Offshore wind zones are reaching sizes at which they start to affect each other and potentially also alter mesoscale weather systems, impacting the energy production. Here, we assess the impact of future wind farm characteristics, like turbine type and capacity density, on cluster-scale wake losses. For this we use the mesoscale model COSMO-CLM at the km-scale resolution, which skillfully models frequency distributions of wind speed and wind direction at turbine level compared to measurement masts, wind lidars and satellite data. It was found that inter-farm wakes can reduce the long-term capacity factor at the inflow edge of wind farms from 59% to between 55% and 40% depending on the degree of clustering and the size of the upwind farms, for a layout equipped with 5MW turbines at a capacity density of 8.1 MW / km². Moving to next-generation wind turbines (15MW) partly mitigates this degradation, as the total generation over all windfarms (TWh) is increased by 19% under the same wind farm capacity density. On the other hand, increases in the capacity density in this future layout lead to a less than proportional (0.8 to 1) increase in the basin-integrated, total generation as a consequence of more intense intra- and inter-farm wake effects. Generally, wind farm characteristics play an essential role in inter-farm wake losses, which should be included in future wind farm planning.

How to cite: van Lipzig, N. and Borgers, R.: Impact assessment of future wind farm characteristics on cluster-scale wake losses in the North Sea, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6943, https://doi.org/10.5194/egusphere-egu23-6943, 2023.

This study explores the potential of slow-varying components of the earth system to predict the monthly mean wind speeds over the seven homogenous climate zones of India at subseasonal to seasonal time-scales. The following set of predictors are selected for that purpose: sea-surface temperature, mean sea-level pressure, 10 m wind speed, wind speed at 850 hPa, and geopotential height at 850 hPa. With the exception of sea-surface temperature which is obtained from HadISST, the rest of the variables are obtained from the JRA55. Besides, the popular indices such as the Nino 3.4 index and the Dipole mode index are also used as predictors. The forecasts are made at 1, 2, 3, 4, and 5 months of leadtime for the monsoon months of June, July, August, and September when the wind speeds are the highest throughout the country. The regions of significant correlations of the predictor fields with the spatially-averaged wind speeds of each homogenous region are determined using the past 6 month lagged composites. Once identified, the variables over these regions are spatially averaged and are mapped to the 10 m wind speeds from JRA55, since it is the closest representation of observed wind speeds over India. This predictor-based forecasting is carried out using the following approaches: multi-linear regression, decision tree based regression, and K nearest neighbours regression. The models use data from 1958-2018 for training and 2019-2021 for testing. The deterministic predictions are evaluated using mean absolute error (MAE) and the skill compared to a climatological forecast is estimated using the root mean squared error skill score (RMSESS). Results show that different sets of predictor combinations are responsible for giving the best forecasts for individual months and leadtimes. These forecasts have MAE of  around 0.2 m/s and RMSESS values ranging from 0.5-0.7. Although we are looking at deterministic predictions here, a combination of multiple models and predictors used above can lead to the production of ensemble forecasts as well, which will be of further added value to the wind energy sector.

How to cite: Das, A. and Baidya Roy, S.: Exploiting the predictability of global teleconnections to forecast subseasonal to seasonal scale wind speeds over India, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-507, https://doi.org/10.5194/egusphere-egu23-507, 2023.

Wind energy is essential in many decarbonization strategies and potentially vulnerable to climate change. While existing wind climate change assessments rely on regional or global climate models, a systematic investigation of the global-to-regional climate modeling chain is missing. In this presentation, I therefore address the differences in climate change impacts on winds according to  regional and global climate model ensembles under three different future scenarios.

I highlight two key limitations, namely (a) the differing representation of land-use change in global and regional climate models which compromises comparability, and (b) the consistency of large-scale features along the global-to-regional climate modeling chain. To this end, I analyze the large EURO-CORDEX ensemble (rcp85: N=49; rcp45: N=18; rcp26: N=22) along with the driving global models (rcp85: N=7; rcp45: N=5; rcp26: N=7), finding evidence that climate change reduces mean wind speeds by up to -0.8 m/s (offshore) and -0.3 m/s (onshore).

