Analyzing the Spatiotemporal Effects on Detection of Rain Event Duration

Abstract

There has been significant recent interest in using the aggregate information from social media sites to detect and predict real-world phenomena. Temporal and geographic effects are often considered as two possible impact factors on detection of rain event from microblog data. However, the actual contribution of them to rain event detection has yet to be defined. To investigate this issue, one method considering overall effects of time and geography is proposed for detecting the rain event. Our analysis implies that the way people post tweets changes dynamically during a day. The number of tweets grows from the early morning and peak at midnight. Besides, distribution of the population and user responses to the rain event are both not the same in different regions. Our findings therefore suggest that temporal and geographic effects may play an important role in the detection of rain event. We also apply our strategy to forecast the rain events. Our results show that our strategy performs well both in detecting and predicting events of rain. Comparative analysis with existing methods is also presented to demonstrate the effectiveness of our method. Our proposed scheme is therefore practical and feasible to be deployed in the real world.