Build the case for
your next tree investment.

Use it to

Turn a citywide canopy and heat map into a focused conversation about where to invest. A city team can shortlist areas for field assessment, compare policy priorities, and estimate the scale of a street-tree program in one place.

Why this approach

We bring canopy, residents and home-cooling context onto the same grid so staff can compare areas without assembling separate maps. The workflow connects a priority to its evidence and a planting scenario.

Planning note Use the shortlist to scope an assessment; confirm site feasibility and local costs before committing funds.

Use it to

Build a first-round investment shortlist that accounts for both missing shade and the people living with it. Open any area to explain the drivers behind its position in a budget or program-scoping discussion.

Why this approach

An additive score makes each input’s contribution inspectable. Surface heat carries the largest default share because San Ramon is one valley climate where shade measurably cools the ground (canopy and surface heat correlate at about −0.8); missing canopy and age add resident context. A/C is viewable but not scored: the city is uniformly affluent and its A/C prevalence is high and narrow. Population adjusts potential reach. These are explicit project policy choices, not externally prescribed weights.

Planning note Scores compare residential cells within this city; they are not health-risk probabilities or cross-city rankings.

Rank stability A rank is an estimate with a band, not an exact position. 05b_stability.py re-draws each block group’s population and residents 65+ and each tract’s A/C estimate inside their published 90% margins of error (neighbouring areas cut from the same block group move together), jitters each non-zero recommended weight by ±10 points, and re-ranks; 300 draws. The detail panel shows each area’s 5th–95th percentile rank and how often it lands in the top tier (best 42 of 419). In this build the median band is 50 ranks wide, and of the 42 top-tier areas 67% stay top-tier in at least 80% of draws. Surface temperature and tree cover are held fixed (measurement error, no per-area margin published), as is the allocation of residents within a block group. Treat rank differences smaller than an area’s band as not established (the bands are sensitivity ranges under stated assumptions, not confidence intervals or a pairwise test); use the top-tier share to shortlist.

Use it to

Test whether a proposed focus area remains a priority when staff emphasize shade, older residents or access to cooling. Use presets to structure a policy discussion instead of treating one ranking as the only answer.

Why this approach

Editable weights expose the value judgments behind a composite index. The JRC’s guidance supports examining rankings under alternative assumptions; it does not endorse our particular presets. Normalizing weights keeps their relative contribution explicit.

Count people more (the population multiplier)

The four condition sliders say what counts; this slider says how much the number of residents counts. Each area first gets a need figure from its conditions. That figure is then multiplied by a factor that runs from (1 − P) for the least-populated area up to 1 for the most-populated, where P is the slider. At P = 0% every area is judged on conditions alone. At the recommended 45%, the most-populated area counts about 1.8× the least. At 100%, an area with almost no residents scores near zero whatever its heat or shade, because there is almost nobody there to help. Population is normalized between its minimum and the 98th percentile of ranked areas (risk inputs run from their 2nd percentile to their maximum; protective inputs from their minimum to their 98th percentile), so one very crowded area does not flatten everyone else.

Planning note Presets represent alternative priorities. They change rankings, not planting costs.

Use it to

See the outdoor thermal context around proposed investments and compare it with canopy gaps. The separate heat view helps staff distinguish a shade-priority question from a surface-heat question.

Why this approach

Landsat provides a repeatable, spatially consistent thermal record. Summer medians reduce dependence on a single scene. Heat carries the largest default weight here because San Ramon is one valley climate, so within-city surface-temperature comparisons compare like with like. Contra Costa County, by contrast, leaves heat unscored.

Planning note Surface temperature is not air temperature or a live heat alert. Current scenes cover summers 2022–2024.

Use it to

Locate shade deficits and establish a baseline for scoping a canopy assessment. Staff can compare canopy coverage across candidate areas before choosing where to investigate street-tree opportunities.

