does it — soil, plants and their microbes — not a lawn alone.
Roslund et al. 2020 · Science Advances; see also Hanski et al. 2012, PNAS
Reach where people are
Put it where people live.
Studies that use maps to choose sites for new green space put people first. In one such study, how many people live nearby was the strongest single factor — about a third of the decision (a weight of 0.357).
Environmental Systems Research 2024 · a GIS siting study in Lilongwe, Malawi
Target the gap
Most people are missing out.
The 3-30-300 rule: three trees visible from every home, 30% tree cover in every neighbourhood, and quality green space within 300 metres of everyone.
Across 862 European cities, fewer than 15% of people meet it. So new nature should go first to the people without it — not to places that are already green.
Konijnendijk 2023 · the 3-30-300 rule; 862-city test: Nature Communications 2026
Empty lots work
Greening empty land pays off fast.
In Philadelphia, a randomised trial cleaned and planted vacant lots. Neighbours’ feelings of depression fell — by up to 69% in neighbourhoods below the poverty line.
It is cheap, and it scales.
South, Hohl, Kondo, MacDonald & Branas 2018 · JAMA Network Open
for daycare children’s immune systems to show healthier signs after their yards were given forest floor and planting.
Roslund et al. 2020the strongest single factor when a GIS study weighed where new green space helps most (a weight of 0.357).
Environmental Systems Research 2024in 862 European cities. Prioritise the people without nature, not the already green.
Nature Communications 2026among residents near greened vacant lots in Philadelphia neighbourhoods below the poverty line (randomised trial).
South et al. 2018So the aim is simple: add nature where it is missing, where people — above all children — will meet it every day.
The Tallinn area as a grid
5,966 squares, each with its own answers.
The study area is Tallinn plus a ring of suburbs and small towns reaching up to about 8 km beyond the city limits (parts of Harku, Saue, Saku, Kiili, Rae, Jõelähtme and Viimsi parishes, and the town of Maardu). In built-up areas the squares are 125 × 125 metres — about one city block. Where fewer people live they grow to 250 m, 500 m, 1 km and 2 km.
For every square we know what can be reached on foot within 15 minutes: kindergartens, schools, shops, doctors, walking routes, bathing places, transport terminals and much more.
Question 1 counts 25%
Do many people live here?
New green space helps more people when more people live next to it. We count the homes in each square — individual flats, so one apartment block counts many times.
A dense housing block counts for far more than an empty industrial strip.
Question 2 counts 20%
Are there places for children nearby?
The health benefits of nature are strongest for children. We cannot count children directly, so we count kindergartens, childcare, upper-secondary schools, playgrounds, children’s clubs and children’s shops within a 15-minute walk — a good sign of where young families are.
A block between two kindergartens and a school scores higher.
Question 3 counts 35% — the biggest slice
Is nature missing?
If people can already walk to a walking route, a nature spot, a dog-walking area, a bathing place or a theme park, they are well served, and the square scores lower. Where none is within 15 minutes, it scores higher.
It is the biggest single slice because the aim is to reach people who lack nature today.
On the map: darker = fewer such places within a 15-minute walk. It counts places to go, not trees — so the forested edge of the area looks dark too.
Question 4 counts 20%
Are everyday services nearby?
We favour places that are part of everyday life — where people walk to the grocery, pharmacy, doctor, post office, library or a transport terminal.
Greening a high-street corner completes a living neighbourhood; greening a remote field reaches almost nobody.
Blend the four answers
One score from 0 to 100.
Each answer becomes a level from 0 to 100, measured against the best-served squares in the area. The levels are weighted by the size of their slice and added up. Finally the results are spread out so that the lowest squares get 0 and the highest get 100.
Missing nature is the biggest single slice (35%). The two people questions together count 45%, because research says the number of people reached matters most (chapter 01).
A real square that scores 100
Many homes, many children’s places, little nature.
This square in central Tallinn has about 170 homes, many places for children and everyday services close by — and little nature within a 15-minute walk.
Its weighted answers add up to 82 — more than 99% of squares in the area — so it sits at the top of the 0–100 scale.
A real square that scores 0
Nobody lives here, and nature is next door.
This square has no homes, and walking routes, a nature spot and a bathing place are all within easy reach. New greening here would reach almost nobody who lacks nature today.
Its weighted answers add up to 17 — lower than 99% of squares — so it sits at 0.
The result
The Fair-Share map.
Put together, the four answers light up dense neighbourhoods with many homes, children’s places and everyday services. Some nature is often within reach there already — but so many people live there that new nature reaches the most people.
Where nature is missing, a square gets extra points. That matters most around the edges of the dense city, as chapter 06 shows.
