Special Session on:
Data-driven Approaches for Water Resources Forecasting and Knowledge Discovery


David J. Hill, Rutgers, the State University of New Jersey
Kenneth H. Reckhow, RTI International


Recent advances in environmental sensing technologies have lowered the barrier to knowledge discovery by giving us the ability to observe phenomena at time and space scales that have previously been impossible to access. Sensor networks present potentially profound opportunities for improving our understanding and ability to manage large-scale environmental systems. Likewise, national water monitoring programs (e.g., NAWQA, NARS) have resulted in large multivariate data sets, which with appropriate analytic techniques, can yield greater understanding of relationships and improve forecasting. This session focuses on new and emerging computational methods for harnessing the information value of these observations for knowledge discovery and forecasting.  This session encourages submissions that advance (1) integration and in-network processing of real-time data streams, (2) data mining frameworks for processing significant quantities of spatiotemporal data, (3) Bayesian or other statistical frameworks for forecasting future states given past behaviors, (4) machine learning and visualization techniques that can facilitate knowledge discovery, and (5) approaches for forecasting non-stationary processes.  Applications in surface hydrology, groundwater, water quality, and meteorology are encouraged.