Socialising the pixel: The use of citizen science in land use classification and its potential uncertainties

Socialising the pixel: The use of citizen science in land use classification and its potential uncertainties

Visser, T.

Civil Engineering and Geosciences

Water Management

2015-10-22

When making an land use map of an area with the use of remote sensing data, additional data from the ground is needed for this area, the so called ‘ground truth’. By combining the datasets a a land use classification can be done, relating a pixel with the actual situation on the ground. The collection of the ground data is labour intensive since enough data is needed in order to make a correct classification. This data collection could also be done by citizen science (volunteers that collect data for scientific research), which would allow to collect more data on a larger geographic and temporal scale (Cohn 2008). This paper looks at the possible factors that can cause error in the collection of ground truth data for land use. If one knows these factors it might be possible to minimise the error and have an idea on the accuracy of the data collected by citizens. With this in mind the different elements are described with the help of the available literature. This was followed by an experiment to study more in detail the relation between the error and the heterogeneity of the site, the training of the data collectors, and the specific land use class of the site. The experiment consisted of a photo experiment and a field experiment done with the help of Hanoi University of Natural Resources (HUNRE) in the Day River basin in Vietnam. This indicated that heterogeneity certainly made the classification more difficult. The difficulty of the classes was heavily influenced by the list with predefined classes given to the participants. To find the sites it might be easier to use smart phones with GPS applications to help the classification. It is recommended for certain experiments to be repeated in order to gain a deeper insight in these relations.

ground truth
citizen science
land use classification

http://resolver.tudelft.nl/uuid:013c82dd-0a70-4a0a-84a4-e70a462f0c0f

Master Project report

Student theses

student report

(c) 2015 Visser, T.