Fig 1.
Sunrise in the Swiss Alps from one of the authors.
Fig 2.
Workflow schema for data filtering, transformation, aggregation and visualization.
Table 1.
List of keywords, for the four chosen languages and for sunset (top row) and sunrise (bottom row), whose whole occurrence in title, post body, or (hash-) tags defined the initial set of collected reactions.
Fig 3.
Chi-value merged—Flickr user count for "sunrise" (blue) and "sunset" (red), 2007–2018.
Focus on positive chi values, normalized to 1–1000 range, Head-Tail-Breaks, 100 km grid. Most significant five grid cells highlighted for sunrise (diamond) and sunset (square).
Fig 4.
Chi-value merged—Instagram, user count for "sunrise" (blue) and "sunset" (red), Aug-Dec 2017.
Focus on positive chi values, normalized to 1–1000 range, Head-Tail-Breaks, 100 km grid. Most significant five grid cells highlighted for sunrise (diamond) and sunset (square).
Fig 5.
Chi expectation surface for Countries for Flickr (top, 2007–2018) and Instagram (bottom, Aug-Dec 2017) and sunrise (left) and sunset (right).
Based on user counts, over- and underrepresentation, pooled quantiles classification applied to all four maps. Countries with non-significant results are shown with hatching.
Fig 6.
Temporal distribution of collected data for sunset and sunrise from Flickr.
Table 2.
Search terms and comparison of ranking of post count quantities overall for Instagram and Flickr (corresponding code in S9 File).
Fig 7.
TF-IDF for top scoring sunset terms in Zambia and Spain and global country similarity ranking based on cosine similarity to Zambia.
Fig 8.
Relationship between sunset and sunrise (two plots on the left) and Flickr and Instagram (two plots on the right).
Based on ranked (absolute) user count, each dot represents a country (su_a3 codes).