The term “short range” means different things to different users. We consider any horizon inside the range of reasonably accurate weather forecasts to be short range. That can range from the immediately next trading period out to fourteen days. Inside of this window we offer three distinct tools to collectively help you better understand what demand will do in the short range.
We combine available weather forecast data with the TESLA Model to generate highly accurate and reliable point forecasts for horizons extending to the limit of the available weather forecast. Because we fully realize that no model is perfect, we monitor the performance of our point forecasts against real-time observations. If we are tending to over-forecast, our systems are smart enough to dampen our forecasts over the next 24 to 48 hours, and vice-versa.
Our systems can use the same TESLA Model to generate a point forecast for each independent weather forecast that you may have access to. The resulting forecast ensembles can be very useful, providing an envelope of probable demand scenarios.
We have developed algorithms to scan the available histories of observed weather data to identify as many as five days in the past that are meteorologically comparable to the forecast day. Our system will then display the observed demand data from those days alongside our point forecasts. While demand systems do change over time, the Comparable Days tool can lend valuable insight into how a system is capable of responding to specific weather scenarios.
To demonstrate the potential sensitivity of a given demand or generation system to fluctuations in temperature, we routinely generate multiple demand forecasts using a single weather forecast by raising and lowering the temperature at every hour by one degree at a time. The resulting ensemble can help you determine whether warmer or cooler temperatures present a greater risk of demand fluctuations.