Using tweets for rainfall monitoring

Abstract

In Brazil, the summer season is the wettest period in which many disasters can happen, such as landslides and floods. In recent years, even with the rainy season, the metropolitan region of São Paulo (Brazil) suffers a severe water crisis. Given this scenario, monitoring of rainfall is fundamental for taking preventive actions and planning in the various business branches. Thus, the use of computers to develop tools that assist the rainfall monitoring can help extend the coverage of the existing solutions. Moreover, it is known that, every day, the number of social media users is increasing, and consequently increases the amount of content published in these medias. The objective of this study is to analyze the contents of the Twitter social media, especially the tweets related to rainfall events in order to determine whether this information can contribute to the monitoring of rainfall events in Brazil. More than 1 million tweets published in Brazil related to rainfall were collected in a period of 30 days. Gathered tweets were analyzed and evaluated taking into account the data collected by automatic weather stations (AWS or EMA). The results were satisfactory and indicate a relationship between the geolocated tweets and data from AWS.