Why this approach

The USFS/CAL FIRE aerial canopy classification targets tree crowns, which fits a shade-investment question more directly than general greenness. Equal-area aggregation preserves ground-area proportions; the assessed urban boundary defines the denominator. Validation. All 419 ranked areas were compared with the CDFW 2025 lidar vegetation map (returns above 4.6 m): the aerial product tracks it at a rank correlation of 0.91. Eight areas the aerial product could not assess use the lidar canopy directly (“trees · lidar”); no area uses a greenness stand-in.

Planning note Canopy coverage uses aerial and height-map sources; 2 ranked cells on the fringe rely on the legacy height model, labeled in cell details. Percentages describe assessed cover, not individual trees or verified planting space.

Use it to

Bring the scale of resident need into investment decisions. Population helps distinguish a sparsely occupied canopy gap from an area where a similar gap affects the surroundings of many more residents.

Why this approach

ACS five-year data supplies population, age and income on a consistent small-area geography. In San Ramon, residents are placed by mapped residential floor area (dasymetric weighting): OSM building coverage is 0.68 buildings per housing unit across the block groups touching the city, and the lowest-income quartile’s lower building count reflects apartment-heavy housing rather than missing mapping.

Planning note Cell populations are allocated estimates, not household counts. Published ACS margins of error are carried into the likely-rank bands (see Rank stability), which hold heat, tree cover and walking time fixed and spread each block group’s or tract’s re-draw over cells by whole-cell overlap area, an approximation of the point model’s boundary-clipped, building-weighted allocation; the bands are sensitivity ranges under those assumptions, not confidence intervals.

Use it to

Bring older residents into the shortlist for shade investment and outreach planning. The seniors preset lets staff see how prioritizing this population changes the areas they would investigate first.

Why this approach

CDC identifies adults aged 65+ as more susceptible to heat-related problems, supporting age as one input. The project smooths small-cell percentages to avoid unstable extremes; the 100-person smoothing strength is a design choice, not a CDC recommendation.

Planning note Shares are smoothed area estimates. They do not identify individual residents or their health needs.

Use it to

Identify areas where limited home cooling adds a reason to investigate shade and public cooling access. This gives the investment discussion a housing dimension alongside canopy and population.

Why this approach

Census LACE combines AHS air-conditioning information with the ACS’s geographic coverage to estimate prevalence at tract scale. We chose it to use a documented, cooling-specific national product instead of treating local income rank as A/C availability.

Planning note Experimental tract estimates, weighted by occupied homes. It does not establish working or affordable A/C in a particular home.

Use it to

Screen for areas where residents may face a long walk to a potential cooling facility. Use the map to focus a facility-verification exercise or a discussion about where additional relief may be needed.

Why this approach

OSM provides facility locations and a pedestrian graph in a common, reproducible source. Shortest-path routing represents mapped street connections more realistically than straight-line distance alone. This is a facility-access screen, not an emergency service directory.

Planning note Verify designation, hours, fees and accessibility. Category filters change markers only; walking estimates remain based on all mapped sites.

County-listed cooling centers and contacts ↗

The county EHSD bulletin lists 17 designated locations and asks visitors to call for hours. San Ramon’s listed location is the Alcosta Senior and Community Center. Its June 2026 revision says El Cerrito and Kensington libraries lack A/C; that exclusion rule is applied but touches no San Ramon site. The directory is not an open-now feed. access_snap_m records the straight-line cell-to-network connection; connections over 100 m are flagged for approach review, without implying that shorter approaches are verified. Walking estimates use all 25 discovery sites, not just the official directory.

Use it to

Translate a priority area into a concrete program-scale question: what would planting a quarter, half or all of the mapped street capacity involve? Compare tree counts, canopy gain and indicative cost before commissioning site surveys.

Why this approach

Street length connects the scenario to a physical intervention that a public-works team can investigate. Both sides at 10 m spacing gives a transparent, repeatable capacity assumption. It is a project scenario convention, not a verified spacing requirement for every street.

Planning note Capacity is theoretical; confirm ownership, existing trees, utilities and space. Partial and unknown canopy coverage allow conditional scenarios; they do not establish a reliable future total or verified planting capacity.

