The human impact competition was initiated in September 2011. Users were randomly provided with MODIS-sized pixels of 500m resolution and were then asked to answer 3 questions based on what they could see on Google Earth. The questions were:
In addition, users were asked to enter the date of the image and whether the image was high resolution. The resolution indicates potential difficulties in identifying human impact, land cover and/or abandoned land, especially when users are presented with low resolution imagery.
More than 60 people took part in the competition with approximately 55,000 points validated in total. Figure 1 summarizes the number of points validated by the top ten competitors (Table 1) along with a score for quality to determine their final ranking. Quality was measured by comparing a set of control pixels derived from experts with the answers provided by the crowd. Factoring in quality actually resulted in changes to the ranking, with 2 competitors with higher validations dropping out of the top ten and two entering due to their accuracy in identifying land cover and human impact. The users were informed of this possibility on a number of occasions so they clearly understood that quality is of utmost importance.
|1||Simone Ortner||University of Applied Sciences, Wiener Neustadt, Austria|
|2||Rubul Hazarika||Gauhati University, Assam, India|
|3||Anna Cipriani||Università di Modena e Reggio Emilia, Modena, Italy and Lamont Doherty Earth Observatory of Columbia University, Palisades, New York, USA|
|4||Carlos Di Bella||National Institute of Agricultural Technology (INTA), Buenos Aires, Argentina|
|5||Ahmed H. Rabia||University of Naples Federico II, Naples, Italy|
|6||Alfredo G. García||National Institute of Agricultural Technology (INTA), Buenos Aires, Argentina|
|7||Mar'yana Vakolyuk||IIASA and Centre for Aerospace Research of the Earth (CASRE) Kyiv, Ukraine|
|8||Kuleswar Singha||Gauhati University, Assam, India|
|9||María E. Beget||National Institute of Agricultural Technology (INTA), Buenos Aires, Argentina|
|10||Stefan Erasmi||University of G?ttingen, G?tingen, Germany|
Although the campaign was a success with such a large number of points collected over a two month period, a number of issues arose during the competition, which are being used to improve Geo-Wiki. These include:
1. In the human impact competition we asked users to classify the land cover based on the dominant land cover type. We have now added the ability to indicate how many land cover types there are in an area based on the percentage of occurrence. A grid is overlaid onto the area to help users determine these percentages.
2. The addition of other information from external sources such as flickr, picasa, youtube and twitter to help in making interpretations, particularly with low resolution imagery.
3. The provision of social networking tools embedded directly within Geo-Wiki so that users can interact, discuss areas that are hard to classify and ask the Geo-Wiki team any questions during the competition.
Last edited: 14 February 2013
International Institute for Applied Systems Analysis (IIASA)
Schlossplatz 1, A-2361 Laxenburg, Austria
Phone: (+43 2236) 807 0 Fax:(+43 2236) 71 313