Moreover, I provide physical explanations for these changes by identifying two key drivers. First, onshore wind speeds drop in the driving global models in regions and scenarios with strong land use change but show no drop in EURO-CORDEX where land use is held constant. Second, offshore wind reductions follow decreases in the equator-to-pole temperature gradient remarkably well with correlations reaching around 0.9 in resource-rich European countries like Ireland, the United Kingdom and Norway, implying that arctic amplification is a severe risk for European offshore wind energy.

My results suggest that earlier conclusions of negligible climate change impacts on wind energy might be premature if either land use changes strongly or polar amplification is at or above the range sampled in global climate models. In conjunction with earlier work that demonstrated the relevance of multidecadal wind fluctuations caused by climate variability, these results call for a better inclusion of climate risk in wind energy planning.

Reference

Wohland, J. Process-based climate change assessment for European winds using EURO-CORDEX and global models. Environ. Res. Lett. (2022) doi:10.1088/1748-9326/aca77f.

How to cite: Wohland, J.: Climate change impacts on winds in Europe: do global and regional climate models tell the same story?, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-2317, https://doi.org/10.5194/egusphere-egu23-2317, 2023.

Currently, to include the effects of wind farms in the Weather Research and Forecasting (WRF) model, a common choice is to use the Fitch wind farm parametrisation (WFP). This WFP has long been implemented into the WRF's standard repository and has been the subject of several wind resource assessment studies. However, one of the disadvantages of its current (WRF version 4.4) set-up is that it is constrained to one Planetary Boundary Layer (PBL) parametrisation. The Fitch scheme is coupled to the Mellor-Yamada Nakanishi Niino (MYNN) PBL parametrisation because it can inject Turbulent Kinetic Energy (TKE) from the turbines into the atmosphere. More importantly, it is the only PBL where the TKE advection can be activated. This feature is essential since it stores the TKE from one time step to the next and prevents the high TKE concentration at the turbine's location.

One way for the WFPs to become PBL-independent is to move away from focusing on the TKE source term and parametrise the turbulence in some other way. The Explicit Wake Parametrisation (EWP) is a WFP coupled to WRF that, as opposed to the Fitch scheme, does not include an explicit TKE source term and the turbulence is produced via enhanced vertical shear. The EWP is based on the assumption that the advection and diffusion terms in the RANS Navier-Stokes equations dominate the development of the wake. As a result, the drag equation is also related to the diffusion term from a 1.5 turbulence closure. The EWP then needs the turbine's information, wind speed and the turbulent diffusivity coefficient (K_m) from the PBLs to calculate the wind deficit. Given the latter, the EWP can work if K_m is present and comes from at least a 1.5-order turbulence closure PBL. However, studies have yet to attempt to prove this feature since it has only been used with the MYNN scheme.

In this study, we demonstrate the use of the EWP in WRF when other PBL schemes are used and the implications of this approach. We demonstrate this implementation under ideal neutral conditions with similar setups and forcings (surface roughness length, Coriolis parameter and hub-height wind speed) for two local (1.5-order closure) PBL schemes. Similarly, we test the possibility of coupling EWP into two non-local PBL schemes (first-order closure). The study focuses on the wake recovery behaviour, the drag strength and the power produced by an idealized wind farm under the four PBL schemes. Early results show faster wake recovery from non-local PBls than local ones, which could be related to the diffusivity coefficient values and the PBL's mixing rates.

How to cite: García-Santiago, O., Badger, J., and Hahmann, A. N.: Wind farm wake recovery under different Planetary Boundary Layer schemes in WRF, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-14911, https://doi.org/10.5194/egusphere-egu23-14911, 2023.

Wind ramps, or rapid changes in wind speed, are a crucial aspect of atmospheric dynamics and
have significant implications for various wind energy applications. For example, wind ramps tend
to increase uncertainty in power output predictions. Furthermore, they also induce fatigue damage
to wind turbines.

In a recent study, DeMarco and Basu (2018; Wind Energy) used long-term observational
data from four geographical locations to characterize the tails of the wind ramp probability
distribution functions (pdfs). They showed that the pdfs from these various sites (ranging from
offshore to complex terrain) portray quasi-universal behavior. The tails of the pdfs are much
heavier than the Gaussian pdf and decay faster with increasing time increments. The tail-index
statistics, computed via the so-called Hill plots, exhibited minimal height dependency up to
approximately one hundred meters above the land or sea surface level. However, wind ramp
statistics at higher altitudes at Cabauw (the Netherlands) were quite distinct.

