Detecting information trends in online social networks is an important problem that has attracted the attention of both the industry and the research community in recent years. Global trends, information items that are trendy in the entire social network, can be detected using existing data streams techniques. However, detecting global trends is only the ﬁrst step in understanding online social networks. As The First Law of Geography states “Everything is related to everything else, but near things are more related than distant things”. This spatial signiﬁcance has implications in various applications, trend detection being one of them. To this end, in this paper we propose a new algorithmic tool, GeoWatch, to detect geo-trends. GeoWatch is a data streams solution that detects correlations between topics and locations in a sliding window in addition to providing tools for analyzing topics and locations independently. The degree of correlation as well as the sliding window size can be set to arbitrary values thus enabling a ﬂexible framework. GeoWatch has theoretical guarantees for detecting all trending correlated pairs while requiring only sublinear space and running time. Experiments on Twitter show that in addition to providing perfect recall, GeoWatch has near-perfect precision. As the Twitter analysis demonstrates, GeoWatch successfully ﬁlters out topics without geo-intent and detects various local interests such as emergency events, local political demonstrations or cultural events.