Use it to

Put an initial scale of investment beside the shortlist. A shared unit-cost and crown-area assumption lets staff compare scenarios consistently and prepare questions for arborists and procurement teams.

Why this approach

A linear model makes every dollar and canopy increment traceable to the tree count. The cost per tree is a band, not a point: no San Ramon contract price is published online, so the app offers three sourced presets and lets you type your own. The 40 m² mature crown is a planning assumption.

Planning note Indicative cost at assumed maturity; not a bid, lifecycle budget or financial ROI. Growth and crown overlap are not modeled; survival is an assumed share (editable beside the cost, default 70%), not a predicted outcome.

Use it to

Connect tree investment to cooler streets and shade where people walk and wait. Added crowns can reduce direct sun and surface heating; the air-cooling number describes a different, area-wide effect and does not capture the full benefit of standing in shade.

Why this approach

A review of urban cooling simulations estimates about 0.3 °C of afternoon air cooling per 10 percentage points of canopy under clear-sky summer conditions. We use that relationship for an illustrative scenario, separate from the satellite surface-temperature layer.

Planning note Literature-based illustration at maturity, not a locally calibrated forecast or a count of residents cooled.

Use it to

Compare the same geographic units across all map layers. A consistent grid helps staff move from a city overview to an identifiable area for follow-up without changing the unit of analysis.

Why this approach

H3 cells give each area a stable identifier and allow different source geographies to be combined. Keeping developed and undeveloped cells visible preserves geographic context while residential-only ranking keeps the shortlist aligned with resident priorities.

Planning note Cells are not census blocks or neighborhood boundaries. Place names are approximate; gray means unranked.

Use it to

Frame an equity discussion using canopy and resident context, then inspect the candidate areas directly. The figure below is recomputed from the current build; it is an exposure count, not an equity comparison.

Why this approach

The figure counts estimated residents by the canopy of the cell they live in, independently of the composite score. It uses no income data, so it does not by itself establish an equity gap; San Ramon’s canopy story is uneven cover, not thin cover.

Planning note The exposure count is an aerial-assessed figure for this build. No income-equity claim is made for San Ramon in this build.

Use it to

Keep historical context available where the data supports it, without distracting the current investment workflow with an empty layer.

Why this approach

The application checks for actual HOLC grades before offering the control. Mapping Inequality is the documented historical source; San Ramon has no digitised HOLC map there, so this build has no graded cells.

Planning note No HOLC layer does not imply no history of discrimination.

Use it to

Move from a quick shortlist to detailed exploration at your own pace. Switch layers to inspect evidence, adjust weights to test priorities, and change the planting share to scope an intervention.

Why this approach

One panel holds every control, in numbered steps, so a quick shortlist and a detailed exploration use the same screen. Units and basemaps change presentation only. Contextual guide links preserve the map in its original tab, so checking a method does not interrupt the scenario.

Planning note Your mix of priorities and the map layer are kept for this browser tab: a reload lands where you were, and the home button returns to the start screen without resetting them. A new tab starts fresh at the start screen. Planting shares last for the current page session.

Use it to

Read the same ranked list from a different angle: the most people first, the hottest ground first, the least tree cover first, the longest walk first. Filters narrow the list to areas that meet a threshold; search finds one area by name or cell number.

Why this approach

The rank badge is the answer to the question set under What matters most, so it stays fixed. Sorting is a reading aid, which is why the count line says what the list is sorted by and in which direction whenever it is not in rank order.

Planning note A sort is not a ranking. Use What matters most to change what the score rewards; use sort and filters to find areas within the ranking you already have.

Use it to

Share a shortlist or one area with colleagues, keep a record of the weights and planting share behind a recommendation, and reopen the same view. A CSV opens in a spreadsheet; a JSON file reloads into this map. Nothing is uploaded: files are generated in the browser and saved on your computer.

