Characterization and forecasting of national-scale solar power ramp events
Authors: Luca Lanzilao, Angela Meyer
Organizations: School of Engineering and Computer Science, Bern University of Applied Sciences, Quellgasse 21, Biel, 2501, Bern, Switzerland · Department of Geosciences and Remote Sensing, TU Delft, Stevinweg 1, Delft, 2628 CN, South-Holland, The Netherlands
The rapid growth of solar energy is reshaping power system operations and increasing the complexity of grid management. As photovoltaic (PV) capacity expands, short-term fluctuations in PV generation introduce substantial operational uncertainty. At the same time, solar power ramp events intensify risks of grid instability and unplanned outages due to sudden large power fluctuations. Accurate identification, forecasting and mitigation of solar ramp events are therefore critical to maintaining grid stability. In this study, we analyze two years of PV power production from 6434 PV stations at 15-minute resolution. We develop quantitative metrics to define solar ramp events and systematically characterize their occurrence, frequency, and magnitude at a national scale. Furthermore, we examine the meteorological drivers of ramp events, highlighting the role of mesoscale cloud systems. In particular, we observe that ramp-up events are typically associated with cloud dissipation during the morning, while ramp-down events commonly occur when cloud cover increases in the afternoon. Additionally, we adopt a recently developed spatiotemporal forecasting framework to evaluate both deterministic and probabilistic PV power forecasts derived from deep learning and physics-based models, including SolarSTEPS, SHADECast, IrradianceNet, and IFS-ENS. The results show that SHADECast is the most reliable model, achieving a CRPS 10.8% lower than that of SolarSTEPS at a two-hour lead time. Nonetheless, state-of-the-art nowcasting models struggle to capture ramp dynamics, with forecast RMSE increasing by up to 50% compared to normal operating conditions. Overall, these results emphasize the need for improved high-resolution spatiotemporal modelling to enhance ramp prediction skill and support the reliable integration of large-scale solar generation into power systems.
Figures & tables
Figure 1: (a) Satellite-based SSI field and (b) PV power output of the 6434 PV stations, observed on 26 June 2020 at 14:00 UTC. Note that the PV power output is normalized using the station-specific 95th percentile of the power time series. (c) Elevation map showing values in meters above sea level. The black and red lines denote national borders.
Figure 2: (a) Time series of aggregated PV power production from 6434 PV stations, normalized by the maximum value observed over the displayed period, and (b) the corresponding magnitude of power variations, normalized by the ramp threshold ΔPtr . Results are shown for the first two weeks of July 2020. The horizontal dashed red line indicates the threshold used to identify ramp events, while the vertical dashed grey lines mark midnight of each day.
Figure 3: (a) Empirical probability density of magnitude of the power variation over a time interval of 15 minutes normalized by the threshold ΔPtr , measured over the year 2019-2020 for the aggregate PV power generation of 6434 PV stations. (b) Number of ramp events aggregated by hour of the day, with distinction between ramp-up and ramp-down events.
Figure 4: Hexagonal bin distribution of ramp magnitude normalized with the threshold ΔPtr as a function of hour of the day for (a) MAM, (b) JJA, (c) SON, and (d) DJF. Color shading indicates the number of samples per hexagonal bin on a logarithmic scale. The dashed white line traces the locus of maximum sample density at each hour, highlighting the most probable ramp magnitude throughout the diurnal cycle. The dashed red line marks the threshold used to identify ramp events. The bin widths are 15 minutes along the x-axis and 0.028 in normalized ramp units along the y-axis.
Figure 5: (a-c) Satellite-based SSI fields and (d-f) PV power output of the 6434 PV stations, observed on 23 May 2020 at three time stamps: 09:00 UTC, 10:00 UTC and 11:00 UTC. The black lines denote national borders. Note that the PV power output is normalized using the station-specific 95th percentile of the power time series.
Figure 6: (a,b) nRMSE and (d,e) nCRPS averaged across all PV stations and over the subsets of forecasts corresponding to (a,d) ramp events and (b,e) non-ramp events. Panels (c,f) show the relative difference in nRMSE and nCRPS, respectively, between forecasts during ramp and non-ramp conditions. The relative difference is defined as (mA−mB)/mB , where m denotes a generic metric and A and B refer to ramp and non-ramp subsets, respectively. For deterministic models, nCRPS is replaced by nMAE. Dashed–dotted lines indicate deterministic models, while dashed lines represent probabilistic models.
