arXiv subjects

Luca Lanzilao

Publications and source records attributed to Luca Lanzilao.

9 recordsLinked to original sources

arXiv2026

Skillful forecasting of offshore winds from satellite scatterometer constellations

Accurate intraday forecasts of offshore wind are becoming increasingly important for power system operation and the integration of growing shares of offshore wind energy. Operational forecasts rely predominantly on numerical weather prediction (NWP), which is not optimized for lead times of minutes to hours, where initial-condition accuracy dominates forecast skill. Although satellite scatterometer observations are routinely assimilated into NWP, they have not previously been used directly for forecasting. Here we present WindCastNet, the first satellite-based nowcasting framework for offshore wind speed and direction, introducing a new paradigm for intraday forecasting that learns from spatiotemporally irregular satellite observations. WindCastNet predicts offshore wind fields from observations acquired by satellite scatterometer constellations. WindCastNet employs a partial convolutional long short-term memory network that exploits microwave radar observations from the European, Chinese, and Indian scatterometers despite their irregular spatial coverage, asynchronous sampling, and variable revisit times. Spatial observation masks and inter-observation intervals are encoded, while a continuous temporal representation enables forecasts at arbitrary lead times. Evaluated over the North Sea, WindCastNet reduces the root-mean-square error by 23% and 7% relative to the HARMONIE MEPS model at lead times of 1 and 2 h, respectively, and outperforms persistence by 9-15% during the first three forecast hours. Forecast skill decreases under strong-wind conditions and spatially non-uniform flow. These results demonstrate that satellite scatterometer constellations can provide an independent and competitive source of short-term offshore wind forecasts, opening new opportunities for renewable energy forecasting but also broader marine weather applications, including tropical cyclone nowcasting.

cs.LG↗

arXiv2026

Characterization and forecasting of national-scale solar power ramp events

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.

cs.LG↗

arXiv2026

Intraday spatiotemporal PV power prediction at national scale using satellite-based solar forecast models

We present a novel framework for spatiotemporal photovoltaic (PV) power forecasting and use it to evaluate the reliability, sharpness, and overall performance of seven intraday PV power nowcasting models. The model suite includes satellite-based deep learning and optical-flow approaches and physics-based numerical weather prediction models, covering both deterministic and probabilistic formulations. Forecasts are first validated against satellite-derived surface solar irradiance (SSI). Irradiance fields are then converted into PV power using station-specific machine learning models, enabling comparison with production data from 6434 PV stations across Switzerland. To our knowledge, this is the first study to investigate spatiotemporal PV forecasting at a national scale. We additionally provide the first visualizations of how mesoscale cloud systems shape national PV production on hourly and sub-hourly timescales. Our results show that satellite-based approaches outperform the Integrated Forecast System (IFS-ENS), particularly at short lead times. Among them, SolarSTEPS and SHADECast deliver the most accurate SSI and PV power predictions, with SHADECast providing the most reliable ensemble spread. The deterministic model IrradianceNet achieves the lowest root mean square error, while probabilistic forecasts of SolarSTEPS and SHADECast provide better-calibrated uncertainty. Forecast skill generally decreases with elevation. At a national scale, satellite-based models forecast the daily total PV generation with relative errors below 10% for 82% of the days in 2019-2020, demonstrating robustness and their potential for operational use.

