Geocoding Basics: Plot Addresses in QGIS

Introduction

How can you show the location of an address such as “3-1-1 Umeda, Kita Ward, Osaka City, Osaka Prefecture” in GIS? An address is a type of location data, but QGIS cannot display it on a map as it is. To show an address in GIS, you first need to convert it to coordinates such as latitude and longitude.

This article shows how to convert address data to coordinates and add it to QGIS as point data.

Converting an address to latitude and longitude

The process, or tool, that converts text such as an address, postal code, or facility name to coordinates such as latitude and longitude is called address matching or geocoding.

You cannot place points directly in GIS from address text alone. Once you convert the text to coordinates, you can display the locations in GIS software such as QGIS.

The reverse process, which converts latitude and longitude to an address, is called reverse geocoding.

Major geocoding (address matching) services

Several geocoding services exist in Japan and abroad, both paid and free.

This section introduces the major ones.

University of Tokyo CSV Address Matching Service

This address matching service is provided by the University of Tokyo. Upload a CSV file of addresses to the site, and you can download a CSV file with latitude and longitude added.

Converting takes nothing more than uploading a file, so the service is intuitive and easy to understand. It is also free to use, which is a major advantage.

jageocoder

This open-source Python library is geocoding software that uses the address geocoder integration feature included in PyGeoNLP.

It can also parse place names from text.

Nominatim

This geocoding API uses OpenStreetMap address data.

You can use it for free in the environment provided by the OpenStreetMap Foundation, but there are limits such as the number of requests (one request per second). For heavy use, setting up your own server is recommended.

Summary of the services

The first service, the University of Tokyo CSV Address Matching Service, is relatively easy for beginners to use: you select the CSV file of addresses you want to convert and run it. jageocoder and Nominatim are APIs intended for use in web pages and applications.

The University of Tokyo CSV Address Matching Service, jageocoder, and Nominatim are all free to use, but paid geocoding services exist as well. Conversion accuracy and pricing differ by service, so choose the one that suits your purpose.

Try geocoding

This section walks through geocoding with the University of Tokyo CSV Address Matching Service, which is easy for beginners to use.

Prepare the data

First, prepare a CSV file with the list of addresses to geocode.

This article uses the List of Designated Evacuation Shelters (Chuo Ward) from the Sapporo City ICT Utilization Platform DATA-SMART CITY SAPPORO. Download the data and check the contents of the CSV file.

In this data, the first column is 種別 (type), the second is 施設名 (facility name), and the third, 所在地 (location), holds the address. Make a note of which column holds the addresses.

Run the geocoding

Go to the University of Tokyo CSV Address Matching Service page.

Set the following required items.

Target area (対象範囲)

• Specify the rough area of the addresses in your CSV. If the addresses cover all of Japan, select Nationwide, block level (全国街区レベル). If they are within a single prefecture, select Each prefecture, block level (各都道府県 街区レベル).
• With options marked “(latitude/longitude, world geodetic system)” (（経緯度・世界測地系）), the geocoded coordinates are returned as latitude and longitude. With options marked “(public survey coordinate system, world geodetic system)” (（公共測量座標系・世界測地系）), they are returned as coordinates in the Japan Plane Rectangular Coordinate System.
• To get coordinates in the old geodetic system, select an option marked “old geodetic system” (旧測地系).

Column number containing the address (住所を含むカラム番号)

• Specify the number of the column that holds the addresses in your CSV. As confirmed above, the addresses are in the third column, so enter 3 as a half-width number.

File to convert (変換したいファイル名)

• Click the Choose File button (ファイルを選択) and select the CSV file that contains the addresses.

You can also set the following items as needed.

Character encoding of the input file (入力ファイルの漢字コード)

• Specify the character encoding of the CSV file. If you do not need to specify one, select Auto (自動設定).

Character encoding of the output file (出力ファイル漢字コード)

• Specify the character encoding of the returned file. If you do not need to specify one, select Same as input file (入力ファイルと同じ).

For more details on the CSV Address Matching Service, see the service page.

When you finish the settings, click the Submit button (送信). The geocoding result file downloads.

Check the geocoding results

Open the downloaded file in Excel or a similar application. New columns such as LocName have been added to the original data.

The columns are described below. For details, see this page.

Plot the results in QGIS

Geocoding returned the latitude and longitude of each address, so now plot the locations in QGIS from this information.

For details on plotting latitude and longitude locations in QGIS, see the following article as well.

1. Open Data Source Manager from the menu, then select Delimited Text.
2. For File name, specify the CSV file converted by geocoding.
3. Under Geometry Definition, set X field to the column that holds the longitude values (fX) and Y field to the column that holds the latitude values (fY).
4. For Geometry CRS, select the latitude and longitude coordinate system (EPSG:4326 - WGS 84).

Click Add. QGIS plots points at the latitude and longitude positions.

Common problems

The coordinates in the result are not latitude and longitude

When you check the result of geocoding with the University of Tokyo CSV Address Matching Service, the fX and fY columns, which should hold latitude and longitude, may contain other numbers. In that case, the coordinates may have been output in the public survey coordinate system.

If you select a public survey coordinate system for the target area in the parameter settings when you run the geocoding, the service outputs numbers that are not latitude and longitude, as shown here.

These numbers are values in the Japan Plane Rectangular Coordinate System, which is a projected coordinate system. For more on it, see the following article.

If you geocoded with a public survey coordinate system, select the appropriate coordinate system for Geometry CRS when you load the file in the QGIS Data Source Manager.

Points land in unexpected places

Depending on the geocoding result, the service may fail to determine the exact location of an address, or may match an address in the wrong place. Points can then land far from where you expected.

For example, the addresses geocoded here are for shelters in Chuo Ward, Sapporo, but the following figure shows points outside Sapporo as well.

In this case, addresses such as 南2条 (Minami 2-jo) and 北1条 (Kita 1-jo) in the geocoded 所在地 column also exist in Hokkaido outside Sapporo. This information alone was not enough to identify Sapporo, so the service probably treated them as addresses in other areas.

As this shows, ambiguity in how an address is written can cause it to be converted to the latitude and longitude of an unintended area. Try to improve the accuracy of the conversion, for example by including the prefecture and municipality names in the addresses you geocode.

It is also a good idea to plot the results on a map and check that the points land in the right places. If a point lands in the wrong place, manually moving the feature to the correct location is also an effective fix. For how to move features, see the following article.

Note that to edit point data added from CSV latitude and longitude, you must export the layer first.

Conclusion

This article explained geocoding, which converts addresses to coordinates such as latitude and longitude. To use data that contains address information in GIS, first geocode it to convert the addresses to coordinates. You can then handle it as location data.

This conversion is the first step in data analysis and visualization on a map, so it is a basic and important skill for using GIS.