Extreme weather events, like extreme rainfalls, are severe weather hazards and also the triggers for other natural disasters like floods and tornadoes. Accurate forecasting of such events relies on the understanding of the spatiotemporal evolution processes in climate system. Learning from climate science data has been a challenging task, because the variations among spatial, temporal and multivariate spaces have created a huge amount of features and complex regularities within the data. In this study we developed a framework for learning patterns from the spatiotemporal system and forecasting extreme weather events. In this framework, we learned patterns in a hierarchical manner: in each level, new features were learned from data and used as the input for the next level. Firstly, we summarized the temporal evolution process of individual variables by learning the location-based patterns. Secondly, we developed an optimization algorithm for summarizing the spatial regularities, SCOT, by growing spatial clusters from the location-based patterns. Finally, we developed an instance-based algorithm, SPC, to forecast the extreme events through classification. We applied this framework to forecasting extreme rainfall events in the eastern Central Andes area. Our experiments show that this method was able to find climatic process patterns similar to those found in domain studies, and our forecasting results outperformed the state-of-art model.

Reliable tornado forecasting with a long-lead time can greatly support emergency response and is of vital importance for the economy and society. The large number of meteorological variables in spatiotemporal domains and the complex relationships among variables remain the top difficulties for a long-lead tornado forecasting.

The development of disastrous flood forecasting techniques able to provide warnings at a long lead-time (5-15 days) is of great importance to society. Extreme Flood is usually a consequence of a sequence of precipitation events occurring over from several days to several weeks. Though precise short-term forecasting the magnitude and extent of individual precipitation event is still beyond our reach, long-term forecasting of precipitation clusters can be attempted by identifying persistent atmospheric regimes that are conducive for the precipitation clusters. However, such forecasting will suffer from overwhelming number of relevant features and high imbalance of sample sets. In this paper, we propose an integrated data mining framework for identifying the precursors to precipitation event clusters and use this information to predict extended periods of extreme precipitation and subsequent floods. We synthesize a representative feature set that describes the atmosphere motion, and apply a streaming feature selection algorithm to online identify the precipitation precursors from the enormous feature space. A hierarchical re-sampling approach is embedded in the framework to deal with the imbalance problem. An extensive empirical study is conducted on historical precipitation and associated flood data collected in the State of Iowa. Utilizing our framework a few physically meaningful precipitation cluster precursor sets are identified from millions of features. More than 90% of extreme precipitation events are captured by the proposed prediction model using precipitation cluster precursors with a lead time of more than 5 days.