We call it the Fair-Share score: the research-based view, built from the evidence in chapter 01.
Four kinds of place
From most people reached to fewest.
The four kinds of place follow the colours of the map, from the highest score to the lowest.
Next, each kind of place on the map.
80–100 331 squares
Deep green: plant first.
The most homes, children’s places and everyday services within reach. New biodiverse green here reaches the most people and completes the neighbourhood.
60–80 631 squares
Light green: a strong case.
Many homes and services nearby, and room for more nature. Plan these together with the deep-green squares.
40–60 1,120 squares
Pale yellow: fewer people, less nature than the greens.
Far fewer homes than the deep-green squares, and fewer services — but less nature within reach. Greening helps most here when it comes with new homes, shops, care or transport.
0–40 3,884 squares
Orange: few people to reach.
Few or no homes — more than a quarter of these squares have none. Some already have nature next door; others have nothing nearby, but few people to use it. Green here when a chance comes up.
Why don’t empty fields score zero? Where nothing green is within reach, the “missing nature” question adds points even if nobody lives there — so squares without homes still land around 24. That is a known limit of the method (see Method).
Four kinds of place · highest score first
The most homes, children’s places and everyday services within reach.
Many homes and services nearby; room for more nature.
Pair greening with new homes, shops, care or transport.
Few or no homes. Green when a chance comes up.
How we rebuilt them
Five categories each, 20% each.
Each team named five categories. We matched every category to the data we have and gave each one an equal share. Here the rule is simple: the more of a category — homes in the square, everything else within a 15-minute walk — the higher the score.
Same colour = same data. Team 1’s two dark-blue slices both use homes.
The teams defined their categories during the session; the scores were computed afterwards. The teams did not see or tune any numbers.
Team 1
The Active city.
Home streets, education, food shops, services and dense housing.
On the map, high scores shrink to the built-up, populated city and concentrate in active, well-served neighbourhoods.
Team 1 · what the data could hold
Two gaps to know about.
Parking lots have no matching data and were dropped. There is no street layer, so “home streets” uses homes as a stand-in — the same data as “dense housing”. In effect, homes count twice.
Separating them would need a street layer and a parking layer.
Team 2
The Everyday-life city.
Home (kodu), food (toit), movement (liikumine), interests (huvid) and education (haridus).
A near-twin of Team 1: the same lived-in core of the city lights up.
Team 2 · one catch
Movement works the opposite way to Fair-Share.
For liikumine (movement), more walking routes, bathing places and nature spots nearby gives a higher score. In Fair-Share, missing nature gives a higher score.
If Team 2 meant “green where people lack places to move outdoors”, this category should be reversed.
On the map: darker = more places for outdoor movement — almost the reverse of the missing-nature map in chapter 02.
Team 1 · matched to our data
| Team category | Matched to | Status |
|---|---|---|
| Streets (home streets) | Homes, as a stand-in | ~ Stand-in |
| Education | Kindergarten, childcare, upper-secondary, higher, vocational & continuing education | ✓ Matched |
| Shops — food (parking lots) | Grocery, market, eatery, bar | ~ Partial ✗ parking |
| Services | Pharmacy, post, bank, fuel, beauty, home services, public toilet… | ✓ Matched |
| Dense housing | Homes | ✓ Matched |
Fair-Share · based on research
The Fair-Share city.
A fair city gives everyone — above all children — nature on their doorstep.
It reads the city by who lives there and what they lack: besides the dense, busy core that every view shares, it also picks quieter family areas with little nature nearby — places the teams miss.
What it misses: it can pick less visible places, and it counts places to go outdoors, not actual trees and plants.
Team 1’s view
The Active city.
Green should reinforce the parts of the city that are already busy and built-up — homes, streets, schools, shops and services.
It reads the city by its built fabric and footfall, and lights up dense, central, well-served blocks that many people pass every day.
What it misses: it adds green where some may already exist, and can widen the gap for areas without it. Streets and parking are not in the data.
Team 2’s view
The Everyday-life city.
Green should support daily wellbeing — home, food, movement, hobbies and learning all within reach.
It reads the city by quality of everyday life, and lights up complete neighbourhoods that already have amenities and leisure.
What it misses: it rewards places that are already well equipped, so it may overlook the ones that lack recreation.
None of the three is more correct. Each puts a different value first.
Team 1 ↔ Team 2
Nearly the same map.
Each dot is one square. Where Team 1 scores high, Team 2 almost always does too: they agree 0.96 on a scale where 1 means a perfect match (a correlation).
Their shortlists share 88% of squares. In practice, they are one view under two names.
