Toward Integration of Seasonal Climate Forecasts into Energy Decision Support Systems

Weather-sensitive applications including transport, energy and agriculture can benefit from improved weather information. For instance, the energy sector concerns of cooling demands in summer and heating demands in winter require months of advanced planning. Such demands are tied to the human comfort-zone, which is a window defined by temperature [15°C to 25°C] and relative humidity [40% to 60%]. The purpose of the present work is to understand what weather conditions outside the comfort zone are of specific interest to the energy sector. In this study, we apply Kernel Density Estimation (KDE) to temperature and relative humidity to represent the distribution of these variables and calculate the percentage of time the weather is outside of the comfort zone. The KDE analysis is performed on 35 years (1980-2014) of the Climate Forecast System (CFS) reanalysis data to understand the distribution patterns, its interannual variability and extreme years. In the interest of energy planning, the same technique is then applied to the CFS sub-seasonal and seasonal ensemble prediction, providing forecasts and uncertainty estimates months in advance.

The CFS resolution (half-degree) and number of ensemble members (four) may not be sufficient for specific needs in some regions. We therefore take the state-of-art WRF model and downscale the CFS seasonal forecast, creating higher resolution predictions with more ensemble members to provide improved probability forecast. Applying KDE analysis on these WRF forecasts will demonstrate the added value of multiple ensemble downscaling.