VALID-LAND: assessing the fitness of citizens … · VALID-LAND Workshop - AGILE - June 2015 Data sets used Data File name Original Reference System OSM polygons London_osm_ply_2015

June 2015

VALID-LAND: assessing the fitness of citizens

observatories for LAND cover/use mapping &

VALIDation

Organizing committee

Jamal Jokar Arsanjani, GIScience Research Group, University of Heidelberg, Germany

Marco Painho, NOVA IMS – Information Management School, Portugal

Jacinto Estima, NOVA IMS – Information Management School, Portugal

Cidália Fonte, University of Coimbra, Portugal

Linda See, International Institute for Applied Systems Analysis, Austria

Lucy Bastin, Aston University, UK

Vyron Antoniou, Hellenic Army Geographical Service, Greece

mailto:[email protected]







VALID-LAND Workshop - AGILE - June 2015

Objectives

• Analyse different VGI initiatives that can potentially provide land cover/use related information

• Compare the LULC information provided by these different VGI initiatives for a study area (London) and other free data available


Workplan

• Introductory lectures

• Analysis of the different data and nomenclatures used

• Classification of photographs

• Data integration

• Discussion


Volunteered Geographic Information

(VGI)• Definition

▫ Geographic Information provided by volunteers Goodchild, M.F., 2007. Citizens as sensors: the world of

volunteered geography. GeoJournal 69, 211–221

• Other terms can also be found- Contributed Geographical Information - Collaborative Mapping- Public Participation Geographic Information

Systems

- Involuntary geographic information- ...


Types of VGI

• Photographs and descriptions

▫ Panoramio (http://www.panoramio.com/)

▫ Flickr (http://www.flickr.com/)

▫ Geograph (http://www.geograph.org/)

▫ Degree Confluence Project (http://confluence.org/)

http://www.panoramio.com/

http://www.flickr.com/

http://www.geograph.org/

http://confluence.org/


Types of VGI

• Vector Maps

▫ OpenStreetMap (http://www.openstreetmap.org/)

▫ Wikimapia (http://wikimapia.org)

▫ Wikiloc (http://www.wikiloc.com)

▫ …

• Classification made by volunteers

▫ Geo-Wiki Project (http://www.geo-wiki.org/)

http://www.openstreetmap.org/

http://wikimapia.org/

http://wikimapia.org

http://www.wikiloc.com/


Study area

• Region of London (UK)

▫ Bounding Box with limits:

0.217 W, 51.466 N to 0.043 W, 51.526 N


Data sets usedData File name Original Reference System

OSM polygons London_osm_ply_2015 Geographic Coordinates: WGS84

OSM linear features London_osm_ln_2015 Geographic Coordinates: WGS84

GMESUA – Urban AtlasLondon_UA_clip.shp

Projected Coordinate System: ETRS_1989_LAEALondon_UA_clip.lyr

Ordnance Survey raster

basemaps

TQ_27

Projected Coordinate System: British National GridTQ_28

TQ_37

TQ_38

Location of Panoramio photos Panoramiox.shp Geographic Coordinates: WGS84

Location of Geograph photos Geographxshp Geographic Coordinates: WGS84

Location of Flickr photos Flickrx.shp Geographic Coordinates: WGS84

All files were converted to the British National Grid (Projected Coordinate System)


Data description – Urban Atlas (UA)

• Project of the Global Monitoring for Environment and Security (GMES) - European Environment Agency (EEA)

• Aims to provide high resolution Land Use Land Cover maps (LULCM) for Pan-European regions▫ with more than 100 000 habitants

• Freely available through the European Environmental Agency website:▫ (http://www.eea.europa.eu/data-and-

maps/data/urban-atlas) in vector format



• Minimum mapping units:▫ 0,25ha for area features▫ 100m for linear features▫ geometric scale of 1:10 000

• The classification of UA is made considering a nomenclature separated into levels▫ European Union, 2011. Mapping guide for a European

Urban Atlas. Available from: http://www.eea.europa.eu/data-and-maps/data/urban-

atlas [Accessed 2 June 2015].