Fair-Share ↔ the teams
Related, not opposed.
Fair-Share agrees less closely: 0.83 with Team 1 and 0.68 with Team 2. Still, all three favour the same dense, lived-in city.
The shared core
831 squares every view agrees on.
Each shortlist holds about 1,200 squares. 831 of them are on all three lists: about 7 in 10.
These are the densest, busiest squares — a large agreed core to start on, whatever you believe.
The fault line
The decisive 222.
222 squares — about 1 in 5 of the Fair-Share shortlist — are on neither team’s list. This is where the fairness question makes the difference: they have homes, but far less nature within reach than the shared core.
They lie in parts of Kristiine, Põhja-Tallinn and Lasnamäe, and in newer suburbs just outside the city, such as Peetri and Laagri.
The teams’ own extra picks are the opposite: squares with nature on the doorstep.
A closer look · Peetri
Homes, but almost no nature and few services.
Around Peetri, a new suburb just outside the city limits in Rae Parish, 27 squares that only Fair-Share picks have homes and children’s places, yet almost no walking routes, nature spots or bathing places — and few shops or services — within a 15-minute walk.
A score that rewards busy places cannot see them. A score that looks for missing nature cannot miss them.
On the team scores these squares get only about 20 (medians: Team 1 24, Team 2 19). A square needs about 47–49 to make a team’s shortlist.
The real finding
Not three views — a shared core plus one choice.
The two team scores give almost the same map. Only one real choice is left:
Do you also fund the 222 squares the fairness question reveals? Fund them, and you close gaps in family areas that lack nature. Skip them, and greening concentrates in the already-thriving core.
What each choice builds
Two real options, two kinds of city.
Follow Fair-Share, and new green also reaches quieter family areas that lack nature: a more equal city. Follow the teams — they pick almost the same squares — and green gathers in busy, well-served blocks: a more concentrated city.
Every square · Team 1 score against Team 2 score
Every square · Fair-Share against each team
What each top-20% shortlist is made of
Two real options
- Green goes to
- the shared dense core, plus quieter family areas that lack nature (the 222)
- Who gains most
- children and residents now without nature nearby
- Main risk
- some chosen sites are less visible
- Green goes to
- dense, central, busy blocks with services and leisure
- Who gains most
- the many people already in busy areas
- Main risk
- areas without nature stay without it
Start with the 831 squares every view agrees on. Then decide about the 222.
National · Fair-Share
Where new nature would reach the most people, nationwide.
Tallinn and the larger towns light up — more than half of the top band is the Tallinn area — because that is where homes, children’s places and services are.
The countryside (faded here) scores below 40: few homes, few services. It also reads as “nature missing”, because the data count mapped places to go, such as trails and beaches, not forests or fields.
Smaller towns are specks at this scale — zoom into any of them in the interactive map below.
National · Team 1
The Active city, nationwide.
Only the built-up centres score 60 or more. Almost everything else scores low, because this view needs homes, schools, shops and services close together.
National · Team 2 · Tartu
The Everyday-life city, here in Tartu.
Again a near-twin of Team 1. Zoomed in on Tartu, Estonia’s second city: the centre and the dense housing areas score 60 or more; the rest does not.
A score depends on the comparison
Compared with all of Estonia: 1,205 top squares.
Compared with the whole country, the Tallinn area is the densest, best-served place in Estonia, so 1,205 of its squares reach 80–100 (shown here).
Same squares, same method
Compared within the Tallinn area: only 331.
Measured against the Tallinn area instead of all of Estonia, only 331 squares stay on top. Neither is wrong: use the national map to compare towns, and the Tallinn-area map to choose sites within it — as this report does.
Three more checks
Can we actually plant there?
Only empty land, or land that can change use, can be planted — so that check comes first. Ownership decides how: on public land the city can act; on private land it must agree with the owner. How many people live around a site tells how many each project reaches.
None of the three is in the data yet — they are the next layers to add.
Check it on the ground
Look inside each square.
The score measures access to nature, not trees or free land. Each outlined square is a whole city block: on recent aerial photos, look inside it for yards, lots and verges that could be planted — and check they are not already green.
Here, outlined in white: squares on all three shortlists in Mustamäe, over the Estonian Land and Spatial Development Board’s aerial photo.
Test the weights
Nature 35, homes 25, children 20, services 20: a judgement.
Research supports what the four questions ask. The exact weights are the research team’s judgement, and can be adjusted together with the city and residents.
Try your own mix below and see how the shortlist moves.
Try your own weights ↓From priority squares to real sites
09Explore
Explore the maps yourself.
