tippecanoe

Builds vector tilesets from large collections of GeoJSON features. This is a tool for making maps from huge datasets.

Intent

The goal of Tippecanoe is to enable making a scale-independent view of your data, so that at any level from the entire world to a single building, you can see the density and texture of the data rather than a simplification from dropping supposedly unimportant features or clustering or aggregating them.

If you give it all of OpenStreetMap and zoom out, it should give you back something that looks like "All Streets" rather than something that looks like an Interstate road atlas.

If you give it all the building footprints in Los Angeles and zoom out far enough that most individual buildings are no longer discernable, you should still be able to see the extent and variety of development in every neighborhood, not just the largest downtown buildings.

If you give it a collection of years of tweet locations, you should be able to see the shape and relative popularity of every point of interest and every significant travel corridor.

Installation

The easiest way to install tippecanoe on OSX is with Homebrew:

$ brew install tippecanoe

Usage

$ tippecanoe -o file.mbtiles [file.json ...]

If no files are specified, it reads GeoJSON from the standard input. If multiple files are specified, each is placed in its own layer.

The GeoJSON features need not be wrapped in a FeatureCollection. You can concatenate multiple GeoJSON features or files together, and it will parse out the features and ignore whatever other objects it encounters.

Options

Naming

• -l name: Layer name (default "file" if source is file.json or output is file.mbtiles). If there are multiple input files specified, the files are all merged into the single named layer.
• -n name: Human-readable name (default file.json)

File control

• -o file.mbtiles: Name the output file.
• -f: Delete the mbtiles file if it already exists instead of giving an error
• -F: Proceed (without deleting existing data) if the metadata or tiles table already exists or if metadata fields can't be set
• -t directory: Put the temporary files in directory.
• -P: Use multiple threads to read different parts of each input file at once. This will only work if the input is line-delimited JSON with each Feature on its own line, because it knows nothing of the top-level structure around the Features. Performance will be better if the input is a named file that can be mapped into memory rather than a stream that can only be read sequentially.

Zoom levels and resolution

• -z zoom: Maxzoom: the highest zoom level for which tiles are generated (default 14)
• -Z zoom: Minzoom: the lowest zoom level for which tiles are generated (default 0)
• -B zoom: Base zoom, the level at and above which all points are included in the tiles (default maxzoom). If you use -Bg, it will guess a zoom level that will keep at most 50,000 features in the densest tile. You can also specify a marker-width with -Bgwidth to allow fewer features in the densest tile to compensate for the larger marker, or -Bfnumber to allow at most number features in the densest tile.
• -d detail: Detail at max zoom level (default 12, for tile resolution of 4096)
• -D detail: Detail at lower zoom levels (default 12, for tile resolution of 4096)
• -m detail: Minimum detail that it will try if tiles are too big at regular detail (default 7)
• -b pixels: Buffer size where features are duplicated from adjacent tiles. Units are "screen pixels"--1/256th of the tile width or height. (default 5)

All internal math is done in terms of a 32-bit tile coordinate system, so 1/(2^32) of the size of Earth, or about 1cm, is the smallest distinguishable distance. If maxzoom + detail > 32, no additional resolution is obtained than by using a smaller maxzoom or detail.

Properties

• -x name: Exclude the named properties from all features
• -y name: Include the named properties in all features, excluding all those not explicitly named
• -X: Exclude all properties and encode only geometries

Point simplification

• -r rate: Rate at which dots are dropped at zoom levels below basezoom (default 2.5). If you use -rg, it will guess a drop rate that will keep at most 50,000 features in the densest tile. You can also specify a marker-width with -rgwidth to allow fewer features in the densest tile to compensate for the larger marker, or -rfnumber to allow at most number features in the densest tile.
• -g gamma: Rate at which especially dense dots are dropped (default 0, for no effect). A gamma of 2 reduces the number of dots less than a pixel apart to the square root of their original number.