Urban_Atlas_2006_mapping_guide_v2_final.pdf



• Levels 1, 2 and 3 ▫ Further description: European Union, 2011

Urban Atlas nomenclature used

Level 1 Level 2 Level 3

1.Artificial Surfaces (1)

1.1 Urban Fabric (11)

1.1.1 Continuous urban fabric (111)

1.1.2 Discontinuous urban fabric (112)

1.1.3 Isolated Structures (113)

1.2 Industrial, commercial, public, military

and private units or transport units are

predominant (12)

1.2.1 Industrial, commercial, public, military and private

units (121)

1.2.2 Road and rail network and associated land (122)

1.2.3 Port areas (123)

1.2.4 Airports (124)

1.3 Mine, dump and construction sites (13)

1.3.1 Mineral extraction and dump sites (131)

1.3.3 Construction sites (133)

1.3.4 Land without current use (134)

1.4 Artificial non-agricultural vegetated

areas (14)

1.4.1 Green urban areas (141)

1.4.2 Sports and leisure facilities (142)

2. Agricultural, semi-

natural areas, wetlands (2)

2.0 Agricultural, semi-natural areas,

wetlands (20)2.0.0 Agricultural, semi-natural areas, wetlands (200)

3. Forests (3) 3.0 Forests (30) 3.0.0 Forests (300)

5. Water (5) 5.0 Water (50) 5.0.0 Water (500)


Data description – OpenStreetMap

(OSM)• OpenStreetMap data can be downloaded in several

ways:▫ using tools available in some GIS software▫ from the OSM web page indicating a Bounding Box▫ using Geofabrik (http://download.geofabrik.de/)▫ …

• Different sources may provide data structured in slightly different ways

• The data used in the workshop:▫ retrieved from the OSM planet file - downloaded from

Geofabrik.de

http://download.geofabrik.de/


Data description – OpenStreetMap

(OSM)• Data (two shape files):

▫ One with linear features

London_osm_ln_2015.shp

▫ One with polygonal features

London_osm_lply_2015.shp


Data description – Ordnance Survey

(OS)• The UK Ordnance Survey is the official mapping

agency of the UK

• A large proportion of the data are now freely available for downloading from this site:

▫ https://www.ordnancesurvey.co.uk/opendatadownload/products.html

https://www.ordnancesurvey.co.uk/opendatadownload/products.html


Data description – Ordnance Survey

(OS)• Data used:

▫ Data from the OS VectorMapTM District product (as a raster file)

version date September 2014

▫ Data are selected by OS grid squares where the majority of London falls in grid square TQ

▫ The data are also available as shapefiles but for this exercise the raster files are used for context


Data description – Panoramio

• Panoramio is a community-powered site for exploring places through photography

• The photos illustrate places

• Photos can be accessed and explored through a website (http://www.panoramio.com/)

▫ a world map for browsing by location is available and a public API:

(http://www.panoramio.com/api/data/api.html)

REST interface to communicate with the Panoramio database

http://www.panoramio.com/

http://www.panoramio.com/api/data/api.html


Data description – Panoramio

• To get the metadata for the available photos within a certain boundingbox a GET operation needs to be done, such as:▫ http://www.panoramio.com/map/get_panoramas.php?set=public&from=0&to=20&minx=-

180&miny=-90&maxx=180&maxy=90&size=medium&mapfilter=true

▫ The boundingbox is defined by minx, miny, maxx and maxy representing respectively minimum longitude, minimum latitude, maximum longitude and maximum

latitude

▫ For "set“ the following can be used: public (popular photos), full (all photos), user ID number.

▫ For "size“ the following can be used: original, medium (default value), small, thumbnail, square, mini_square

▫ The number of photos to be displayed can be defined using "from=X" and "to=Y“ where Y-X is the number of photos included

The value 0 represents the latest photo uploaded to Panoramio. For example, "from=0 to=20" will extract a set of the last 20 photos uploaded to Panoramio, "from=20 to=40" the previous set of 20 photos

▫ The maximum number of photos possible to retrieve in one query is 100.

http://www.panoramio.com/map/get_panoramas.php?set=public&from=0&to=20&minx=-180&miny=-90&maxx=180&maxy=90&size=medium&mapfilter=true


Data description – Flickr

• Flickr is an initiative based on an online application that allows photo storage and sharing throughout the Web

• Users can freely attribute classes to photos using a tag mechanism that further helps to organize and search for content in the application

• Special tags can be used to add location information in the form of latitude and longitude, either automatically using GPS-enabled devices or manually using specific available tools

• Photos can then be associated with a point location in a map using these spatial tags called geotags


Data description – Flickr

• Photos can be accessed and explored through a website (https://www.flickr.com/) ▫ a world map for browsing is available and a public

API: (https://www.flickr.com/services/api/)

more oriented for Web development and coding purposes.