Switch between the three scores, see which squares are on each shortlist (Who agrees), compare the Tallinn area with all of Estonia, change the background, and click any square to see its answers. In Your mix, set your own weights.
Drag to change how much each question counts. The shares always add up to 100%.
Hold Ctrl (Windows) or ⌘ (Mac) while scrolling to zoom; use two fingers on a phone. With a keyboard: focus the map, move with the arrow keys and press Enter to pick the square in the centre. Scores are relative to the area chosen: Tallinn-area squares are scored within the Tallinn area (as in the story), Estonia squares against the whole country.
10For technical reviewers
Method, data and limitations.
This part documents the method so that a GIS analyst or academic reviewer can reproduce and critique it. It follows the technical report (Regreening_Methodology_Report.docx) and the delivered PyQGIS scripts, and records where this version corrects the June 2026 report.
Method at a glance
Input data and coordinate system
The analysis layer is a multi-resolution polygon grid with INSPIRE/EEA-style identifiers (e.g. GRD_ID “125mN6611375E595375”; kilometre cells are coded in km, “1kmN6603E0542”). Cells are 125 m in built-up areas and 250 m, 500 m, 1 km, 2 km (and nationally 4 km and 8 km) where fewer people live. The CRS is L-EST97 / EPSG:3301 (confirmed from the GeoPackage metadata). Two extents are used:
| Extent | File | Cells | Used for |
|---|---|---|---|
| Tallinn area — Tallinn and a ring of surrounding suburbs up to about 8 km beyond the city limits (24.44–25.01 E, 59.27–59.53 N) | acessiblity tallinn.gpkg (+ acessiblity tallinn.csv, join by fid) | 5,966 (125 m–2 km) | every Tallinn-area map and number in chapters 00–06 and 08 |
| Estonia | AGRD_EE_v6B-Grid_results_20260531_1235.geojson | 46,213 (125 m–8 km) | chapter 07 and the explorer’s “All Estonia” (the QGIS layers “Aca”, “Team 1”, “Team 2”) |
Each extent is scored within itself, so the same square has different scores in the two files (chapter 07). Each POI-…_W15 field is a continuous “within-15-minute-walk” accessibility value (0 to about 108 in the Tallinn area; the commonest non-zero values are whole numbers such as 1, 2 and 3, which suggests a distance-weighted count). Its exact definition is unconfirmed (see limitations); the model treats it as an interval-scaled access intensity. Dwelling units (DUs_Sum) give the demand measure. In the delivered GeoPackage and GeoJSON all dwelling and POI fields are numeric (only GRD_500m and GRD_1km are text, see below); field names contain Estonian diacritics (ä, õ, ü, š) and parentheses. When the layer was first loaded as delimited text, some numeric fields were read as text (see Implementation in QGIS).
Empty fields in the delivered grid (v6B). Eight scored attributes are 0 in every cell of both files, so they contribute nothing to any score: the four outdoor-sport fields (Välisport: Staadion, Tenniseplats, Välispalliväljak, Välisujula), the three indoor-sport fields (Sisesport, Jõusaal, Ujula) and basic school (Põhikool). GRD_500m and GRD_1km hold only “-”, so surrounding density is not yet available. Worth checking with the data provider whether these were dropped in v6B.
Normalisation — as used for the published scores
Each term is the sum of its member attributes per cell, divided by that term’s 95th percentile across the extent and capped at 1. A robust benchmark keeps a handful of extreme cells from flattening everyone else — the mitigation this report recommends for extent-relative scaling. The weighted sum × 100 gives the raw score; the published score stretches the raw score so that the 1st percentile maps to 0 and the 99th to 100, clamped.
D̂ homes, K̂ children’s places, Ĝ access to nature (inverted as 1 − Ĝ), Û everyday services (“walkable context”), Ê education, F̂ food, Ŝ services, L̂ movement (liikumine), Ĥ interests (huvid).
| Benchmark (P95) | Homes | Children | Nature access | Services | Stretch P1–P99 (Fair-Share raw) |
|---|---|---|---|---|---|
| Tallinn area (re-run) | 210 | 58.96 | 13.30 | 57.31 | 22.74 – 73.36 |
| Estonia | 91 | 29.48 | 6.64 | 32.89 | 21.02 – 70.61 |
The Tallinn-area row is what re-running the script on its 5,966 cells gives. The scores stored in that file reproduce only to within ±0.6 points (cause unknown — perhaps an earlier file state); this page uses the stored scores as delivered. A re-run would give 833 core, 223 Fair-Share-only and 329 top-band squares instead of 831, 222 and 331. Re-running the script on the national file reproduces all 46,213 stored scores exactly.