Doing more

• -ac: Coalesce adjacent line and polygon features that have the same properties
• -ar: Try reversing the directions of lines to make them coalesce and compress better
• -ao: Reorder features to put ones with the same properties in sequence, to try to get them to coalesce
• -al: Let "dot" dropping at lower zooms apply to lines too

Doing less

• -ps: Don't simplify lines
• -pS: Don't simplify lines at maxzoom (but do simplify at lower zooms)
• -pf: Don't limit tiles to 200,000 features
• -pk: Don't limit tiles to 500K bytes
• -pd: Dynamically drop some fraction of features from large tiles to keep them under the 500K size limit. It will probably look ugly at the tile boundaries.
• -pi: Preserve the original input order of features as the drawing order instead of ordering geographically. (This is implemented as a restoration of the original order at the end, so that dot-dropping is still geographic, which means it also undoes -ao).
• -pp: Don't split complex polygons (over 700 vertices after simplification) into multiple features.
• -q: Work quietly instead of reporting progress

Example

$ tippecanoe -o alameda.mbtiles -l alameda -n "Alameda County from TIGER" -z13 tl_2014_06001_roads.json

$ cat tiger/tl_2014_*_roads.json | tippecanoe -o tiger.mbtiles -l roads -n "All TIGER roads, one zoom" -z12 -Z12 -d14 -x LINEARID -x RTTYP

GeoJSON extension

Tippecanoe defines a GeoJSON extension that you can use to specify the minimum and/or maximum zoom level at which an individual feature will be included in the vector tile dataset being produced. If you have a feature like this:

{
    "type" : "Feature",
    "tippecanoe" : { "maxzoom" : 9, "minzoom" : 4 },
    "properties" : { "FULLNAME" : "N Vasco Rd" },
    "geometry" : {
        "type" : "LineString",
        "coordinates" : [ [ -121.733350, 37.767671 ], [ -121.733600, 37.767483 ], [ -121.733131, 37.766952 ] ]
    }
}

with a tippecanoe object specifiying a maxzoom of 9 and a minzoom of 4, the feature will only appear in the vector tiles for zoom levels 4 through 9. Note that the tippecanoe object belongs to the Feature, not to its properties.

Point styling

To provide a consistent density gradient as you zoom, the Mapbox Studio style needs to be coordinated with the base zoom level and dot-dropping rate. You can use this shell script to calculate the appropriate marker-width at high zoom levels to match the fraction of dots that were dropped at low zoom levels.

If you used -B or -z to change the base zoom level or -r to change the dot-dropping rate, replace them in the basezoom and rate below.

awk 'BEGIN {
    dotsize = 2;    # up to you to decide
    basezoom = 14;  # tippecanoe -z 14
    rate = 2.5;     # tippecanoe -r 2.5

    print "  marker-line-width: 0;";
    print "  marker-ignore-placement: true;";
    print "  marker-allow-overlap: true;";
    print "  marker-width: " dotsize ";";
    for (i = basezoom + 1; i <= 22; i++) {
        print "  [zoom >= " i "] { marker-width: " (dotsize * exp(log(sqrt(rate)) * (i - basezoom))) "; }";
    }

    exit(0);
}'

Geometric simplifications

At every zoom level, line and polygon features are subjected to Douglas-Peucker simplification to the resolution of the tile.

For point features, it drops 1/2.5 of the dots for each zoom level above the point base zoom (which is normally the same as the -z max zoom, but can be a different zoom specified with -B if you have precise but sparse data). I don't know why 2.5 is the appropriate number, but the densities of many different data sets fall off at about this same rate. You can use -r to specify a different rate.

You can use the gamma option to thin out especially dense clusters of points. For any area where dots are closer than one pixel together (at whatever zoom level), a gamma of 3, for example, will reduce these clusters to the cube root of their original density.

For line features, it drops any features that are too small to draw at all. This still leaves the lower zooms too dark (and too dense for the 500K tile limit, in some places), so I need to figure out an equitable way to throw features away.

Any polygons that are smaller than a minimum area (currently 4 square subpixels) will have their probability diffused, so that some of them will be drawn as a square of this minimum size and others will not be drawn at all, preserving the total area that all of them should have had together.

