Guelph Trees Layer

Turns the City of Guelph Trees inventory -- the street and park trees the City owns or shares on a property line, 42,712 on 2026-10-09 -- into map-tile layers that OpenStreetMap mappers can add to the JOSM and iD editors as a reference overlay, plus a gap page showing where OSM has no trees at all and which mapped trees lack the species and girth the City records.

Live layer and how to add it: https://guelph-maps.github.io/guelph-trees-layer/

Part of the guelph-maps organisation. It is a clone of guelph-bus-stops-layer (live): the same download -> slim -> compare -> vector -> raster -> site -> publish pipeline, the same ArcGIS service, the same Overpass, link-gate and publish rules. What differs follows from the scale and from what a tree carries:

• slim writes names the way OSM does and the girth in OSM's unit;
• compare matches by distance alone, and rolls the result up into 250 m cells, because 34,000 unmatched trees are not a work list;
• raster draws crowns, coloured by genus and sized from the trunk.

Why build it at all: the import audit found OSM's ~50,000 Guelph trees are almost all one mapper's hand-tracing of bare natural=tree -- genus on 3, species on 12 -- and that only 19% of the City's trees have an OSM tree within 5 m. Half the city has none.

What it produces

• Vector tiles (MVT, z14-19) — interactive in iD; click a tree to read its tags (below).
• Raster tiles (PNG, z13-18) — a crown per tree, coloured by genus, drawn to an approximate crown size from the trunk diameter at z16-18. A readable backdrop for JOSM.
• A landing page with copy-paste "add this layer" instructions for both editors and the colour legend.
• A gap-review page (/gaps/): a map of 250 m cells and a table of them, drilling into one cell's trees, and its history page (/gaps/history.html) charting the counts, one point per weekly run. Each cell links to JOSM for its square, and each tree to openstreetmap.org, iD and JOSM (needs JOSM running with Preferences → Remote Control enabled).

What a tree carries

Key	Example	Is it an OSM tag?
genus	Quercus	yes
species	Quercus rubra, Platanus x acerifolia	yes
circumference	0.94	yes -- metres, DBH × π / 100
common_name	Red Oak	no -- often a cultivar's name ("Green Mountain Maple" is a sugar maple), so not species:en
dbh_cm	30	no -- the City's trunk diameter, cm
planted	2014	no
owner	city, boundary	no
address	133 Clairfields Dr W	no -- the City's nearest address
city_asset_id	FT10001	no -- the City's inventory number. It is not on the tree, so it is not a ref
objectid	1	no -- the ArcGIS row, for finding it again

Names are converted once, in slim.taxon: capitals to OSM's case, hybrids with a plain x (Platanus x acerifolia is on ~21,000 OSM trees, the × form on none), varieties back to their species, and "MALUS SPECIES" or a quoted cultivar to the genus alone. Three City names cannot be right and are corrected: RHAMUS and GYMNOCLADUS DIOCUS are misspellings, and the 31 CONIUM MACULATUM rows -- poison hemlock, a herb -- are called "HEMLOCK", the eastern hemlock's name, so they get genus=Tsuga and no species. A DBH over 300 cm (one row says 879) gets no circumference.

A word on copying. Individual-tree imports are contested in OSM, and copying species or girth off this overlay is, in effect, an attribute import. The City licence is compatible (below), but compatible is not consensus. The layer is a reference: add a tree you can see on imagery, a species you have verified. The landing page and the gap page both say so.

The gap page

Overpass is asked for every natural=tree node in the city (nodes only -- a tree mapped as an area is a wood). Each City tree is matched to an OSM tree nearest-first and one-to-one within 5 m (MATCH_M): every pair within 5 m is sorted by distance and taken unless one side is spoken for. Street trees stand 8-10 m apart, often closer, so "nearest OSM tree for each City tree" would hand one node to two trees. The recalibration measured the gap at 3, 5 and 10 m and it barely moves; the radius is not doing the work.

Then the city is cut into 250 m cells (CELL_M), counted from the bounding box's south-west corner:

Cell status	Meaning
empty_cell	≥ 10 City trees (CELL_MIN_CITY) and no OSM tree at all
thin_cell	≥ 10 City trees, some OSM trees, but under half (THIN_SHARE) of the City's have one within 5 m
covered	the rest with ≥ 10 City trees
sparse	fewer than 10 City trees (off by default)

thin_cell is decided by the matched share, not the counts. OSM's trees include woodlots and back yards the City does not inventory, and a cell's counts can agree while its trees do not -- the trap the audit's first tree review fell into.

Inside a cell, per tree:

Tree status	Meaning
missing	no OSM tree within 5 m
nospecies	matched, but the OSM tree has no genus, species or taxon, and the City knows the genus
nogirth	matched, but the OSM tree has no circumference, and the City has a DBH

A matched tree can need both. There is no osm_only: the 40,000-odd OSM trees with no City tree beside them are woodlots, private land and campus, not mistakes.

The page loads cells.geojson (~900 squares) on open and trees.json (the per-tree rows, ~5 MB) only on the first drill-down, so it never draws 42,000 markers.

This is the beginning of the README. Read the rest at the original. Read the full README ↗