In the present study, we investigate if state-of-the-art reanalysis datasets capture the
intrinsic traits of wind ramp pdfs. Specifically, we make use of the newly released Copernicus
European Regional ReAnalysis (CERRA) dataset in conjunction with the popular fifth-generation
ECMWF reanalysis (ERA5) dataset. These datasets allow us to describe the characteristics of wind
ramp pdfs at high altitudes (up to 500 m). Given the disparity of the spatial resolution of CERRA
(~5.5 km) and ERA5 (~32 km) datasets, we are also able to demonstrate the impact of spatial
resolution on simulated tail index characteristics. Lastly, the influence of natural climate patterns
such as El-Nino and La-Nina on wind ramp pdfs are examined.

How to cite: Baki, H., Basu, S., and Lavidas, G.: Statistical characterization of simulated wind ramps, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-17208, https://doi.org/10.5194/egusphere-egu23-17208, 2023.

Probabilistic forecasts based on ensemble simulations of numerical weather prediction models have become a standard tool in weather forecasting and various application areas. However, ensemble forecasting systems tend to exhibit systematic errors such as biases, and fail to correctly quantify forecast uncertainty. Therefore, a variety of post-processing methods has been developed to correct these errors and improve predictions [1]. In particular, machine learning methods based on neural networks have been demonstrated to lead to substantial improvements compared to classical statistical techniques [2].
While post-processing can successfully correct the biases and dispersion errors in the weather variables, its effect but has not been evaluated thoroughly in the context of subsequent forecasts, such as wind and solar power generation forecasts and it is not obvious how to best propagate forecast uncertainty through to subsequent power forecasting models. Therefore, the work presented here will evaluate multiple strategies for applying ensemble post-processing to probabilistic wind and solar power forecasts. We use Ensemble Model Output Statistics (EMOS) as the post-processing method and evaluate four possible strategies: only using the raw ensembles without post-processing, a one-step strategy where only the weather ensembles are post-processed, a one-step strategy where we only post-process the power ensembles and a two-step strategy where we post-process both the weather and power ensembles. The presentation is based on recent work in Phipps et al. (2022) [3] and ongoing other work.

References

[1] Vannitsem, S., et al. (2021). Statistical Postprocessing for Weather Forecasts - Review, Challenges and Avenues in a Big Data World. Bulletin of the American Meteorological Society, 102, E681–E699.
[2] Rasp, S. and Lerch, S. (2018). Neural networks for post-processing ensemble weather forecasts. Monthly Weather Review, 146, 3885–3900.
[3] Phipps, K., Lerch, S., Andersson, M., Mikut, R., Hagenmeyer, V. and Ludwig, N. (2022). Evaluating ensemble post-processing for wind power forecasts. Wind Energy, 25, 1379-1405. 

How to cite: Lerch, S.: Evaluating ensemble post-processing for probabilistic energy prediction, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6644, https://doi.org/10.5194/egusphere-egu23-6644, 2023.

Hydrogen as a versatile, greenhouse gas-free energy carrier will play an important role in our future economy. Yet sustainable, competitive production and distribution of hydrogen remains a challenge. Highly integrated solar water splitting systems aim to combine solar energy harvesting and electrolysis in a single device, similar to a photovoltaic module.^[1] Such a system can produce hydrogen locally without the requirement to be connected to the electricity grid. Unlike large electrolysis that draws power from the grid, the power density of such a device is reduced so far that it does not require active cooling, but its operating temperature will closely follow outdoor conditions.

Here, we present our work on high-efficiency integrated solar water splitting devices based on multi-junction solar absorbers. The light-absorbing component is sensitive to the shape of the solar spectrum and generally becomes more efficient at lower temperatures. Catalysis, on the other hand, benefits from higher temperatures. These conflicting trends wih respect to the temperature impact the design of the solar hydrogen production system. We analyse how the local climate affects production efficiency^[2] and show in a lab study that adequate system design allows efficient operation at temperatures as low as -20°C.^[3] These insights can help to design small-scale distributed solar hydrogen production for both temperate regions, but also more extreme climatic conditions.

