National scale flood forecasting in the world of data, models, HPC and AI – shaping a more resilient tomorrow

Flooding is the most frequent natural disaster in New Zealand. Two trends have recently emerged in the development of new flood forecasting systems: i) a shift from deterministic to ensemble (probabilistic) streamflow predictions, and ii) a move towards national/continental scale systems that attempt to forecast flows for all streams over a given domain. We will present a selection of NIWA’s work on forecasting flood and hazards, including key aspects of the development of a national short-term flow forecasting system prototype for New Zealand rivers (0-2 days). The forecast system, running on HPC, is updated every 6 hours and produces river flow forecasts for nearly 60,000 rivers across the hydrologically diverse New Zealand environment. National scale forecasting systems require complex automated workflows, with data assimilating models, and uncertainty quantification techniques. We will highlight some of the key science and technical challenges of this work, and the role of data, models, HPC and AI, in shaping a more resilient tomorrow.