Figure 7: Satellite-based SSI observations and model forecasts over the area of interest at lead times ranging from 15 to 120 minutes. The forecasts are issued at 09:00 UTC on 23 May 2020. For the probabilistic models, the ensemble member chosen is the one with the lowest RMSE. Note that the IFS-ENS SSI fields represent a one-hour temporal average, whereas all other fields are instantaneous. The black lines denote national borders. The row highlighted with the red label is used as the ground truth. Missing panels reflect differences in the temporal resolution of the satellite observations and forecast models. For illustration purposes, the SolarSTEPS, SolarSTEPS-pa and IrradianceNet models are abbreviated to SolarS, SolarS-pa and IrrNet, respectively.
Figure 8: PV power measurements and model forecasts at lead times ranging from 15 to 120 minutes. The forecasts are issued at 09:00 UTC on 23 May 2020. For probabilistic models, the same ensemble member used in Figure 7 is shown. The row highlighted with the red label is used as the ground truth. Missing panels reflect differences in the temporal resolution of the satellite observations and forecast models. Note that the PV power output is normalized using the station-specific 95th percentile of the power time series.
Figure 9: Satellite-based SSI observations and model forecasts over the area of interest at lead times ranging from 15 to 120 minutes. The forecasts are issued at 11:00 UTC on 3 October 2020. For the probabilistic models, the ensemble member chosen is the one with the lowest RMSE. Note that the IFS-ENS SSI fields represent a one-hour temporal average, whereas all other fields are instantaneous. The black lines denote national borders. The row highlighted with the red label is used as the ground truth. Missing panels reflect differences in the temporal resolution of the satellite observations and forecast models.
Figure 10: PV power measurements and model forecasts at lead times ranging from 15 to 120 minutes. The forecasts are issued at 11:00 UTC on 3 October 2020, a day with high-variability weather. For probabilistic models, the same ensemble member used in Figure 9 is shown. The black lines denote national borders. The row highlighted with the red label is used as the ground truth. Missing panels reflect differences in the temporal resolution of the satellite observations and forecast models. Note that the PV power output is normalized using the station-specific 95th percentile of the power time series.
Figure 11: Satellite-based SSI observations and model forecasts over the area of interest at lead times ranging from 15 to 120 minutes. The forecasts are issued at 08:00 UTC on 7 June 2019. For the probabilistic models, the ensemble member chosen is the one with the lowest RMSE. Note that the IFS-ENS SSI fields represent a one-hour temporal average, whereas all other fields are instantaneous. The black lines denote national borders. The row highlighted with the red label is used as the ground truth. Missing panels reflect differences in the temporal resolution of the satellite observations and forecast models.
Figure 12: PV power measurements and model forecasts at lead times ranging from 15 to 120 minutes. The forecasts are issued at 08:00 UTC on 7 June 2019. For probabilistic models, the same ensemble member used in Figure 11 is shown. The black lines denote national borders. The row highlighted with the red label is used as the ground truth. Missing panels reflect differences in the temporal resolution of the satellite observations and forecast models. Note that the PV power output is normalized using the station-specific 95th percentile of the power time series.
Appendix figures & tables1 asset
Supplementary material from the paper’s appendix.
Appendix
Figure 13: Comparison of (a-c) SSI and (d-f) aggregated PV power forecasts computed for the three case studies discussed in Section 5. The SSI is spatially interpolated to each station location, averaged across all stations, and normalized by its maximum value within the corresponding 2-hour period. The PV power generated is summed over all stations and normalized by the maximum measured aggregated PV power within the same time window. The vertical dashed red line marks the occurrence time of the ramp event.
Solar power forecasting predicts photovoltaic generation and irradiance from weather, imagery, plant data, and recent output to support reliable grid and energy decisions.
