It’s springtime in Japan and I recently saw a series of posts in the The Economist magazine and on Twitter about the latest dataset release for the Kyoto cherry blossom time (‘Hanami‘ 花見, “flower viewing” in Japanese). The remarkable thing about this dataset is that it contains observations of the day of maximum blossom going back to 812CE, over 1200 years of data that has been carefully reconstructed by the Japanese researchers, Aono, Kazui, and Saito. Maximum blossom in Kyoto this year (2021) happened on March 26th, part of a general trend towards early blossoming over recent decades (which we can clearly see in the 50 year trend line in bright pink). Maximum in Blossom in March has been rare for much of recorded history, a trend which may be associated with global heating in the industrial era. The boxplot below clearly shows the distinction between the last thirty years (1990-2021CE) and the period (812-1989), with the mean max blossom day for the former period falling 10 days earlier than the latter.
This analysis was achieved using the raw data source csv file, the substantial work “Historical Series of Phenological data for Cherry Tree Flowering at Kyoto City” by Aono and Kazui, 2008; Aono and Saito, 2010, that I found referenced on the Datagraver website. It’s really just another in the series of practice data files that I use as I continue to build my Python data science skills, but this one was particularly pretty and striking in its final visualisation, as well as being of further interest to me due to personal connections with Japan.
Technical note: principal tools for this analysis were the Pandas, Matplotlib and Seaborn libraries running on Jupyter Notebooks within the Google Colabs environment. The Python knowledge is largely from www.datacamp.com (which is subscription, but I highly recommend).