cs.LG↗

arXiv2024

Wind-farm wake recovery mechanisms in conventionally neutral boundary layers

Synthetic-aperture radar images and mesoscale model results show that wind-farm wakes behave very differently than single-turbine wakes, e.g. with wakes that seemingly narrow and do not disperse over long distances. In the current work, we aim at better understanding the physical mechanisms that govern wind-farm wake behaviour and recovery. Hence, we study the wake properties of a 1.6 GW wind-farm operating in conventionally neutral boundary layers with four capping-inversion heights, i.e. 203, 319, 507 and 1001 m. In shallow boundary layers, we find strong flow decelerations which reduce the Coriolis force magnitude, leading to an anticlockwise wake deflection in the Northern Hemisphere. In deep boundary layers, the vertical turbulent entrainment of momentum adds clockwise-turning flow from aloft into the wake region, leading to a faster recovery rate and a clockwise wake deflection. To estimate the wake properties, we develop a simple model that fits the velocity magnitude profiles along the spanwise direction. Based on this, we observe that the wake narrows along the downstream direction in all cases. Further, a detailed momentum budget analysis shows that the wake is mostly replenished by turbulent vertical entrainment in deep boundary layers. In shallow boundary layers, the capping inversion limits vertical motions and wakes are mostly replenished by mean flow entrainment in the spanwise direction. Moreover, in these cases, we observe a counterclockwise flow rotation near the left edge of the wake, which persists at each location downstream of the farm, giving rise to local strong streamwise velocity gradients along the spanwise direction.

physics.flu-dyn↗

arXiv2023

A meso-micro atmospheric perturbation model for wind farm blockage

As wind farms continue to grow in size, mesoscale effects such as blockage and gravity waves become increasingly important. Allaerts & Meyers (J. Fluid Mech., 2019) proposed an atmospheric perturbation model (APM) that can simulate the interaction of wind farms and the atmospheric boundary layer while keeping computational costs low. The model resolves the meso-scale flow, and couples to a wake model to estimate the turbine inflow velocities at the micro-scale. This study presents a new way of coupling the mesoscale APM to a wake model, based on matching the velocity between the models throughout the farm. This method performs well, but requires good estimates of the turbine-level velocity fields by the wake model. Additionally, we investigate the mesoscale effects of a large wind farm, and find that aside from the turbine forces and increased turbulence levels, the dispersive stresses due to subgrid flow heterogeneity also play an important role at the entrance of the farm, and contribute to the global blockage effect. By using the wake model coupling, we can explicitly incorporate these stresses in the model. The resulting APM is validated using 27 LES simulations of a large wind farm under different atmospheric conditions. The APM and LES results are compared on both meso- and turbine-scale, and on turbine power output. The APM captures the overall effects gravity waves have on wind farm power production, and significantly outperforms standard wake models.

physics.flu-dyn↗

arXiv2023

A parametric large-eddy simulation study of wind-farm blockage and gravity waves in conventionally neutral boundary layers

We present a suite of large-eddy simulations of a wind farm operating in conventionally neutral atmospheric boundary layers (CNBLs). A fixed 1.6 GW wind farm is considered for 40 different atmospheric stratification conditions to investigate effects on wind-farm efficiency and blockage, as well as related gravity-wave excitation. A tuned Rayleigh damping layer and a wave-free fringe region method (Lanzilao & Meyers, Bound. Layer Meteor. 186, 2023) are used to avoid spurious excitation of gravity waves, and a domain-size study is included to evaluate and minimize effects of artificial domain blockage. A fully neutral reference case is also considered, to distinguish between a case with hydrodynamic blockage only, and cases that include hydrostatic blockage excited by gravity waves. We discuss in detail the dependence of gravity-wave excitation, flow fields, and wind-farm blockage on capping-inversion height, strength and free-atmosphere lapse rate. In all cases, an unfavourable pressure gradient is present in front of the farm, and a favourable pressure gradient in the farm, with hydrostatic contributions arising from gravity waves at least an order of magnitude larger than hydrodynamic effects. Using respectively non-local and wake efficiencies $η_\textit{nl}$ and $η_\textit{w}$ (Allaerts & Meyers, Bound. Layer Meteor. 166, 2018), we observe a strong negative correlation between unfavourable upstream pressure rise and $η_\textit{nl}$, and a strong positive correlation between the favourable pressure drop in the farm and $η_\textit{w}$. Using a simplified linear gravity-wave model, we formulate a simple scaling for $η_\textit{nl}/η_\textit{w}$, which matches reasonably well with the LES results.

physics.flu-dyn↗

arXiv2023

Bayesian uncertainty quantification framework for wake model calibration and validation with historical wind farm power data