[1] M.M. May et al., Nature Communications 6 (2015), 8286.
[2] M. Kölbach et al, Sustainable Energy & Fuels 6 (2022), 4062.
[3] M. Kölbach, K. Rehfeld, M.M. May, Energy & Environmental Sciences 14 (2021), 4410-4417.

How to cite: May, M. M., Schmitt, E., Grabenstein, J., Höhn, O., Barry, J., Kölbach, M., and Rehfeld, K.: Integrated solar hydrogen production: Impact of the local climate, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-14800, https://doi.org/10.5194/egusphere-egu23-14800, 2023.

In recent years, solar power applications are growing rapidly worldwide, to meet the increasing power demand and the sustainable development planning. Estimation of solar radiation availability at surface level, its characteristics and various factors that affect it, play a key role in designing and achieving the optimal performance of systems employing solar energy. Various solar -PV related - applications are using radiative transfer modeling to characterize the radiation field, since accurate surface solar irradiance measurements are not always available, especially in remote regions. Understanding the effect of aerosols to the solar energy potential is highly important for the energy sector as well as for a variety of fields.    In areas and periods where cloudiness is limited and they are in the proximity of particle sources, the significance of aerosol effect is very high.

The objective of this study is to assess the impact of the variability of aerosols on the solar Direct Normal Irradiance (DNI), Global Horizontal Irradiance (GHI) and solar energy, using spectral solar measurements and aerosol optical properties retrievals, in the framework of the one-year experimental campaign (December 2020-December 2021) of the ASPIRE (Atmospheric parameters affecting SPectral solar IRradiance and solar Energy, https://aspire.geol.uoa.gr) project, which was held in Athens, Greece.

Main findings include an assessment of differences among different PV technology and their calculated outputs using actual and standard spectra, linking the differences with aerosol optical properties (optical depth, spectral dependence, absorption). Aerosol optical depth is the major factor of such differences for all PV technologies. Spectral aerosol characteristics affect differently PV technologies as a consequence of different spectral responsivities.

Finally, aerosol effect on solar nowcasting models have been investigated by comparing spectral solar measurements and aerosol properties with model inputs and outputs.

How to cite: Kouklaki, D., Raptis, I.-P., Kazadzis, S., Fountoulakis, I., Papachristopoulou, K., and Eleftheratos, K.: The Aspire campaign: Assessing the effects of aerosols on solar radiation and energy in SE Europe., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5952, https://doi.org/10.5194/egusphere-egu23-5952, 2023.

Solar energy from photovoltaics (PV) is a major contributor to the power production, e.g., in Germany, with a growing share. It is a major contributor to renewable power production but highly volatile as it is heavily influenced by atmospheric conditions. Especially shading by clouds can change within seconds to minutes and cause ramps in irradiance and solar power production. Accurate short-term predictions (nowcasts) of irradiance for the next minutes can help to alleviate the impact of this volatility and improve the integration of solar power into energy grids. One approach for nowcasting is the use of all-sky imagers (ASI), ground based fisheye cameras which capture the current cloud situation. Therefore, cloud information is extracted from current images, future cloud states are extrapolated and converted into an irradiance nowcast. Despite substantial progress in the quality of the applied methods, current ASI nowcasting models still exhibit significant nowcast errors and struggle to reliably outperform persistence nowcasts for all situations. Therefore, we assessed the implications for nowcast performance of two common fundamental simplifications of ASI nowcasting models. Firstly, cloud evolution is often modelled by advection, i.e. simple displacement over time. Growth, shrinking or reshaping of clouds is usually neglected in the models. Additionally, the ASI viewing geometry may introduce a misrepresentation of the depicted cloud scene, which is also commonly neglected. The ASI views surrounding clouds from a single ground position and under varying angles. For direct irradiance however, the horizontal distribution of clouds and their intersection in the direction of the sun is essential. While ASI images are usually reprojected to comply with the required horizontal representation, the original difference in actual and required viewing geometry cannot be fully compensated. E.g., breaks between distant clouds may not be clearly visible by the ASI although modulating the irradiance. Kurtz et al. (2017) demonstrated a major impact by this geometric limitation. We applied a nowcasting model to synthetic ASI images of a simulated cloud scene to extend this previous study and analyze the errors introduced by both of the two commonly used simplifications of ASI nowcasting models. A large fraction of the nowcasting error is attributable to the simplifications, which implies a systematic baseline error of common ASI nowcasting models. While the implementation of more evolved cloud evolution and a better representation of relevant cloud geometry are challenging, this work indicates, that efforts to implement these improvements in ASI nowcasting models are a chance for a leap in performance of future nowcasting models.