Any polygons that have over 700 vertices after line simplification will be split into multiple features so they can be rendered efficiently, unless you use -pp to prevent this.

Features in the same tile that share the same type and attributes are coalesced together into a single geometry. You are strongly encouraged to use -x to exclude any unnecessary properties to reduce wasted file size.

If a tile is larger than 500K, it will try encoding that tile at progressively lower resolutions before failing if it still doesn't fit.

Development

Requires protoc and sqlite3. Rebuilding the manpage uses md2man (gem install md2man).

MacOS:

brew install protobuf

Linux:

sudo apt-get install libprotobuf-dev protobuf-compiler libsqlite3-dev

Then build:

make

and perhaps

make install

Examples

Check out some examples of maps made with tippecanoe

Name

The name is a joking reference to a "tiler" for making map tiles.

tile-join

Tile-join is a tool for joining new attributes from a CSV file to features that have already been tiled with tippecanoe. It reads the tiles from an existing .mbtiles file, matches them against the records of the CSV, and writes out a new tileset.

The options are:

• -o out.mbtiles: Write the new tiles to the specified .mbtiles file
• -f: Remove out.mbtiles if it already exists
• -c match.csv: Use match.csv as the source for new attributes to join to the features. The first line of the file should be the key names; the other lines are values. The first column is the one to match against the existing features; the other columns are the new data to add.
• -x key: Remove attributes of type key from the output. You can use this to remove the field you are matching against if you no longer need it after joining, or to remove any other attributes you don't want.
• -i: Only include features that matched the CSV.

Because tile-join just copies the geometries to the new .mbtiles without processing them, it doesn't have any of tippecanoe's recourses if the new tiles are bigger than the 500K tile limit. If a tile is too big, it is just left out of the new tileset.

Example

Imagine you have a tileset of census blocks:

curl -O http://www2.census.gov/geo/tiger/TIGER2010/TABBLOCK/2010/tl_2010_06001_tabblock10.zip
unzip tl_2010_06001_tabblock10.zip
ogr2ogr -f GeoJSON tl_2010_06001_tabblock10.json tl_2010_06001_tabblock10.shp
./tippecanoe -o tl_2010_06001_tabblock10.mbtiles tl_2010_06001_tabblock10.json

and a CSV of their populations:

curl -O http://www2.census.gov/census_2010/01-Redistricting_File--PL_94-171/California/ca2010.pl.zip
unzip -p ca2010.pl.zip cageo2010.pl |
awk 'BEGIN {
    print "GEOID10,population"
}
(substr($0, 9, 3) == "750") {
    print "\"" substr($0, 28, 2) substr($0, 30, 3) substr($0, 55, 6) substr($0, 62, 4) "\"," (0 + substr($0, 328, 9))
}' > population.csv

which looks like this:

GEOID10,population
"060014277003018",0
"060014283014046",0
"060014284001020",0
...
"060014507501001",202
"060014507501002",119
"060014507501003",193
"060014507501004",85
...

Then you can join those populations to the geometries and discard the no-longer-needed ID field:

./tile-join -o population.mbtiles -x GEOID10 -c population.csv tl_2010_06001_tabblock10.mbtiles

tippecanoe-enumerate

The tippecanoe-enumerate utility lists the tiles that an mbtiles file defines. Each line of the output lists the name of the mbtiles file and the zoom, x, and y coordinates of one of the tiles. It does basically the same thing as

select zoom_level, tile_column, (1 << zoom_level) - 1 - tile_row from tiles;

on the file in sqlite3.

tippecanoe-decode

The tippecanoe-decode utility turns vector mbtiles back to GeoJSON. You can use it either on an entire file:

tippecanoe-decode file.mbtiles

or on an individual tile:

tippecanoe-decode file.mbtiles zoom x y

If you decode an entire file, you get a nested FeatureCollection identifying each tile and layer separately. Note that the same features generally appear at all zooms, so the output for the file will have many copies of the same features at different resolutions.