Airport Amenities

airport_amenities_osm.md

Airport Shops on OSM

Tallinn Airport is said to be the 'cosiest' airport in the world. If you search for that on Google they come up first. I noticed none of the shops or other amenities around the gates are marked in OSM.

I wrote some code to see if this was an issue at airports of a similar size or larger nearby in other EU countries. I got a lot of false-positives but this code might help anyone else looking to enrich the Airport metadata in OSM.

TLL is surrounded by dense urban areas and a major shopping mall so I built a polygon mask around the airport gates and counted the shops within it.

Airports that came back with no shops within their gate area mask where:

• Bremen (BRE)
• Bromma (BMA)
• Dresden (DRS)
• Gdansk Lech Walesa (GDN)
• Kiruna_Airport (KRN)
• Kraków-Balice (KRK)
• Lulea (LLA)
• Nurnberg (NUE)
• Oulu (OUL)
• Rovaniemi (RVN)
• Stuttgart (STR)
• Tromsø Langnes (TOS)
• Ulemiste (TLL)

I checked Bremen, Krakow and Oulu and they look to be fairly well populated though I've only ever been in Bremen's airport and that was 20 years ago.

Bremen: https://www.openstreetmap.org/node/10796900775#map=17/53.053980/8.784814

Krakow: https://www.openstreetmap.org/node/10796900775#map=19/50.072307/19.801341

Count Shops in Gates Area

Natural Earth Download Guide: https://tech.marksblogg.com/natural-earth-free-gis-data.html

$ sudo apt update
$ sudo apt install \
    aws-cli \
    python3-pip \
    python3-virtualenv \
    unzip

$ virtualenv ~/.osm
$ source ~/.osm/bin/activate
$ pip install \
    duckdb \
    requests \
    shapely

import duckdb
import requests
from   shapely import convex_hull, MultiPoint, Polygon, Point


def osm_get(latitude:float,
            longitude:float,
            radius_meters:int=3_000,
            item:str='nwr[shop]'):
    query = f"""
    [out:json];
    (
      {item}(around:{radius_meters},{latitude},{longitude});
    );
    out center;
    """

    resp = requests.get('https://overpass-api.de/api/interpreter',
                        params={'data': query})

    if resp.status_code == 200:
        return resp.json()['elements']
    
    raise Exception(resp.text)


def get_point(shop):
    lon, lat = (shop['lon'], shop['lat']) \
                if 'lon' in shop.keys() \
                else (shop['center']['lon'], shop['center']['lat'])
    return Point(lon, lat)


sql = '''
SELECT   name,
         abbrev,
         ST_X(geom) as lon,
         ST_Y(geom) as lat
FROM     ST_READ('/home/mark/natural_earth/10m_cultural/ne_10m_airports/ne_10m_airports.shp')
WHERE    ST_DISTANCE(ST_POINT(24.801791, 59.416408), geom) < 20
AND      scalerank > 3
AND      gps_code LIKE 'E%'
ORDER BY name
'''

con = duckdb.connect(database=':memory:')
_ = con.sql('INSTALL spatial; LOAD spatial;')

shop_count = []

for rec in con.sql(sql).to_df().iloc():
    features = osm_get(rec['lat'],
                       rec['lon'],
                       3_000,
                       'nwr[aeroway=gate]')

    gates_area = convex_hull(MultiPoint([(x['lon'], x['lat'])
                                         for x in features]))

    if not gates_area.is_empty:
        features = osm_get(gates_area.centroid.y,
                           gates_area.centroid.x,
                           3_000,
                           'nwr[shop]')

        shops = [shop for shop in features if gates_area.contains(get_point(shop))]
        shop_count.append([rec['name'], rec['abbrev'], len(shops)])

print('\n'.join(['%s (%s)' % (name, abbrev)
                 for name, abbrev, count_ in shop_count
                 if count_ < 1]))

Tallinn Airport's OSM data:

To compare, Helsinki's Vantaa Airport has pretty much every shop and border post listed that I can remember.