Kurtz, B., Mejia, F., and Kleissl, J.: A virtual sky imager testbed for solar energy forecasting, Solar Energy, 158, 753–759, https://doi.org/10.1016/j.solener.2017.10.036, 2017.

How to cite: Gregor, P., Zinner, T., and Mayer, B.: How good can we get? – An analysis of systematic errors in common models for all-sky imager based irradiance nowcasting for solar energy, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-14645, https://doi.org/10.5194/egusphere-egu23-14645, 2023.

In 2021, Saudi Arabia, a leading global oil producer, announced its Middle East Green Initiative with many objectives including reducing carbon emissions by divagating the country away from an oil-based economy and towards renewable. Saudi Arabia has a high potential to become one of the global largest solar energy producers, as it is geographically located on a sunbelt. By 2030, the Saudi government targets building eight solar plants across the country which are expected to produce more than 3,600 MW, enough to power more than 500, 000 homes. However, the vast desert environment in Saudi Arabia increases the dust and aerosol loadings in the atmosphere, which affects the performance of the solar irradiance performance of photovoltaic panels due to the scattering of the solar radiation and the dust deposition on the solar panels. In this work, ground-based data from weather stations located in five Saudi cities: Dammam, Riyadh, Jeddah, Najran, and Arar along with data from the Moderate Resolution Imaging Spectroradiometer (MODIS) are used to estimate solar irradiance and its correlation with atmospheric and meteorological conditions like air temperature, wind, and aerosol physical parameters. We investigate the effect of three major dust storms that blew over different regions in Saudi Arabia on 20 March 2017, 23 April 2018, and 15 April 2021 on solar irradiance. It is found that there is a strong correlation between aerosol optical parameters like Aerosol Optical Depth (AOD), Ångström exponent, and solar irradiance. Maximum AOD (about 2) is recorded over Jeddah on 19 March 2017, (about 2.3) over Riyadh on 20 March 2017, (about 1.5) over Riyadh on 24 April 2018, and (about 0.9) over Najran on 15 April 2021. Large dust events are found to reduce air temperature by a few degrees in the regions affected by dust loadings. The study found large dust loading decreases the DNI, and GHI components on the solar irradiance, while increasing the DHI component over the cities of Jeddah, Riyadh, and Najran. This could be an indication that scattering from dust particles could play a significant role in the solar irradiance intensity. 

How to cite: Labban, A. and Farahat, A.: Effect of Major Dust Events on Atmospheric Temperature and Solar Irradiance Components over Saudi Arabia, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-1854, https://doi.org/10.5194/egusphere-egu23-1854, 2023.

Solar eclipse causes high magnitude fluctuations in the Surface Solar Irradiance (SSI) for a short duration and consequently reduces the output of solar PV systems. Grid operators try to estimate the impending loss in PV power generation prior to the occurrence of an eclipse in order to schedule conventional generators for compensating the loss. The worldwide installed capacity of grid connected solar PV systems is expected to steeply rise in the coming decade as a result of the various policy initiatives aimed to tackle the climate change. In future electric supply networks with a high penetration of solar PV systems, such large ramps in generation could impact the stability of the network. Although a solar eclipse is a purely deterministic phenomenon, it’s impact on the satellite retrieval of Surface Solar Irradiance (SSI) is complicated due to the possibility of cloud presence in the regions affected by the eclipse. The extraterrestrial solar irradiance is reduced by the moon during an eclipse. On the one hand this causes clouds to appear darker and they get assigned lower reflectance values than they should have in reality. This leads to predicting higher values for the solar irradiance under these clouds than expected. On the other hand, the eclipse also reduces the clear sky irradiance reaching the earth surface. We developed a method to make corrections for both of these effects on the High Resolution Visible (HRV) channel images from Meteosat-11 The results are validated against ground measurements of irradiance provided by BSRN, IEA-PVPS, DTN and the National Weather Services networks. The validation is performed for sites with locations across Europe and for the last two eclipses.  