The June 2026 report described division by each term’s maximum (observed maxima: dwellings 774, children 83.9, nature access 22.4, walkable context 114.9). The published scores use the 95th-percentile benchmark and the 1–99 stretch above (Regreening_all_scores.py, PCTL = 95, SCALE_TO_FULL = 'minmax').
Weighting — why these weights
The weights are not arbitrary; they encode a hierarchy the evidence supports, while the exact magnitudes remain an explicit, testable calibration.
Biodiversity deficit (0.35, the largest). The programme’s purpose is to add biodiversity where people lack it, so deprivation carries the heaviest weight. Two lines of evidence justify this: the health case — direct contact with environmental biodiversity measurably improves children’s immune regulation [1, 2] — and the equity case — under the 3-30-300 standard, fewer than 15% of residents across 862 European cities have adequate green access [4, 5].
Household demand (0.25) and children reach (0.20). GIS multi-criteria siting finds population a dominant determinant of where new green space delivers benefit (population density was the top AHP factor, weight 0.357, in [7]; equity-based park siting also weighs supply against the population served [6]); demand is weighted second. Children are a distinct, high-value subset of that population — the health evidence is child-specific — but a complement to, not a replacement for, the population term. Together the two “people” terms (0.45) slightly exceed the deficit, so the model balances where the need is against how many people are reached: need takes the largest single weight, people the largest combined weight.
Walkable context (0.20). Green space is used and cared for best when embedded in active, walkable neighbourhoods; the 15-minute-city literature motivates rewarding integration into daily life, set equal to children as a supporting rather than leading factor.
Status of the weights. The literature supports the direction of the four terms; the precise values (35/25/20/20) are an expert judgement by the research team, offered as a transparent hypothesis to be stress-tested with a sensitivity analysis or a formal AHP elicitation. The model is linear-additive and fully compensatory.
What drives the published Fair-Share score
Because the model is additive and the deficit term is highest where few people live, the score is driven mainly by the people and services terms. In the Tallinn area the published score correlates +0.78 with the homes term, +0.62 with services, +0.60 with children’s places and −0.20 with the missing-nature term (nationally +0.83 / +0.59 / +0.58 / −0.30).
| Fair-Share band (Tallinn area) | Squares | Median homes | Median missing-nature level | No homes |
|---|---|---|---|---|
| 80–100 | 331 | 202 | 49 | 0% |
| 60–80 | 631 | 130 | 50 | 3% |
| 40–60 | 1,120 | 58 | 65 | 10% |
| 20–40 | 2,889 | 10 | 92 | 27% |
| 0–20 | 995 | 8 | 54 | 34% |
The median missing-nature level across all squares is 70. The shared core of 831 has a median of 40 and no square without nature access; the 222 Fair-Share-only squares have 77 (27 of them with none); the 368 team-only squares have 10. The deficit term therefore acts mainly at the margin — which is exactly where the Fair-Share view differs from the teams. If the programme’s intent is “plant where nature is missing”, a multiplicative form or a demand-gated deficit (PRESENCE_GATE) would make that term decisive; see limitations.
Definitions and scoring detail
- “Homes / demand” = dwelling units, not buildings. DUs are individual dwellings (flats); one apartment block holds many. The demand term uses
DUs_Sum, the dwellings within each cell — the cell’s own count, not a walking catchment. The data contain no counts of children; the children term measures access to places for children. - Catchment. The children, nature and walkable-context terms use the
_W15fields — destinations reachable within a 15-minute walk. - What the nature term contains — and excludes. Only outdoor-nature and recreation access: walking routes, dog-walking, nature destinations, swimming and bathing, theme park, and outdoor sport (stadium, tennis, ball courts, outdoor pool — empty in v6B). Ambulance (Kiirabi), car rental (Autorent) and fuel (Kütus) enter neither the nature term nor the Fair-Share formula. There is no vegetation or park-area input.
- Substitution inside a term. Each term is an additive sum across destination types, so it is fully compensatory — many childcares can offset the absence of a school; no “at least one of each type” rule is applied. If diversity should be required, count distinct types present, cap each subtype, or use a geometric mean.
- Workshop process. Teams 1 and 2 defined their priority categories during the session; the scores were computed afterwards by mapping those categories onto the available attributes (equal weights, presence-based). The teams did not see or tune numeric scores. Team 2’s movement term uses the same destinations as the nature term (minus the theme park), scored the other way round.
