In the realm of meteorology, do weather models adapt to climate changes? At ConWX we rely significantly on high-quality weather model data to fuel our forecasting algorithms, which drive our advanced machine-learning processes for both Day-Ahead Hourly (DAH) and Intraday (ID) predictions. Coupled with our clients’ data, this synergy enables us to make near-future predictions. We are all well aware that weather operates chaotically, undergoing constant fluctuations. This becomes especially pronounced when considering the drivers of renewable energy.
Even slight wind fluctuations of just 1 m/s can introduce a staggering 30% uncertainty in power production, particularly on the steep segments of the production curve. Similarly, solar energy production is heavily impacted, with an unexpected cloud causing a temporary yet substantial decline in output. Furthermore, the occurrence of massive summer storms casts shadows of uncertainty, creating imbalances that persist for hours. The considerable investments in renewables have led to instances of negative prices during noon in various markets this summer. With volatility on the rise each year, traders face increasingly challenging decisions.
It’s a fundamental truth that we can never achieve 100% accuracy in predicting tomorrow’s weather, and forecasts should always be perceived within the context of probabilities. Currently, we are in the final stages of refining our icing algorithms for power loss forecasts in the upcoming season. In this domain, dealing with false positives presents a significant challenge, and predicting power loss due to icing is even more intricate. Nonetheless, through the collaborative efforts of adept employees and cooperative clients, we have managed to produce icing forecasts with minimal false positive icing alerts and accurate power loss predictions. These specialized forecasts save millions of Euros annually for our clients and play a pivotal role in grid stability.
Returning to the topic of climate change and its impact on our weather models: As climate change progresses, the atmosphere becomes host to heightened energy potential. This energy becomes the driving force behind colossal storms, distinct from those of yesteryears that originated from low-pressure systems. Instead, we now witness massive thunderstorms laden with destructive winds and hail. These storms are not only becoming more frequent but their intensity and destructiveness are also on the rise.
In our pursuit of accurate forecasts, we not only purchase and utilize external weather models but also operate an in-house model setup. This practice spans three continents where we provide power forecasting services. The motivation behind this approach is clear: We aim to furnish our customers not only with superior average forecasts but also alternative solutions in cases where mainstream models within the sector falter. Clients granted access to our models receive supplementary insights, often employed to compensate for deviations from the industry-standard model provided by the European Centre for Medium-Range Weather Forecasts (ECMWF). This approach can mitigate the costs of market imbalances, a benefit that resonates across the industry.
Additionally, our advanced weather models, particularly one designed by Ivan Ristic featuring uniquely developed IR micro and convective schemes, excel at storm prediction. Returning to the initial analogy, much like anticipating boiling water in a pot, we can identify the likelihood and potential intensity of convective storms. Yet, the exact timing and location remain elusive. Interestingly, the substantial storms that struck Stuttgart on 23.08.2023 and Croatia in July eluded prediction by global models.
However, our model equipped with innovative convective schemes rose to the challenge admirably. A model can never predict all details due to the chaotic process, but risks can be highlighted and forecasting will not be easier with more energy in the atmosphere from the climate changes. Especially in the zones between the highly stable heating areas and the unsettled areas to the north.
This is where most of us live. This not only aids industries during extreme events but also holds the potential for providing valuable pre-warnings to society at large.