▫ Many methods are available in the API interface one of the most interesting is the “flickr.photos.search”

methods that allows to search for available photos using a set of criteria. Only photos visible to the calling user will be returned.

https://www.flickr.com/


Data description – Geograph

• The project aims to organise and preserve representative images and associated information for every square kilometre of Great Britain, Ireland, and the Isle of Man.

•Geograph is a spatially explicit web application in contrast with other spatially implicit sources like Flickr or Picasa Web. Thus the scope is to describe places using photos.

•The geographical database is freely available to the public either as a full data dump or via dedicated API.


Data description – Geograph

• Available metadata include:▫ image id▫ user id▫ title▫ moderation▫ date taken▫ date submited▫ OS grid reference▫ WGS84 lat▫ WGS84 long▫ Sequence Number

▫ OS easting▫ OS northing▫ Accuracy Used▫ Viewpoint easting▫ Viewpoint northing▫ Viewpoint accuracy▫ View_direction▫ Class▫ Comments▫ Tags


Exercise 1

• Aim:

▫ Examine Urban Atlas nomenclature and data

• Steps:

▫ Get familiar with Urban Atlas nomenclature

▫ Get familiar with Urban Atlas data for the study area


Exercise 1

A. Urban Atlas (UA) nomenclature analysis

1. Find the handout which provides the classes in the Urban Atlas and read through these

2. How many classes are there in level?

3. How are roads characterized in level 1, 2 and 3?

4. At what level are airports captured?


Exercise 1

B. Map the Urban Atlas Data in QGIS

1. In QGIS load the shapefile called London_UA_OSGB_7par.shp from the Exercise1 folder.

2. Apply the style file called UrbanAtlas_rgb_1.7.qml from the Exercise1 folder.

3. Where are main industrial areas located?4. Using a rough visual estimate, what proportion of

the land is ‘Land without current use’?5. Are there any wetlands in this part of London?6. Are there classes that are not present?


Exercise 2

• Aim:▫ Determine the usefulness of photographs for validation of

the Urban Atlas

▫ Gain experience with different sources of photographic VGI

• Steps▫ Examine photographs from three different sources:

Flickr, Geograph and Panoramio.

▫ Classify the photos based on levels 1, 2 and 3 of the Urban Atlas nomenclature, without spatial context.

▫ Plot the photographs in QGIS with reference data as context

▫ Modify the photos classifications according to the context.


Exercise 2A. Classification of photographs

1. From the Exercise2 folder, load the shapefile called SamplePhotos.shp in QGIS2. Open the attribute table3. For each photograph, copy the URL to a browser and view the contents of the

photograph4. Using the Urban Atlas nomenclature (see in the Exercise 2 folder file

Urban_Atlas_2006_mapping_guide_v2_final.pdf), classify the photographs into levels 1, 2 and 3.

5. Make sure the table is editable and enter the values directly into the three columns labelled UA_L1, UA_L2 and UA_L3.

6. For each level enter a first option and an alternative one in columns “L1Choice2”, “L2_Choice2” and “L3Choice2” (if you are not sure about the class to assign to the photo)

7. In the field called Comment, enter any remarks regarding, for example, how easy it was to classify the photograph or what difficulties you encountered


Exercise 2

B. Plot the Photographs on Reference Data and Adjust the Classifications

1. Load the four raster files (TQ27, TQ28, TQ37, TQ38) from the Exercise2 folder as context.

These rasters contain UK Ordnance Survey data.2. For each photograph,

Examine the classification for all levels using the raster map

If appropriate, modify the photographs classification using the context information available on the reference raster

In the comments field, indicate whether you changed your classification


Exercise 2

C. Compare your Classifications with the Urban Atlas

1. Load the shapefile called London_UA_OSGB_7par.shp from the Exercise1 folder.

2. Apply the style file called UrbanAtlas_rgb_1.7.qml from the Exercise1 folder

3. Intersect the photo shapefile with the Urban Atlas

4. Calculate the agreement for levels 1, 2 and 3


Exercise 2

D. Discussion Points

1. How many classifications did you change once you had more context?

2. How well did your classifications match the actual Urban Atlas classes?

3. In those situations where the match was poor, what were the reasons?


Exercise 3

• Aim:▫ Classify a larger set of photos from Flickr, Geograph

and Panoramio so that these can be compared with the Urban Atlas

▫ Gain a better understanding of the value of photographs for validation of land cover/land use