What the method produces
For each cell: a 0–100 score per index and the Fair-Share per-term points (c_*); a map class and a dominant driver can be derived. The actionable outputs are (1) the prioritised maps; (2) each index’s top-20% shortlist — 1,194 Tallinn-area cells for Fair-Share; and (3) the consensus shortlist of 831 cells that all three indices rank top-priority, recommended as the first wave of action.
| Tallinn area · top-20% shortlists | Cells |
|---|---|
| On all three shortlists (consensus core) | 831 |
| Fair-Share + one team only | 141 (133 with Team 1, 8 with Team 2) |
| Fair-Share only (“the decisive 222”) | 222 (some lie just outside the city limits, e.g. around Peetri in Rae Parish and Laagri in Saue Parish) |
| Teams only, not Fair-Share | 368 (222 both teams, 13 Team 1 only, 133 Team 2 only) |
| Correlation | Pearson: Team 1 ↔ Team 2 0.96 · Fair-Share ↔ Team 1 0.83 · Fair-Share ↔ Team 2 0.68 — rank (Spearman): 0.96 · 0.70 · 0.53 |
Classification and diagnostics
Maps use five fixed 20-point bands (0–20, 20–40, 40–60, 60–80, 80–100; QGIS graduated renderer, a value on a boundary falls in the lower band). Shortlists use the quintile rule of Regreening_tiers.py: a cell is shortlisted when its score is at or above the interpolated 80th percentile (Tallinn-area cut-offs: Fair-Share 54.6, Team 1 46.6, Team 2 49.0), which is why a shortlist can hold a few more than 20% of cells when scores tie. Quantile shortlists are preferred to fixed cut-offs because cells without homes carry a deficit baseline — a raw 35 points, published around 24 in the Tallinn area.
Because the score is additive it can be decomposed into the points each term contributes (c_demand 0–25, c_children 0–20, c_greendef 0–35, c_context 0–20). With the 1–99 stretch these sum to regreen_raw, not to the published score.
Colours. The web maps keep the QGIS meaning (orange → yellow → green) with deeper steps, so that neighbouring bands stay distinguishable over a street map and for colour-blind readers.
Implementation in QGIS
An initial single Field Calculator expression returned NULL for every feature. Three compounding causes were identified: the source layer was read-only (delimited text / WKT); numeric fields stored as text made arithmetic evaluate to NULL; and field references with Estonian diacritics did not resolve inside a nested-aggregate expression. The model was re-implemented as PyQGIS scripts that match field names via Unicode NFC normalisation, cast values tolerantly (treating “-”, blanks and text as 0), compute the benchmarks in Python, and write the results through the data provider. Each script prints its settings, feature counts, score range and any unmatched field names.
| Output field | Type | Meaning |
|---|---|---|
regreen_score / team1_score / team2_score | Double | Published index, 0–100 (stretched) |
regreen_raw / team1_raw / team2_raw | Double | Weighted sum before the stretch |
c_demand / c_children / c_greendef / c_context | Double | Points contributed by each Fair-Share term (sum = raw) |
regreen_class / regreen_tier | Integer / Text | Quintile tier 1–5 and its label — written only when Regreening_tiers.py is run; not in the delivered files |
top_driver / top_share | Text / Double | Largest-contributing term and its share — written only by Regreening_contributions.py; not in the delivered files |
Assumptions and limitations
| Issue | Effect on results | Recommended mitigation |
|---|---|---|
| Green access is a proxy | No vegetation / NDVI or park-area input; access to mapped outdoor destinations stands in for “biodiversity”. Forests and fields without mapped destinations read as “nature missing”. | Cross-check the deficit term against NDVI or land cover; consider adding a vegetation index. |
| Empty source fields | Eight POI fields (outdoor sport ×4, indoor sport ×3, basic school) are 0 everywhere; GRD_500m / GRD_1km are empty. | Ask the data provider whether they were dropped in v6B; re-run once restored. |
| Deficit saturates in empty land | Where no destination is reachable, the full 0.35 term is awarded regardless of population, lifting remote or unpopulated cells. | Gate the score on dwellings > 0, or multiply the deficit term by a demand factor (the scripts’ PRESENCE_GATE option). |
| Fully compensatory model | Linear addition lets strong people and services terms outweigh the deficit: the top band has more nature access than average (see “What drives the published score”). | Test a multiplicative or geometric-mean variant; always report term values beside the score. |
| Extent-relative normalisation | Scores compare only within their extent; Tallinn-area and national scores differ for the same cell. | Partly addressed by the 95th-percentile benchmark; use fixed reference values for cross-study comparison. |
| Expert weighting | The 0.25 / 0.20 / 0.35 / 0.20 weights are subjective. | Sensitivity analysis (± weight perturbation) and/or AHP or stakeholder calibration. |
| Accessibility metric semantics | The exact definition of the W15 values (gravity/decay vs count) is unconfirmed. | Confirm the metric and its units with the data provider. |
| MAUP / grid effects | Results depend on cell size and grid origin; sub-cell variation is hidden. A 125 m cell is a whole block, not a site. | Test sensitivity to resolution; validate against parcel-level data. |
| Playground = demand | Playgrounds are treated as a children-presence signal, not green supply. | Confirm with stakeholders; optionally test as partial green supply. |
Quality assurance
- Additivity: the four contribution fields sum to
regreen_raw(by construction; within 0.2 after rounding). - Field resolution: each script prints any field names it could not match, catching schema or encoding drift.