Ranked list — the Export & share section

Save scenario (JSON) saves the ranked list with the full scenario header, for reopening. All ranked areas (CSV) saves every ranked cell in the order the current weights put them, one row per cell, with plain column names (see the column key below). Open a saved scenario… reads a scenario file back in and restores the weights, population weighting, any planting shares you set and the selected area; the status line under the buttons says what was loaded and whether the data version matches.

Selected area — in the area panel

Save this area (JSON) and Spreadsheet row (CSV) save one record for the cell you have open: its metrics, where it ranks under the weights in force and under the shipped defaults, the sources and coverage behind each number, and the planting scenario at the share you chose (trees, canopy gain in percentage points, indicative cost, illustrative cooling). Reloading an area file restores the weights and reopens that area.

Why this approach

A ranking only means something with its weights, assumptions and data version attached, so every export carries them: the weights in force, the population multiplier, the preset if one matches, the planting constants, per-cell source and coverage flags, and the SHA-256 fingerprint of the data file that was loaded. Reloading says plainly if the data version differs.

Column key

Spreadsheet columns use the plain names in the first column; scenario (JSON) files use the internal keys in the second and carry this key as field_key.

Planning note An export documents an exploratory scenario. It does not verify planting sites, costs or facility status, and it carries the same limitations as the map it came from.

Field checks Each ranked area has a Field check box in its detail panel: a status (To visit; Checked · feasible, partly feasible or not feasible; Planted) and a free-text note. They are saved in the browser you typed them in, per city, and are never uploaded. Every area and shortlist export carries them as field_check_status, field_check_note and field_check_updated, and the printed briefing shows them under each shortlisted area, so a colleague can see what was verified on the ground and what was seen. Clearing the browser’s site data clears them; export first.

Use it to

Compare living vegetation with tree cover to identify areas worth investigating for shade. A greener area can contain grass or shrubs and still need trees; the two layers help staff ask that question without confusing greenness with canopy.

Why this approach

Satellite greenness supplies broad vegetation context across the grid, including outside aerial canopy coverage. It is publicly available in Explore, leaving the default priority workflow focused. The active planting calculator uses mapped street length, not the buffered right-of-way area fields.

Planning note Greenness does not establish percent cover or planting feasibility; exported street-buffer fields do not establish ownership or plantability.

Use it to

Answer “how many trees are we planting in District 2, what does it cost, and how many residents does it reach” in one row, under whatever priorities are in force and with the areas you starred.

Why this approach

San Ramon publishes its four council districts on the city GIS service; the app rolls the ranking up by district and can outline them on the map. The city has elected council members by district since November 2020, with lines redrawn after the 2020 census. Member names are fetched with the boundaries and dated; they change at elections. Each area belongs to the district containing its centre; areas outside every district (county islands, unincorporated land, edges) are grouped as such rather than dropped, because their residents are in the ranking. Need per district is the mean of its areas’ 0–100 index; “top 100” counts its areas among the 100 highest. Trees and cost add up the starred areas at their own planting shares.

Planning note Districts are political boundaries. The need lives where it lives; an even split across districts is a policy choice, not a finding. Boundaries and member names are re-fetched by the pipeline, never edited by hand, and SOURCES.md in the data folder records the fetch.

Use it to

See need and feasibility together. An area that needs shade and has plenty of street room is an early win; one that needs shade and has little room needs parks, schoolyards or private planting instead, and should be planned differently rather than dropped.

Why this approach

Street room is the number of tree spots that the mapped street centrelines inside the study boundary could hold at 10 m spacing on both sides. It is the same ceiling the planting scenario starts from, shown directly. Streets come from OpenStreetMap; utilities, driveways, sidewalk width and existing trees are not known, so it is an upper bound, not a surveyed count. The words “little”, “some” and “plenty” are thirds of this study area’s ranked areas that have any mapped street; “no mapped street” means none inside the boundary. On the map, deeper colour means less room, keeping the rule that deeper always means more need.

Planning note Street room is a planning screen for where to look first, not a count of plantable sites. A city’s own right-of-way, utility and sidewalk layers turn it into a planting plan; ask public works for them.