Precipitation Forecasting for Hydrologic Modeling in West-Central Florida using Seasonal Climate Outlooks

Abstract. Seasonal precipitation forecasts play a vital role in short-term decision-making for water resources management, agriculture, and wildfire preparedness. NOAA's seasonal precipitation forecasts can be used at the local scale to further develop precipitation forecasts. Rather than evaluating the forecasting skill at the scale at which forecasts are provided, this study applies NOAA forecasts at the local basin scale and evaluates the skill of such localized forecasts. This study evaluates the skill of NOAA's 3-month precipitation outlooks at a 0.5-month lead for the Alafia and Hillsborough River Basins in west-central Florida, using hindcasts from 1995 to 2019. Forecast performance is assessed seasonally using categorical and probabilistic metrics against basin-scale observed precipitation from a Bayesian gauge-radar dataset. To translate categorical outlooks into basin-scale rainfall estimates, two non-parametric ensemble generation methods are introduced: Proportional Tercile Sampling (PTS) and Dominant Tercile Sampling (DTS). These methods sample from pre-generated rainfall realizations conditioned on seasonal forecasts to capture uncertainty and support operational planning. Results indicate that forecast skill peaks during the dry season (October to February), particularly for wet-tercile forecasts issued during El Niño years. DTS performs best during high-skill seasons by leveraging dominant climate signals, while PTS proves more reliable during low-skill periods. Based on these findings, a hybrid strategy is recommended: apply DTS during late fall and winter to capitalize on strong climate signals and use PTS during other seasons to maintain reliability and operational value. This study contributes a strategic approach to applying NOAA's forecasts in the study area and demonstrates that this method of applying NOAA's forecasts at the local scale is general and can be applied to other regions.