Wind Characteristics Research Articles

Downscaling of surface wind forecasts using convolutional neural networks

Abstract. Near-surface winds over complex terrain generally feature a large variability at the local scale. Forecasting these winds requires high-resolution numerical weather prediction (NWP) models, which drastically increase the duration of simulations and hinder them in running on a routine basis. Nevertheless, downscaling methods can help in forecasting such wind flows at limited numerical cost. In this study, we present a statistical downscaling of WRF (Weather Research and Forecasting) wind forecasts over southeastern France (including the southwestern part of the Alps) from its original 9 km resolution onto a 1 km resolution grid (1 km NWP model outputs are used to fit our statistical models). Downscaling is performed using convolutional neural networks (CNNs), which are the most powerful machine learning tool for processing images or any kind of gridded data, as demonstrated by recent studies dealing with wind forecast downscaling. The previous studies mostly focused on testing new model architectures. In this study, we aimed to extend these works by exploring different output variables and their associated loss function. We found that there is no one approach that outperforms the others in terms of both the direction and the speed at the same time. Finally, the best overall performance is obtained by combining two CNNs, one dedicated to the direction forecast based on the calculation of the normalized wind components using a customized mean squared error (MSE) loss function and the other dedicated to the speed forecast based on the calculation of the wind components and using another customized MSE loss function. Local-scale, topography-related wind features, which were poorly forecast at 9 km, are now well reproduced, both for speed (e.g., acceleration on the ridge, leeward deceleration, sheltering in valleys) and direction (deflection, valley channeling). There is a general improvement in the forecast, especially during the nighttime stable stratification period, which is the most difficult period to forecast. The result is that, after downscaling, the wind speed bias is reduced from −0.55 to −0.01 m s−1, the wind speed MAE is reduced from 1.02 to 0.69 m s−1 (32 % reduction) and the wind direction MAE is reduced from 25.9 to 15.5∘ (40 % reduction) in comparison with the 9 km resolution forecast.

Experimental study of the impact of upstream mountain terrain and urban exposure on approaching wind characteristics

The approaching wind profiles have a significant impact on the wind load of the structures as well as the urban climate, pollutant diffusion, and thermal comfort. Therefore, the study of its evolution is gaining attention as the city sparrows. This paper aims to study the effects of topography and morphology on the characteristics of the approaching wind. Wind tunnel tests were conducted to observe the mean velocity profiles downstream of three kinds of common terrain, which are roughness change from open to urban exposure, open exposure behind mountain terrain, and urban exposure behind mountain terrain. The results show that the ESDU transitional model overestimates velocity inside the IBL and underestimates IBL heights compared with experimental results. The rough exposure behind the mountain leads to smaller recirculation zones and a shorter reattachment location downstream compared with that of open exposure behind mountain. In addition, the evolution of mean velocity profiles over urban exposure behind the mountain is also compared with the results estimated according to current wind load provisions. Unfortunately, none of the current wind load provisions yields a reasonable estimation of the mean velocity profiles. The study aims to provide useful references in the study of the wind environment in mountainous urban area and the estimation of wind load on structures, as well as the future codification of wind load provisions.

Wavelet characterizing the non-stationary features of mountain valley winds both in frequency and time domain

This study aimed to quantify and evaluate the non-stationary characteristics of wind speed in mountain canyon terrain. A non-stationarity index was proposed based on continuous wavelet transform to quantitatively evaluate the non-stationarity of measured wind speed samples in both time and frequency domains. Subsequently, the effect of sample length and observation time interval of wind speed samples on the non-stationarity index was investigated. In addition, the non-stationarity of mountain canyon winds was compared with that of normal climate winds under the same sample length and observation time interval. The results indicate that the non-stationarity index based on wavelet transform can quantitatively evaluate the non-stationary characteristics of wind speed. With an increase in the sample length of wind speed series, the non-stationarity of wind speed increase gradually, and its probability distribution becomes more positively skewed. When the sample length of wind speed does not exceed 10 min, the probability distribution of non-stationarity index for mountain valley terrain and normal climate winds is not significantly different, which implies that it is reasonable for most countries to use a 10-min statistical length to calculate average wind speed. However, the difference in the probability distribution of non-stationarity index between mountain and normal climate winds significantly increases when the sample length exceeds 10 min. Besides, the non-stationarity index can reflect the non-stationary characteristics of wind speed series at different frequency components. Higher frequency components in wind speed exhibit stronger non-stationary characteristics. Moreover, mountain winds exhibit stronger non-stationary characteristics than normal climate winds in amplitude and frequency, especially the high-frequency components in wind speed series with long sample lengths.