How to cite: Roy, A., Hammer, A., Schroedter-Homscheidt, M., Lezaca, J., Azam, F., Lünsdorf, O., Heinemann, D., and Saint-Drenan, Y.-M.: Improving the satellite retrieval of surface solar irradiance during an eclipse, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-17290, https://doi.org/10.5194/egusphere-egu23-17290, 2023.

The European Horizon 2020 project Smart4RES (http://www.smart4res.eu), which started in 2019 and runs until April 2023, aims at improving modelling and forecasting of weather variables necessary to optimize the integration of weather-dependent renewable energy (RES) production (i.e. wind, solar) into power systems and electricity markets. It gathers experts from several disciplines ranging from meteorology, data science, power systems a.o. It aims to contribute to the pathway towards energy systems with very high RES penetrations by 2030 and beyond.

This presentation has a double objective:

(1) To present a comprehensive overview in terms of KPI improvements of the final results obtained by the project. These results cover thematic objectives including:

• Improvement of weather and RES forecasting;
• Streamlined extraction of optimal value from the data through data sharing, data market places, and novel business models for the data;
• New data-driven optimization and decision-aid tools for market and grid management applications;
• Validation of new models in living labs and assessment of forecasting value vs costly remedies to hedge uncertainties (i.e. storage). 

The results obtained are numerous. Without being exhaustive, they include: improved forecasting of weather variables with focus on extreme situations and also through innovative measuring settings (i.e. a network of sky cameras); A seamless approach to couple outputs from different ensemble numerical weather prediction (NWP) models with different temporal resolutions; Advances from ultra-high resolution NWPs based on Large Eddy Simulation; Approaches for RES production forecasting aiming at efficiently combining highly dimensionally input (various types of satellite images, NWPs, spatially distributed measurements etc.); Seamless probabilistic RES forecasting covering multiple time frames and data inputs; Resilient energy forecasting. In the front of applications methods are proposed to optimally use forecasts for the management of storage systems coupled with renewables, for the optimal trading of renewables in multiple markets and for grid management optimization and dynamic security assessment. Prescriptive analytics and explainable AI methods are proposed to optimize decision making.  A cost benefit analysis is performed to assess the contribution of different types of data in forecasting problems.

(2) To present hierarchized proposals for future research directions. An international workshop is organized by the project (14/04/2023), where experts are invited to assess where RES predictability stands today and propose research directions for the future. In this presentation we will present the conclusions of this workshop. This will be a useful insight for academics, industrials as well as policy makers in the field.

How to cite: Kariniotakis, G. and Camal, S. and the Smart4RES Team: Renewable energy forecasting: results of the Smart4RES project and future research directions., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9658, https://doi.org/10.5194/egusphere-egu23-9658, 2023.

In the past decade, significant advances were made in improving the S2S and seasonal prediction using mainly numerical weather prediction models (NWP) and in some cases climate models for generating the predictions. Recently, the application of these models in real time forecasting through the S2S Real-Time Pilot Initiative (Robbins et al., 2020) was evaluated and is ongoing. There are, however, drawbacks. Computational costs for performing one forecast cycle are high (RAM, storage, ensemble for uncertainty) and limit the spatial, and to some extent temporal, resolution which are currently roughly 1.5° in spatial and at most 6-hourly in temporal resolution. Both resolutions are not sufficient for small scale renewable production sites.

In renewable energy applications, these time scales are getting more important as they can adapt their resource management strategies based on predictions of possible load/heating and cooling demand via anomalies to temperature, wind, precipitation amount, effects on the markets can be better estimated for trading, and scheduling of maintenance works. Thus, at least higher spatial resolutions could help improving the management and planning of these tasks.