The expected growth in wind energy capacity requires efficient and accurate models for wind farm layout optimization, control, and annual energy predictions. Although analytical wake models are widely used for these applications, several model components must be better understood to improve their accuracy. To this end, we propose a Bayesian uncertainty quantification framework for physics-guided data-driven model enhancement. The framework incorporates turbulence-related aleatoric uncertainty in historical wind farm data, epistemic uncertainty in the empirical parameters, and systematic uncertainty due to unmodelled physics. We apply the framework to the wake expansion parameterization in the Gaussian wake model and employ historical power data of the Westermost Rough offshore wind farm. We find that the framework successfully distinguishes the three sources of uncertainty in the joint posterior distribution of the parameters. On the one hand, the framework allows for wake model calibration by selecting the maximum a posteriori estimators for the empirical parameters. On the other hand, it facilitates model validation by separating the measurement error and the model error distribution. In addition, the model adequacy and the effect of unmodelled physics are assessable via the posterior parameter uncertainty and correlations. Consequently, we believe that the Bayesian uncertainty quantification framework can be used to calibrate and validate existing and upcoming physics-guided models.

physics.flu-dyn↗

arXiv2022

An improved fringe-region technique for the representation of gravity waves in large-eddy simulation with application to wind farms

Large-eddy simulations of the atmospheric boundary layer are often performed using pseudo-spectral methods, which adopt a fringe-region approach to introduce inflow boundary conditions. However, we notice that a standard fringe-region technique excites spurious gravity waves when stratified atmospheres are considered, therefore enhancing the amount of energy reflected from the top of the domain and perturbing the velocity and pressure fields downstream. In this work, we develop a new fringe-region method that imposes the inflow conditions while limiting spurious effects on the surrounding flow. This is achieved by locally damping the convective term in the vertical momentum equation. We first apply the standard and wave-free fringe-region techniques to two-dimensional inviscid-flow simulations subjected to 169 different atmospheric states. A similar study is performed on a three-dimensional domain using a couple of atmospheric states. In all cases, the new fringe-region technique outperforms the standard method, imposing the inflow conditions with a minimal impact on the surrounding flow. Moreover, we also investigate the performance of two already existing non-reflective upper boundary conditions, that is a Rayleigh damping layer (RDL) and a radiation condition (RC). Results highlight the importance of carefully tuning the RDL to limit the distortion of the numerical solution. Also, we find that the tuned RDL outperforms the RC in all cases. Finally, the tuned RDL together with the wave-free fringe-region method are applied to an LES of a wind farm operating in a conventionally neutral boundary layer, for which we measure a reflectivity of only 0.75%.

physics.flu-dyn↗

arXiv2020

A new wake-merging method for wind-farm power prediction in presence of heterogeneous background velocity fields

Many wind farms are placed near coastal regions or in proximity of orographic obstacles. The meso-scale gradients that develop in these zones make wind farms operating in velocity fields that are rarely uniform. However, all existing wake-merging methods in engineering wind-farm wake models assume a homogeneous background velocity field in and around the farm, relying on a single wind-speed value usually measured several hundreds of meters upstream of the first row of turbines. In this study, we derive a new momentum-conserving wake-merging method capable of superimposing the waked flow on a heterogeneous background velocity field. We couple the proposed wake-merging method with four different wake models, i.e. the Gaussian, super-Gaussian, double-Gaussian and Ishihara model, and we test its performance against LES data, dual-Doppler radar measurements and SCADA data from the Horns Rev, London Array, and Westermost Rough farm. Next to this, as an additional point of reference, the standard Jensen model with quadratic superposition is also included. Results show that the new method performs similarly to linear superposition of velocity deficits in homogeneous conditions but it shows better performance when a spatially varying background velocity is used. The most accurate estimates are obtained when the wake-merging method is coupled with the double-Gaussian and Gaussian single-wake model. The Ishihara model also shows good agreements with observations. In contrast to this, the Jensen and super-Gaussian wake model underestimate the farm power output for all wind speeds, wind directions and wind farms considered in our analysis.

physics.flu-dyn↗