In a past exploratory activity, we investigated the use of social media platforms (Twitter and Flickr) in combination with authoritative sources in order to monitor environmental phenomena, here especially forest fire detection. Despite the specific knowledge that was gained in mining particular social media platforms, this project resulted in a generally applicable workflow for implementing a Digital Earth Nervous System.
A case study of the 2010 forest fire season illustrated how the mined data could be used in combination with official land cover data sets and information on forest fire risks to identify those forest fires that are relevant to people. For this particular case, we found an agreement of 80% with JRC’s European Forest Fire Information System (EFFIS), but could also derive interesting findings about false positive and false negatives. On the one hand, the developed approach based on social media could identify several small forest fires new urban settlements that were not visible in EFFIS. On the other hand, a very large fire contained in the official European system was not spotted. The burned area was completely inside a military-protected zone, i.e. people on social media were not interested in this particular fire.
This exercise provided insides on handling new and complementary information sources and work that is required to sense particular phenomena using new media. It allowed reflections on the potential and limitations of these growing information sources.