Within the SSSEA project (SubSeasonal to Seasonal Ensemble prediction and Application), in project phase I, different methods for post-processing and downscaling the S2S challenge data to 1 km resolution and actual values instead of anomalies were implemented. The statistical methods EPISODES, GMOS, and SAMOS were adapted to be able to work with different time scales compared to their initial implementations (seasonal/hourly) and machine learning based methods were developed from scratch using a feed forward neural network, a Unet-based model, and a Random Forest. Temperature, precipitation, and in the currently ongoing project phase II, the wind components of the ECMWF S2S model were downscaled to daily analysis fields based on the INCA model.

For wind energy applications, specific indices were developed and applied to the downscaled results.  Verification and definition of suitable metrics is crucial to assess the skills of the different methodologies considered and a wide range of aspects and metrics were considered. Results on both grid and station verification for appx. 250 sites in Austria across nearly all altitude ranges show that all post-processing models are able to improve the ECMWF ensemble forecasts for the parameters considered, though, depending on lead time and season, differences in the models’ skill are visible. Furthermore, for most of the initial times and leadtimes in the forecast/testing period of 2020 we were able to outperform also the climatology. To assess the impact on renewable energy production, different indices were derived and evaluated with focus on wind energy and hydrology in project phases I and II. Results of SSSEA show clearly the added value of the post-processed and downscaled subseasonal predictions for both parameters and specified indices.

How to cite: Schicker, I., Dabernig, M., Papazek, P., Schellander-Gorgas, T., and Tiefgraber, M.: Post-processing and high-resolution downscaling of subseasonal ensemble forecasts with focus on renewables using statistics and machine learning, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12949, https://doi.org/10.5194/egusphere-egu23-12949, 2023.

Standard practice of decision-making in energy systems relies largely on complex modeling chains to address technical constraints and integrate numerous sources of uncertainty. The increased penetration of Renewable Energy Sources (RES) such as solar and wind plants adds complexity due to the weather dependency of their electricity production. Artificial Intelligence (AI) based tools have proven their efficiency in different applications in the energy sector ranging from forecasting to optimization and decision making. They permit to simplify modeling chains and to improve performance due to higher learning capabilities compared to state-of-the-art methods. However, decision-makers of the energy sector need to understand how decision-aid tools construct their outputs from the data. AI-based tools are often seen as black-box models and this penalizes their acceptability by end-users (traders, power system operators a.o.). The lack of interpretability of AI tools is a major challenge for the wider adoption of AI in the energy sector and a fundamental requirement to better support humans in the decision-aid process. Agents of energy systems expect very high levels of reliability for the various services they provide. As energy systems are impacted by multiple uncertainty sources (e.g. available power of RES plants, weather and meteorological conditions, market conditions), developed AI tools should not only be performant on average situations but be able to guarantee robust solutions in the case of an extreme event. Therefore, our research focuses on understandable representations of data-driven decision-aid models for human operators in the energy sector. In order to enhance the interpretability of the AI models, a technique borrowed from the computer science domain is explored and further developed. Genetic programming and more precisely Symbolic Regression is used to derive a symbolic representation for the data-driven model that can take the form of a single equation. This equation results according to a specific reward function. The optimal solutions are selected naturally mimicking the biological theory of survival of the fittest. The main outcome is the production of symbolic representations of the AI models that require minimum changes when applied to different case studies. In this presentation a real-world use case is considered, to demonstrate the added value of the proposed tools for decision-making when trading the production of wind and solar power plants to the day-ahead market. An annual period of data is considered to train and test the proposed model. The typical modeling chain involves as many as 12 models for forecasting RES production, weather and meteorological conditions, together with stochastic optimization to derive trading decisions. A single AI-based model here replaces this complex chain. Such simplification is a significant enhancement to the modeling chain interpretability and facilitates trust to the human decision-maker. This work is carried out in part in the frame of the European project Smart4RES (Grant No  864337) supported by the H2020 Framework Program and in part in the frame the Marie-Curie COFUND project Ai4theSciences (Grant No  945304)

How to cite: Parginos, K., Kariniotakis, G., Bessa, R., and Camal, S.: Towards a paradigm of explainable AI applied in energy meteorology, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13992, https://doi.org/10.5194/egusphere-egu23-13992, 2023.