• Steps▫ Classify Photographs from Flickr▫ Classify Photographs from Geograph▫ Classify Photographs from Panoramio▫ Compare the Results from the three photograph

sources


Exercise 3

A. Classify Photographs from Flickr1. Load the shapefile called Flickrx.shp where x will be a number

assigned to you.2. Open the attribute table and create three new integer columns

called L1, L2 and L3.3. Find the column containing the URL of the photographs.4. Copy the first URL to a browser, examine the photograph and

determine the L1, L2 and L3 class.5. There are 100 photographs in the file – complete as many as

you wish. If a photograph is unclassifiable, assign code 9 to L1.6. Load the Urban Atlas (in the Exercise1 folder) and intersect the

two layers. 7. Calculate the agreement between the photographs and the

Urban Atlas for L1, L2 and L3 and summarize the results in a table.


Exercise 3

B. Classify Photographs from Geograph1. Repeat steps 1 to 7 as outlined section A2. In step one, the shapefile will be called

Geographx.shp where x will be a number assigned to you

3. The other main difference is in step 3 because the URLs are not contained in the shapefile.

Instead load the search screen for Geograph: http://www.geograph.org.uk/search.php

Copy the value from the field called gridimage_ into the search box.

This will provide a link to the photograph, which you can display.

http://www.geograph.org.uk/search.php


Exercise 3

C. Classify Photographs from Panoramio

1. Repeat steps 1 to 7 as outlined section A

2. In step one, the shapefile will be called Panoramiox.shp where x will be a number assigned to you


Exercise 3

D. Compare the Results from the Three Photograph Sources

1. Merge the tables that you created in step 7 of sections A to C into a single file, e.g. in Excel

2. Which source of photographs provided the highest L1 agreement?

L2 agreement?

L3 agreement?

3. How many photographs were unclassifiable and what were the main reasons?

4. How useful do you think these photographs are for validation?


Exercise 4

• Aim:▫ Compare the Urban Atlas with data from OpenStreetMap in

order to determine how well they agree the differences in agreement between the three levels where they disagree and the possible reasons for the

disagreement▫ Gain a better understanding of the value of photographs for

validation of land cover/land use• Steps

▫ Create a Grid of Points for Comparison▫ Extract the Data from Urban Atlas and OpenStreetMap▫ Calculate the Agreement between the Urban Atlas and OSM▫ Investigate Areas of Disagreement


Conversion of OSM data to UA classes

• Linear features

▫ The attributes and their values need to be identified

▫ The attributes and their values need to be mapped into the UA classes

▫ The linear features need to be converted into areas

This is done considering a buffer around each line

The width of the lines needs to be identified

Assigment by attribute value


Conversion of OSM data to UA classes

• Poligonal features

▫ The attributes and their values need to be identified

▫ The attributes and their values need to be mapped into the UA classes

• Linear and poligonal features files are merged

• The features are “Dissolved” according to the UA classes assigned to the features


Exercise 4

A. Create a Grid of Points for Comparison

1. Load a shapefile that contains the bounding box for London (LondonBB.shp) from the Exercise4 folder.

2. Create a random point of grids of 1000 points.


Exercise 4

B. Extract the Data from Urban Atlas and OpenStreetMap

1. Intersect the grid points with the Urban Atlas (found in the Exercise1 folder).

2. Use the output from step 1 as the input to a second intersection between this file and the OpenStreetMap file (called OSMMerged.shp in the Exercise1 file).

3. The results should be a shapefile with a grid of 1000 points that have both the attributes from UA and from OSM. Open the attribute file to make sure the intersections have been successful.


Exercise 4

C. Calculate the Agreement between the Urban Atlas and OSM

1. Create three new columns in the point shapefile created in step 2 of Section B called L1Agree, L2Agree and L3Agree.

2. Use the Field Calculator to assign a value of 1 to the columns if the Urban Atlas and OSM agree on the classifications and 0 if they disagree for each of the three levels, i.e. L1, L2 and L3.

3. Create a summary table for the three levels.4. What is the percentage agreement for each level?


Exercise 4

D. Investigate Areas of Disagreement▫ In this section you will look for places where the

Urban Atlas and OSM disagree and try to determine the reasons why

1. Using the point shapefile, colour/symbolize the points according to agreement and disagreement.

2. Look at points of disagreement and determine which one is correct and the reasons why. Make a list of these for discussion at the end.


Comparison of Results

• Photo comparison

• UA/OSM comparison

VALID-LAND: assessing the fitness of citizens … · VALID-LAND Workshop - AGILE - June 2015 Data sets used Data File name Original Reference System OSM polygons London_osm_ply_2015

Documents