Usefulness of Social Sensing Using Text Mining of Tweets for Detection of Autumn Phenology

Abstract

Can social sensing detect the spatio-temporal variability of autumn phenology? We analyzed data published on the Twitter social media website through the text mining of non-geotagged tweets regarding a forested, mountainous region in Japan. We were able to map the spatial characteristic of tweets regarding peak leaf coloring along an altitudinal gradient and found that text mining of tweets is a useful approach to the in situ collection of autumn phenology information at multiple locations over a broad spatial scale. Potential uncertainties in this approach were examined and compared to other online research sources and methods, including Google Trends and information on widely available websites and live camera images. Finally, we suggest ways to reduce the uncertainties identified within our approach and to create better integration between text mining of tweets and other online research data sources and methods.

Perspective and review: how to develop our understanding of temporal changes in the relationship between people and the landscape under societal and climate change in Northeast Asia?

“…In an analysis of Twitter and Facebook data, which mainly consist of text, we can visualize people's interests by analyzing the frequency of terms and network graphs of morphemes, as well as examine people's emotional states by analyzing the context (Arthur et al, 2018;Shin et al, 2021;Albahli, 2022;Malik et al, 2022). People's interests cover various kinds of fields, such as politics, the economy, culture, nature, disaster, and society, and these interests can be considered to represent the relationship between people and the landscape in a given era.…”

“…In addition, the contents of Twitter, Facebook, Instagram, Flickr, and YouTube often include falsehoods and ethical issues (i.e., hate claims, and slander and libel against specific individuals or groups) (Al-Rawi et al, 2021;Albahli, 2022;Leahy et al, 2022;Toliyat et al, 2022;Wang et al, 2022). They also are limited and contain uncertainty caused by a lack of clarity in the terms of use for data and copyright, and Twitter, Facebook, and YouTube lack exact geolocation information unlike Instagram and Flickr (Shin et al, 2021;2024a). GT also includes uncertainty caused by changing analytical specifications, which have changed at least three times (https://trends.google.com/trends/).…”

“…As a consequence, this approach contributes to reducing interpretability gap in our understanding based on the satellite observation approach alone (Soliman et al, 2017). A typical example of this type of use is the spatio-temporal distribution of people's interests in cherry flowering and leaf coloring in Japan (ElQadi et al, 2021;Shin et al, 2021;2022c;ElQadi et al, 2023). Third, despite limitations on the period of analysis (e.g., only the past several years), this approach provides an immense amount of data about people's interests in tradition, culture, and customs, as well as temporal changes in these interests.…”

Perspective: Improving the accuracy of plant phenology observations and land-cover and land-use detection by optical satellite remote-sensing in the Asian tropics

“…The interests and movement of people at various locations can also be tracked by analyzing the access statistics of Google (Google Trends: Takada, 2012;Proulx et al, 2013), 14 number of visitors at Wikipedia (Fernández-Bellon and Kane, 2020), and geolocation information of mobile phones (Chang et al, 2021;Pintér and Felde, 2021). For instance, the analysis of Twitter posts was useful for evaluating the spatiotemporal variation of the timing of leaf coloring in Japan (Shin et al, 2021b). Kotani et al (2021) and Shin et al (2022a) analyzed the time-series of Google Trends and/or Yandex statistics (a major search engine in Russia) 15 to assess the spatiotemporal characteristics of people's interest in the use of berries in Arctic and the Russian Far East regions, 10.3389/ffgc.2023.1106723 which the authors used as proxy data of ripening phenology.…”

Understanding spatio-temporal variation of autumn phenology in temperate China from 1982 to 2018

“…The start, end, and length of the autumn brown-down phase in vegetation all hold substantial implications for the annual variations in vegetation growth and carbon uptake (Piao et al, 2008;Wu et al, 2022). However, in comparison to spring phenology, autumn phenology is influenced by multiple factors such as temperature, precipitation, and photoperiod, requiring further observation and understanding (Archetti et al, 2013;Xie et al, 2018;Nagai et al, 2020;Shin et al, 2021a). Previous studies have employed various methods to observe autumn phenology of vegetation, including satellite-based observations (Garonna et al, 2014;Nagai et al, 2015;Tsutsumida et al, 2022), model simulations (Zhu et al, 2019;Gauzere et al, 2020;Shin et al, 2021b), visual assessments of trees (Klosterman and Richardson, 2017;Klosterman et al, 2018;Xie et al, 2018), fixed-point observations with phenology cameras (Nagai et al, 2018;Richardson et al, 2018b;Xie et al, 2018), nearground observations with unmanned aerial vehicles (Klosterman and Richardson, 2017;Klosterman et al, 2018;Xie et al, 2018), and analysis of phenological information published on the Internet (Nagai et al, 2020;Tsutsumida et al, 2022) and social media (Shin et al, 2021a).…”