Due to the rise in the use of renewable energies as an alternative to traditional ones, and especially solar energy, there is increasing interest in studying how to address photovoltaic forecasting in the face of the challenge of variability in photovoltaic energy production, using different methodologies. This work develops a hybrid approach for short and long-term forecasting based on two studies with the same purpose. A multimodal approach that combines images of the sky and photovoltaic energy history with meteorological data is proposed. The main goal is to improve the accuracy of ramp event prediction, increase the robustness of forecasts in cloudy conditions, and extend capabilities beyond nowcasting, to support more efficient operation of the power grid and better management of solar variability. Deep neural models are used for both nowcasting and forecasting solutions, incorporating individual and multiple meteorological variables, as well as an analytical solar position. The results demonstrate that the inclusion of meteorological data, particularly the surface long-wave, radiation downwards, and the combination of wind and solar position, significantly improves current predictions in both nowcasting and forecasting tasks, especially on cloudy days. This study highlights the importance of integrating diverse data sources to improve the reliability and interpretability of solar energy prediction models.
Ines Montoya-Espinagosa, Antonio Agudo
Universitat Pompeu Fabra, Barcelona, Spain. · Institut de Robòtica i Informàtica Industrial, CSIC-UPC, Barcelona, Spain.
Accurate solar irradiance forecasting is essential for the reliable integration of photovoltaic power into modern electricity grids. All-sky imagers (ASI) provide high-resolution observations of clouds, making them well suited for intra-hour forecasting. Recent deep learning approaches have substantially improved forecast accuracy but are often limited by deterministic predictions and a reduced capability to anticipate ramp events. This work proposes FarSky, a generative forecasting framework that leverages latent-space coupling to learn task-aware representations of sky images. A multi-task autoencoder first learns a shared latent representation for image reconstruction and irradiance estimation. A latent diffusion model then generates future latent states conditioned on recent observations, from which irradiance forecasts are directly decoded. Probabilistic forecasts are inherently obtained through stochastic sampling. The framework is developed using a multi-year ASI dataset acquired at the Plataforma Solar de Almería, Spain, and evaluated on two independent test datasets against persistence, state-of-the-art end-to-end, and generative forecasting approaches. FarSky achieves the best overall deterministic and probabilistic forecasting performance, improving forecast skill by up to 11 percentage points. Furthermore, it substantially improves ramp event detection over existing methods, achieving F1-scores above 60%. These results demonstrate the potential of combining generative models with task-aware latent-space coupling for solar forecasting.
Yann Fabel, Bijan Nouri, Milon Miah +4
1DLR, Institute of Solar Research · 2RWTH Aachen University · 3CIEMAT, Energy Department, Renewable Energy Division
Ultra-short-term forecasting of densely distributed photovoltaic (PV) systems remains challenging because external meteorological and satellite products may not adequately resolve rapid local variations, while heterogeneous PV systems exhibit different generation responses under similar atmospheric conditions. Meanwhile, dense PV fleets provide spatially distributed measurements that reveal regional generation dynamics. Here, we reformulate fleet-level PV forecasting from modeling discrete station time series to predicting explicit, spatially structured regional PV generation fields. We develop SatPV-Field, a regional-to-local forecasting framework that transforms distributed PV measurements into image-like generation fields and models their spatiotemporal evolution. By combining historical PV fields with satellite-derived future shortwave radiation guidance and meteorological forecasts, the framework predicts future regional generation states, which subsequently provide location-specific guidance for individual-station forecasting. Experiments involving 496 distributed PV systems, including 446 source stations and 50 held-out target stations, demonstrate that these regional fields can be effectively forecast and used to improve station-level predictions. The proposed approach reduces station-level MAE by approximately 10.0% relative to the best-performing single-station temporal baseline and by 2.5% relative to a graph-based regional forecasting approach using the same source fleet. These findings establish an explicit spatial-field modeling paradigm that connects regional generation dynamics with individual-site forecasting.
Dayin Chen, Xiaodan Shi, Xin Xu +5
Zhejiang Key Laboratory of Industrial Intelligence and Digital Twin, Eastern Institute of Technology, Ningbo 315200, China. · Ningbo Key Laboratory of Advanced Manufacturing Simulation, Eastern Institute of Technology, Ningbo 315200, China · Department of Building Environment and Energy Engineering, and the International Centre of Urban Energy Nexus, The Hong Kong Polytechnic University, Hong Kong SAR, China. +1