Operational Characteristics Assessment of a Wind–Solar–Hydro Hybrid Power System with Regulating Hydropower

Renewable energy generation technology, as an alternative to traditional coal-fired power generation, is receiving increasing attention. However, the intermittent characteristics of wind and solar energy pose certain challenges to the stable operation of power grids. This requires a better understanding of the operational characteristics of renewable energy to improve the comprehensive efficiency. To achieve this, firstly, four indicators (i.e., average fluctuation magnitude, Richards–Baker flashiness, average climbing rate, and change in the time-averaged value) within a single-evaluation-indicator framework are proposed to quantitatively evaluate the fluctuation characteristics of wind, solar, and hydropower and a wind–solar–hydro hybrid power system. Secondly, a comprehensive evaluation indicator is developed by scientifically assigning and recombining the four indicators using entropy weight theory. Furthermore, the comprehensive evaluation index is applied to the wind–solar–hydro hybrid power system to determine the operational characteristics of subsystems and a complementary system at different time scales. Finally, the load tracking coefficient and coupling degree are used to quantify the complementarity degree of the hybrid power system. It is found that the fluctuation degree of the hybrid power system is smaller than that of the individual power system. Meanwhile, both fluctuation and complementary characteristics are strengthened with the increase in time scale. The proposed methods and results shift the volatility from an abstract concept to concrete representation, providing a new perspective and reference for evaluating the operational characteristics of the hybrid power system to achieve power system planning and scheduling.

Day-ahead wind power forecasting based on feature extraction integrating vertical layer wind characteristics in complex terrain

Accurate wind power forecasts help establish efficient power supply plans and stabilize power systems. For long-term forecasts, the outputs of numerical weather prediction (NWP) models are pipelined as inputs for the statistical post-processing model, underscoring the necessity of understanding forecasts simulated from NWP to enhance power prediction accuracy.This study aims to enhance the quality of wind power forecasts in complex terrains, focusing on identifying and processing appropriate wind features of vertical layers simulated by NWP. In complex terrains with significant terrain variability, it is crucial to meticulously analyze and select the optimal vertical layer for each site or turbine individually, as simulated wind speeds at higher vertical layers than those used in previous research could potentially yield stronger correlations. Furthermore, we introduce a data processing technique that integrates wind characteristics across vertical layers, utilizing Principal Component Analysis (PCA). This approach not only provides physically intuitive results but also demonstrates enhanced performance compared to other feature selection methods. By selecting the appropriate vertical layers and applying the proper feature extraction, for wind farms situated in complex terrains in Korea, the annual normalized mean absolute error can be reduced by up to 1.2 %.

Wind and Solar Energy Generation Potential Features in the Extreme Northern Amazon Using Reanalysis Data

This article examines the potential for wind and solar energy generation in the state of Amapá, Brazil, using ERA5 data from between 1991 and 2020. Key metrics considered include wind power density, capacity factor, photovoltaic potential, and concentrated solar power output. Analyses revealed pronounced wind speeds offshore during summer and in continental regions during spring. Solar irradiance was notably higher in the spring. Differences in wind potential were observed between northern and southern offshore areas. Concentrated solar power efficiency and photovoltaic potential were influenced by location and cloud cover, respectively. Overall, summer presents the best offshore wind energy potential, while spring is optimal for onshore solar energy in Amapá. This study underscores the importance of understanding local climatic patterns when planning energy installations in the region.

