Extended-Range Tropical Cyclone Forecasts by the ECMWF 32-Day Ensemble and its Reforecast Data for Water Resources Management

The objective of this study is to evaluate the twice-weekly extended-range (5-30 day) predictions of western North Pacific tropical cyclone (TC) events and their potential for the disaster prevention and water resources management in Taiwan reservoir watershed areas. Before a TC exists, the extended-range ensemble predictions can be utilized to evaluate the probability of TC formations and subsequent tracks within the next 4 weeks. Given these track forecast positions relative to Taiwan topography, the precipitation and reservoir inflow can also be estimated by climatological/statistical relationships. If a TC has already formed, the extend-range ensemble track forecasts can be used to estimate the precipitation and reservoir inflow probabilities within the next 5-10 days to extend support to the decision-making processes for reservoir flood control operations and water resources management. The first objective of this study is to analyze the skill of the ECMWF 32-day ensemble prediction system TC forecast and quantify the forecast uncertainty. The second objective is to use the 20-year reforecast datasets to investigate the relationships among the historical forecast errors and the large-scale environmental factors (e.g., El Niño, La Niña, Madden Julian Oscillation, etc.). Preliminary evaluations of the ECMWF forecasts prior to the formation of Typhoon Matmo in 2014 show that the model can predict the formation of Matmo in all of the Week-1 through Week-4 forecasts. By utilizing the reforecasts, the signals of TC activities from the raw track forecasts can be calibrated and further improved if the model biases are considered.