New Forecasting Research Presented at UVIG Workshop
As a pioneer and industry-leader in renewable energy forecasting, Vaisala is an active member of the energy integration and research community. At this week's UVIG (Utility Variable-Generation Integration Group) Forecasting Workshop in Denver, CO our Dr. Eric Grimit, Manager of Energy Forecasting, was selected to present some of our latest findings with regard to seasonal wind forecasting and the state of regional solar forecasting in the CAISO market. Read on for more details.
The Rise of Utility-Scale Solar in the CAISO Market & Regional Forecasting
In 2015 utility-scale solar energy capacity exceeded wind energy capacity for the first time ever - 6.7% of total demand versus 5.3% of demand. Over the past five years solar capacity has experienced a fifteen-fold increase, growing exponentially, while wind growth has slowed or even declined. With more utility-scale solar energy on the system than ever before, its variability is becoming more and more important for managing integration and anticipating energy price volatility. Market participants need access to skilled forecasts to make informed scheduling and trading decisions and the public forecast is not always accurate enough - especially on cloudy days when anticipating utility-scale solar availability is even more critical.
A recent forecast validation study shows that Vaisala's new Regional Solar Forecast for CAISO is consistently more accurate than the day-ahead public forecast. For example, in CAISO SP-15 (South of Path 15), which accounts for a majority of the state's solar capacity, Vaisala's forecast successfully predicted reduced production days 60% more often than the public forecast over the last six months of 2015.
In this presentation, Dr. Grimit will show use cases demonstrating the value of a more accurate forecast as well as how having multiple perspectives helps the industry make smarter and more profitable energy trading and scheduling decisions.
Seasonal Forecasting - A Tale of Industry Need and Scientific Limitations
Wind owner-operators have a true need for high quality, long-term seasonal forecasts looking 1-12 months into the future to set budget plans across their portfolios, secure transmission contracts, and determine the need for financial tools like wind hedges to cover shortfalls. These decisions have a heavy impact on a company's financial health and any significant below average conditions over a quarter or longer period of time can cause concern amongst investors and boards.
In this presentation, Dr. Grimit will review the so-called "wind drought" of Q1 and Q2 2015 where wind speeds, especially in areas with a high concentration of operating wind farms, dropped to some of the lowest on record. This event had a significant influence on the industry and drove increased interest in accurate seasonal forecasts.
The record El Niño climate signal predicted for late 2015 and early 2016 promised to give forecasters greater certainty when making long-term forecasts of wind variability. This was due to the high probability that El Niño would trigger weather conditions we have seen during past El Niño events in the U.S. However, while the El Niño still set records, it did not play out as anticipated in terms of the usual weather patterns. Instead it manifested as what is being described by scientists as a "different flavor" of El Niño or what could possibly be a blend of different sea surface temperature patterns.
At the podium, Dr. Grimit will also be presenting the findings of a seasonal forecasting study looking at a hypothetical portfolio of projects and comparing a benchmark method against a new experimental method that blends an expanded set of predictors with machine learning. In this experiment, Vaisala saw marked improvements in accuracy using the experimental approach versus the benchmark approach, especially when looking at the 3-month outlook. While these results are encouraging, Vaisala's research team says there is still much room for improvement over the current state-of-the-art in seasonal forecasting.