- Distribution review: inspect the score histogram, the printed percentile summary (
Regreening_all_scores.py) and the quintile breaks (Regreening_tiers.py) for a sensible spread. - Reproduction: re-running the scoring on the national file reproduces all 46,213 stored scores exactly; the Tallinn-area scores reproduce to within 0.6 points, and the statistics in chapter 06 are computed from the stored scores.
- Ground-truthing: look inside top-tier cells on recent aerial imagery (the explorer’s “Aerial” background) for plantable space that is not already vegetated.
- Sensitivity: re-run with perturbed weights (the explorer’s “Your mix”) and a demand-gated deficit term; compare the stability of the top quintile.
What changed from the June 2026 report
- Reading of the bands. The June report read deep green as “dense, child-rich areas that lack nearby nature” and the lower bands as “services are the gap”. The published scores do not support either: the top band has more nature access than average, and lower scores mainly reflect fewer homes (table above). The story now describes the bands by who would be reached; the original prioritisation matrix wording is kept below for reference.
- Study area. The “Tallinn” file covers Tallinn and a ring of surrounding suburbs and towns (e.g. Saue, Laagri, Saku, Kiili, Peetri, Maardu); it is called the Tallinn area here.
- Empty fields. Sports grounds, indoor sport and basic schools are named in the baskets but hold no data (see above).
- Population weight. The 0.36 population weight comes from the Environmental Systems Research 2024 AHP study [7] (0.357), not from the Frontiers study [6].
- Normalisation, outputs and QA are described as actually implemented (P95 benchmark and 1–99 stretch;
c_*sum to raw; tier and driver fields not delivered).
The June report’s prioritisation matrix (original wording)
| Colour | Original label | Original description | Action |
|---|---|---|---|
| Deep green · highest | 1 · Add greenery — top priority | Dense, child-rich areas that lack nearby nature but already have everyday services within reach. | Plant first |
| Light green · high | 2 · Add greenery — strong case | A clear greening gap with plenty of people to benefit. | Plant / pipeline |
| Light yellow · lower | 3 · Greenery helps — but services are the gap | These areas score lower mainly because everyday services are sparse — not because nature isn’t wanted. | Pair greening with services |
| Orange · lowest | 4 · Services first — greenery is secondary | The dominant shortfall is missing everyday services (and often few residents). | Address services; green opportunistically |
The June report’s comparison of the three narratives
| If you follow… | Fair-Share | Active city (Team 1) | Everyday life (Team 2) |
|---|---|---|---|
| Overlap with the others | shares the 831-square core; adds 222 unique squares | shares 88% of squares with Team 2; 80% also on Fair-Share’s list | shares 88% with Team 1; 70% also on Fair-Share’s list |
| Green goes to… | the shared dense core, plus quieter family areas missing nature | dense, central, high-traffic blocks | complete neighbourhoods with services & leisure |
| Who gains most | children & residents now without nature | the many people already in busy areas | residents enjoying daily amenities |
| Headline win | fairness & child-health equity | visibility & footfall | everyday quality of life |
| Main risk | some chosen sites are less prominent | green-poor areas stay green-poor | the same — under-served areas overlooked |
| City it builds | a more equal green city | a more concentrated green city | a more liveable green city |
Team 2 · categories matched to the data
| Team category | Matched to | Status |
|---|---|---|
| Home (kodu) | Homes (dwellings) | Matched |
| Teenused – TOIT (food) | Grocery, market, eatery, bar | Matched |
| Liikumine (movement) | Walking, dog-walking, nature destinations, swimming (the sport fields are empty in the data) | Partial · scored as presence (more = higher) |
| Huvid (interests) | Hobby, hobby centre, playroom, cinema, museum, theatre, gallery, culture centre | Matched |
| Haridus (education) | Kindergarten, childcare, upper-secondary, higher, vocational & continuing education (the basic-school field is empty) | Matched |
Full attribute reference — every attribute, scored or not
