Geography Reference
In-Depth Information
of measurement errors, corresponding objects should have close locations and
string matching using names could be done.
The review brought out several challenging research gaps that need to be
addressed in this study, such as:
merging opensource geospatial datasets has not been taken up.
global datasets need to be utilized for merging such as OSM, NE and GeoNames.
common attributes identification for reference of databases.
algorithmic resolution of geo-location matching amongst the databases for
higher accuracy.
Methodology
To meet the primary objective of the study which is to insert scale ranking data of
populated places into OSM dataset, it is required that the places appearing in NE be
matched with OSM places. OSM has a much bigger dataset than NE but NE has the
all important attribute of place_rank. Hence both datasets need to be merged. The
merging process is helped greatly by taking help from some fields in the GeoNames
database, for which the reason and procedure is explained below in Fig.
1
. The scale
ranking data of populated places can be found in Natural Earth dataset. For ease of
computation, it has been decided to begin with the populated places of a single
country, Czech Republic (CZ) at present. NE dataset for populated places has 7,312
populated places among which 12 are from CZ.
The work environment has been with Postgis with Postgres on Windows along
with Python. The pre-requisites in datasets and run-time environment are that for
OSM place-ranks the algorithm joins city rankings from Natural Earth into OSM
data with fuzzy matching. The setup computer script initiates by creating a
PostGIS-enabled PostgreSQL database. By default this script assumes it is named
'
. It imports the Natural Earth cities information included in Fig.
1
. Eg: psql -U
postgres -f ne_cities.sql -d osm. It imports OSM places with Imposm if not already
done. These can be from a full planet dump, a regional extract,or an Overpass API
query. It has to be ensured that the Python package
osm
'
unidecode
is installed.
'
'
Eg:
sudo pip install unidecode.
Usage:
python rank-places.py | psql -U
<
pg_user
><
pg_database
>
The columns used while joining the OSM and NE datasets are
way
and
geom
'
respectively Fig.
1
. Both are geometry type fields, however, they could belong to
different spatial reference systems. Hence conversion of the geometry field of one
of the datasets is performed based on the spatial reference system of the other. This
transformation of data brings the geometry fields of both the datasets to the same
'
'
'
