OpenStreetMap

From Wildsong
Jump to navigationJump to search

Here is a page about 2 different ways to use PostGIS with OSM data. I list Osmosis first because it's what I am looking at right now. Then I list osm2pgsql which I found to be workable and easier than osmosis.

I need to find out what the pros and cons are and list them here for you.


Get the data

Get the data for the western United States, grab the PBF either way.

wget http://download.geofabrik.de/north-america/us-west-latest.osm.pbf

Using osmosis

Let the data seep into PostGIS through a semipermeable membrane with osmosis, or use the command line tool of the same name.

Get the tool

This did not work very well for me No build required, that's the Java JAR advantage! (You knew there was an advantage.)

wget http://bretth.dev.openstreetmap.org/osmosis-build/osmosis-latest.zip
mkdir osmosis
cd osmosis
unzip ../osmosis-latest.zip

This is what I ended up doing

git clone https://github.com/openstreetmap/osmosis.git
cd osmosis
./gradlew assemble

Set up the database

createdb osm_us_west
echo "CREATE EXTENSION hstore; CREATE EXTENSION postgis;" | psql -U postgres $DB
echo "ALTER TABLE osm_us_west OWNER TO osm" | psql -U postgres
cd script
psql -U postgres osm_us_west -f pgsnapshot_schema_0.6.sql 
psql -U postgres osm_us_west -f pgsnapshot_schema_0.6_action.sql 
psql -U postgres osm_us_west -f pgsnapshot_schema_0.6_bbox.sql 
psql -U postgres osm_us_west -f pgsnapshot_schema_0.6_linestring.sql

Load the data into PostGIS

Run osmosis to load the PBF file into your PostGIS server. I do it thusly

# Put TEMP files somewhere with lots of space
mkdir tmp
JAVACMD_OPTIONS="-Djava.io.tmpdir=tmp" \
./osmosis -v --read-pbf us-west-latest.osm.pbf --write-pgsql user=osm_reader database=osmosis_us_west

At this point I have the data loaded and to prove that it's usable I need to try QGIS to view it and I need to extract a subset.

Using osm2pgsql tool

Using excellent instructions from http://bostongis.com/ In particular, the Almost idiot's guide...

First, get the data as above, the PBF file for western US. Create a lovely Postgis database and spatially enable it and add hstore.

Build osm2pgsql

No need for any special configuration options, just build and install!

sudo apt-get install cmake libboost-dev libboost-system-dev libboost-filesystem-dev zlib1g-dev libbz2-dev lua5.2 liblua5.2-dev
cd ~/src/GIS
git clone https://github.com/openstreetmap/osm2pgsql.git
cd osm2pgsql
mkdir build
cd build
cmake ..
make
sudo make install

Using osm2pgsql

# bellman
osm2pgsql oregon.osm.bz2 -d osm -U postgres -S /green/bwilson/src/GIS/osm2pgsql/default.style --hstore
osm2pgsql california.osm.bz2 --append -d osm -U postgres -S /green/bwilson/src/GIS/osm2pgsql/default.style --hstore
# dart
createuser -U postgres osm
createdb -U postgres osm
psql -U postgres osm 
CREATE EXTENSION hstore;
CREATE EXTENSION postgis;

# The western US - I had to increase the node cache size
osm2pgsql us-west-latest.osm.pbf \
  -d osm -U osm -S ~/src/GIS/osm2pgsql/default.style --hstore -C 10000 \
  --number-processes 8 --multi-geometry
psql -U postgres osm 
 CREATE INDEX idx_planet_osm_point_tags ON planet_osm_point USING gist(tags);
 CREATE INDEX idx_planet_osm_polygon_tags ON planet_osm_polygon USING gist(tags);
 CREATE INDEX idx_planet_osm_line_tags ON planet_osm_line USING gist(tags);
# The whole planet - I RUN OUT OF MEMORY!! Dang.
osm2pgsql planet-latest.osm -d osm -U osm -S ~/src/GIS/osm2pgsql/default.style --hstore -C 15000

On my little server Bellman, it took 628 seconds to load the Oregon data and it took 3302 seconds to load California. Conclusion: California is much bigger, in fact there is probably room for me there in Nevada county.

I just loaded Oregon data onto my much snazzier server at work and it took 183 seconds.

Nice to have some data in my PostGIS server.

Now I need to get it showing up in Geoserver