geo-wiki as a citizen science platform...geo-wiki as a citizen science platform linda see earth...
TRANSCRIPT
Geo-Wiki as a Citizen
Science Platform
Linda SeeEarth Observations Systems Group
Email: [email protected]
RESTORE+ Kick-off Workshop, 18 April 2017
Geo-Wiki
Visualization
of Global Land Cover,
Biomass, Photos, etc.
Crowdsourcing of
Land Cover
(Google Earth, Bing Maps,
Sentinel-2)
Creation of Hybrid
Land Cover Maps
Validation of Land
Cover Maps
using LACO-Wiki
In-situ data
collection
Via Geo-Wiki
Mobile apps
Serious Games
(Cropland
Capture,
Picture Pile)
Approaches to Citizen-based Data
Collection
• Run campaigns
– Opportunistic vs. directed
– Limited time period
– Incentives
– Linked to research questions or a
scientific purpose
– Online and/or mobile-based
http://www.geo-wiki.org
Initial Campaigns to Collect Data
Number Competition Purpose of the Competition
1 Human Impact Validate a map of land availability for biofuel
production
2 Hotspots of Map
Disagreement
Collect validation points in the areas were the GLC-
2000, MODIS and GlobCover disagree with one
another
3 Wilderness Collect land cover and human impact in map global
wilderness
4 Global Validation
Dataset
Collect data at the same locations as the validation
dataset for the first Chinese 30 m global land cover
map
5 & 6 Hackathon and
IIASA Competition
Collect data on the degree of cultivation and the
degree of human settlement in Ethiopia in the context
of land grabbing
Humanimpact.geo-wiki.org
• First competition – ran for 2 months
• Collected ~ 53,000 points
Human Impact Description
0% No evidence of any human activity visible
1 to 50%
Some visible evidence of human activities such as tracks/roads;
evidence of managed forests; some evidence of deforestation;
some scattered human dwellings, some scattered agricultural
fields; some evidence of grazing
51% to 80%
Increasing density of agriculture from subsistence on the lower
end to intensive, commercial agriculture with large field sizes on
the upper end
81% to 99%Urban areas with decreasing amounts of green space and
increasing density of housing
100%A built up urban area with no green space, typically the business
district of a city
Examples from Google Earth
Humanimpact.geo-wiki.org
Data collected from Campaign #1
Downgrading of Land Availability
for Biofuels
Scenario
Original
figures
(million ha)
Adjusted for
land cover
(million ha)
Adjusted for
field size
(millon ha)
Adjusted for
human impact
(million ha)
S1 320 98 42 34
S2 702 467 201 84
S3 1411 998 N/A 409
S4 1107 786 N/A 264
Fritz, S., See, L., van der Velde, M., Nalepa, R.A., Perger, C., Schill, C., McCallum, I., Schepaschenko, D.,
Kraxner, F., Cai, X., Zhang, X., Ortner, S., Hazarika, R., Cipriani, A., Di Bella, C., Rabia, A.H., Garcia,
A., Vakolyuk, M., Singha, K., Beget, M.E., Erasmi, S., Albrecht, F., Shaw, B., Obersteiner, M. 2013.
Downgrading recent estimates of land available for biofuel production. Environmental Science &
Technology, 47(3), 1688-1694.
