Linking weather generators and hydrological models for streamflow assessments with seasonal climate outlooks

Climate variability and change present crucial challenges in managing water resources and the associated risks. Timely communication of climate forecast information may help mitigate the devastating social-economic impacts from climate extremes. Efficient and effective applications of climate forecast products require that climate information become integrated into assessments of various climate sensitive sectors. In this study, seasonal climate outlooks provided by the Central Weather Bureau (CWB) in Taiwan are integrated with weather generators and hydrological models to forecast stream inflows of the Shihmen Reservoir Watershed with lead times of up to 3 months. The percentage of hits and the Heidke skill score are used to evaluate the seasonal forecast’s skills of CWB climate outlooks. Both the percentage of hits and the HSS shows acceptable skills meaning that the CWB climate outlooks are better than random guesses. The state of monthly mean temperature and precipitation projected by climate outlooks are then used with historical climate statistics for daily weather generations. Two weather generators are investigated in this study. The first one is a semiparametric multivariate weather generator, including a Markov Chain for generating the precipitation state (i.e., no rain, or rain) and a k-nearest neighbor (k-NN) bootstrap resample for generating daily precipitation and temperature. The second one also includes a Markov Chain for generating the precipitation state, but the precipitation amount is estimated by exponential distributions, and the temperature is generated by the first order serial correlation coefficient. Temperature and precipitation time series produced by both weather generators will be investigated for applicability and suitability in the study watershed. Finally, a hydrological model, GWLF (Generalized Watershed Loading Functions, Haith et al., 1992), is applied with generated weather information from climate outlooks to predict stream flows in the Shihmen Reservoir Watershed. With climate outlooks assisted inflow projections, more informed risk-based management decisions can be practiced.