| Category | Dataset attributes (Estonian original in brackets) | Role in the Fair-Share score |
|---|---|---|
| Children & learning | Kindergarten (Lasteaed) · Childcare (Lastehoid) · Basic school (Põhikool — empty) · Upper-secondary (Gümnaasium) · Higher ed (Kõrgkool) · Vocational (Kutseõppeasutus) · Continuing ed (Täienduskoolitus) · Playground (Mänguväljak) · Kids’ hobby & playroom (Huvi, Mängutuba, Huvikeskus) · Children’s shop (Pood-Lapsed) · residents DUs_Count / DUs_Sum | Children’s places ↑ (kindergarten, childcare, basic & upper-secondary school, playground, hobby, children’s shop) + homes (DUs) |
| Health & care | Pharmacy (Apteek) · GP (Perearst) · Specialist (Eriarst) · Dentist (Hambaarst) · Ambulance (Kiirabi) · Care home (Hooldekodu) · Day centre (KOV-Päevakeskus) · Health service (Teenus-Tervis) | Everyday services ↑ — pharmacy and GP only |
| Nature & recreation | Walking routes (Liikumine-Jalutamine) · Dog walking (Koduloomad) · Nature destination (Sihtkoht) · Swimming (Suplus) · Theme park (Teemapark) · Outdoor sport (Välisport: staadion, tennis, ballcourt, pool — empty) | Missing nature ↓ — scored inverted, the core |
| Everyday & food | Grocery (Toidupood) · Market (Turg) · Daily goods (Tarbekaubad) · Eatery (Toidukoht) · Bar / café / pub (Toidukoht-Bar) · Post (Post) | Everyday services ↑ — grocery, daily goods, eatery, post |
| Movement & access | Transit terminal (ÜT-terminal) · Bike parking & rental (Rattaparkla, Rattarent) · Car rental (Autorent) · Fuel (Kütus) | Everyday services ↑ — transit terminal only |
| Community & culture | Local government (KOV) · County centre (Maakonnakeskus) · Church (Kirik) · Culture centre (Kultuurikeskus) · Public library (Raamatukogu) · Cinema (Kino) · Museum (Muuseum) · Theatre (Teater) · Gallery (Galerii) · Hall & stage (Saal, Lava) · Nightclub (Ööklubi) · Police (Politsei) · Fire/rescue (Päästekomando) · Civil shelter (Varjumiskoht) · Cemetery (Kalmistu) · Public toilet & waste (WC, Konteinerid) | Everyday services ↑ — public library only |
| Retail & services | Clothing & textile · Electronics · Furnishings · Books, stationery, music · Plants · Hardware · Sport & bikes · Hobby, pets, vehicles, optics · Beauty & care · Gym & indoor sport (empty) · Home services · Hotel | — (Team 1 “services” and Team 2 “movement” use some of these) |
The role labels describe what enters the Fair-Share formula. Ambulance, dentist, specialist and care home are not scored; car rental and fuel are not part of the nature term.
Companion files and data
Regreening_Methodology_Report.docx— full technical reportRegreening_all_scores.py— all three indices and thec_*per-term points in one run (the published values)Regreening_tiers.py— quintile tiers and the shortlist ruleRegreening_contributions.py— dominant driver; still max-normalised, so do not run it on the published files (it overwritesc_*)Regreening_QGIS_FieldCalculator.txt— Field Calculator expressions with troubleshooting notesacessiblity tallinn.csv— per-cell scores for the Tallinn area (join byfid)
These files are available on request from the EKA EDITUA team (siim.tuksam@artun.ee). The per-square scores behind every map on this page are published as data/tallinn.json and data/estonia.json.
References
- Roslund MI, et al. Biodiversity intervention enhances immune regulation and health-associated commensal microbiota among daycare children. Science Advances. 2020;6(42):eaba2578. doi:10.1126/sciadv.aba2578
- Hanski I, et al. Environmental biodiversity, human microbiota, and allergy are interrelated. PNAS. 2012;109(21):8334–8339. doi:10.1073/pnas.1205624109
- 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;1(3):e180298. PMC6324526
- Konijnendijk CC. Evidence-based guidelines for greener, healthier, more resilient neighbourhoods: introducing the 3-30-300 rule. Journal of Forestry Research. 2023;34:821–830. PubMed 36042873
- Assessing European cities with the 3-30-300 rule underscores the need for enhanced urban greening efforts. Nature Communications. 2026. nature.com
- Optimal site selection strategies for urban parks green spaces under the joint perspective of spatial equity and social equity. Frontiers in Public Health. 2024;12:1310340. frontiersin.org
- Identification of optimal locations for green space initiatives through GIS-based multi-criteria analysis and the analytical hierarchy process. Environmental Systems Research. 2024. doi:10.1186/s40068-024-00377-0