Hackathon.geo-wiki.org
• Organized by USAID
• Challenge:
– Collect information about cropland and
settlement for Ethiopia
– Overlay with location of land
acquisitions
– Look for evidence of effects on local
populations
• Extended to a competition for 3 weeks
Data Collected
Interpolated Cropland Map
One Output: % Global Cropland Product
Fritz et al. 2015 Global Change Biology
Other Geo-Wiki Outputs from Campaigns
See et al. (2014) in ISPRS Photogrammetry and Remote SensingSchepaschenko et al. (2015) in Remote Sensing of Environment
Fritz et al. (2015) in Global Change Biology See et al. (2015) in Technological Forecasting and Social Change
Field Size Wilderness
Forest Cover Hybrid Land Cover
Recent Campaigns to Collect Data
Number Competition Purpose of the Competition
7 Cropland
Capture
Collect data on presence or absence of
cropland
8 Picture Pile Collect data on deforestation in
Tanzania/Indonesia
9 FotoQuest
Austria
Crowd were directed to locations in Austria
including LUCAS sample points
10 SIGMA Collect a validation dataset for cropland
Cropland Capture Game
Multi-platform including mobile devices
Cropland Capture Game Mechanics
Incentives
Leaderboards
Weekly prizes: one random classification is picked; the person who did
this classification receives a prize (last 5 weeks)
3 final winners: at the end of the competition 3 winners were drawn to
receiver bigger prizes (e.g. a tablet and smartphone)
Cropland Capture
3,314 Players
November, 2014 May, 2014
4,648,659Classifications
92%< 2 seconds
187,673Unique images
25 weeks
Defining What is Correct
• Working on the premise of ‘wisdom of the
crowd’
• Gave multiple players the same location
• Correct is what the majority (> 85%) says
(Output agreement)
• Players get +1 point if they classify an image
correctly, -1 point if wrong
• Built up frequency distributions at each location
Image 17365
Denmark
Yes=69
No=1
Maybe=0
Image 36318
Zimbabwe
Yes=21
No=20
Maybe=2
• 45 said
cropland
• 1 said not
cropland
But this is
pasture!
BUT the crowd is
not always
right…
Picture Pile
http://geo-wiki.org/games/picturepile
3.8 millionclassifications
Picture Pile
Deforestation Pile
Disaster Pile for rapid
assessment of building
damage
Pending:
Cropland
Human Impact
Oil palm identification
New scoring mechanism• More reference data based on
expert classifications
FotoQuest Austria (Field-based app)
Takes you to specific points
App design to help users follow the
‘LUCAS’ protocol
Scaling up to Europe with FotoQuest Go!
Results in Laso Bayas et al. (2017)
http://www.mdpi.com/2072-4292/8/11/905
FotoQuest Austria
July, 2015 September, 2015
12 weeks
200+Players
2000+Quests
12,000+Photos
12%Quests completed
Recent Campaigns to Collect Data
Number Competition Purpose of the Competition
7 Cropland
Capture
Collect data on presence or absence of
cropland
8 Picture Pile Collect data on deforestation in
Tanzania/Indonesia
9 FotoQuest
Austria
Crowd were directed to locations in Austria
including LUCAS sample points
10 SIGMA Collect a validation dataset for cropland
SIGMA Campaign• Ran last September 2016
• Collected data at 36K locations
• Validated by at least 3 people and up to 15 people
• Prizes: Amazon vouchers or co-authorship on a paper (Laso Bayas et al.
• submitted to Scientific Data)
Click on the Sub-pixels and Submit
Next Campaigns to Collect Data
• Field Size
– Collect data on the size of crop fields
– Will be launched on 15 May 2017
– Will lead to the development of an improved global
field size map (Fritz et al., 2015, Global Change
Biology)
• Use of Picture Pile for rapid assessment of
building damage (Hurricane Matthew)
– Working in collaboration with Humanitarian
OpenStreetmap Team (HOT), Imperative Space and
the European Space Agency
Geo-Wiki Pictures (Field-based app)
Can create your own
legend
Can indicate an offset for
distance way from a
feature
Can view, organize and
share photos in Geo-Wiki
Pictures branch
Lessons Learned
• Low cost solutions High volume of data
generated
• Crowdsourcing/citizen science well suited to
tasks of visual interpretation
• Need some form of incentives
• Gamification can help motivate participants
• Rigorous pre- and post-launch testing
• FEEDBACK, FEEDBACK, FEEDBACK
• Effective use of marketing/the media to launch
and run campaigns