Identification of high-wind features within extratropical cyclones using a probabilistic random forest – Part 2: Climatology over Europe

Abstract. Strong winds associated with extratropical cyclones are one of the most dangerous natural hazards in Europe. These high winds are mostly associated with five mesoscale features: the warm (conveyor belt) jet (WJ); the cold (conveyor belt) jet (CJ); cold frontal convection (CFC); strong cold-sector (CS) winds; and, in some cases, the sting jet (SJ). The timing within the cyclone's life cycle, the location relative to the cyclone core and further characteristics differ between these features and, hence, likely also their associated forecast errors. In Part 1 of this study (Eisenstein et al., 2022a), we introduced the objective and flexible identification tool RAMEFI (RAndom-forest-based MEsoscale wind Feature Identification), which distinguishes between the WJ, CFC and CS as well as CJ and SJ combined. RAMEFI is based on a probabilistic random forest trained on station observations of 12 storm cases over Europe. Being independent of spatial distribution, RAMEFI can also be applied to gridded data. Here, we use RAMEFI to compile a climatology over 19 extended winter seasons (October–March 2000–2019) based on high-resolution regional reanalyses of the German Consortium for Small-scale Modelling (COSMO) model over Europe. This allows the first ever long-term objective statistical analysis of the mesoscale wind features, including their occurrence frequency, geographical distribution and characteristics. For western and central Europe, we demonstrate that the CS is prominent in most winter storms, while CFC is the least common cause of high winds, both in terms of frequency and affected area. However, probably due to convective momentum transport, CFC is on average the cause of the highest gusts after the CJ and has the highest gust factor. As expected, CFC high-wind areas show high levels of humidity and overcast conditions. In contrast, the CS is characterised by sunnier conditions interspersed by patchy cumulus clouds, leading to a broader cloud cover distribution than for other features. The WJ produces the weakest winds on average but affects a larger area than CJ. Central Europe is more strongly affected by WJ and CFC winds, while the CJ usually occurs farther north over the North and Baltic seas, northern Germany, Denmark and southern Scandinavia. System-relative composites show that the WJ and CFC tend to occur earlier in the cyclone life cycle than the CJ and CS. Consistently, the CS is the most common cause of high winds over eastern Europe, where cyclones tend to occlude, represented by a narrowing warm sector and weakening cold front. The WJ mostly occurs within the south-eastern quadrant of a cyclone bordered by the narrow CFC in the west. However, the location of CFC varies greatly between cases. The CS occurs in the south-western quadrant, while the CJ appears closer to the cyclone centre, sometimes stretching into the south-eastern quadrant. This objective climatology largely confirms previous, more subjective investigations but puts these into climatological context. It allows a more detailed analysis of feature properties and provides a solid foundation for model assessment and forecast evaluation in future studies.

Incorporating externalities when optimising power system expansion plans

Power system expansion planning has traditionally been conducted in two phases: optimisation of the future generation system to satisfy forecast demands, followed by that of a complementary transmission system. However, it is now computationally realistic to optimise both aspects simultaneously, even for large and complex systems. At the same time, the range of candidate generation plant types to be considered has widened considerably, with the technical and economic characteristics of wind, solar and biomass plants now having to be compared with those of conventional thermal and hydro plants. It is also becoming increasingly important to take explicit account, within the planning process, of the so-called ‘externalities’ such as emission limits, budget constraints and demand-management measures, in addition to the environmental costs and secondary benefits associated with individual projects. This paper describes the results of a study to demonstrate the feasibility of explicitly considering a range of externalities when mathematically optimising integrated generation and transmission system expansion plans, using the Vietnamese power system as an example.

Transonic plasma winds with cosmic-rays and waves

Transonic plasma winds are studied under the influence of gravitational potential well in the presence of cosmic rays and self excited Alfvén waves. We present interesting features of transonic winds originating from the gravitational potential well for a three-fluid system. The model is comprising of thermal plasma, cosmic rays and Alfvén waves. The analysis is carried out on the hydro-dynamical basis with a particular emphasis on cosmic-ray streaming instability and damping mechanism. Boundary conditions on the base of the gravitational potential well will help to explore steady-state transonic wind solutions. A critical analysis on various mass outflow rate is also presented which is applicable to know the behaviour of normal and star-burst galaxies. The dependence of the critical or sonic point is studied for different sets of physical parameters. For the graphical interpretation of the numerical results different contour plots presented to analyze impacts of different mass flow rates. We find that transonic solutions exist for a wide range of parameters. We examine the case very close to the base of the gravitational potential well and the effect of the cosmic ray diffusion is neglected in current study.

Long-term statistics of atmospheric conditions over the Baltic Sea and meteorological features related to wind wave extremes in the Gulf of Gdańsk

The goal of this study is to describe wind wave climate and wave extremes of the Gulf of Gdańsk in the southern Baltic Sea and associated meteorological conditions over the Baltic Sea. We obtain the characteristic features of 34 severe historical storms in the Gulf of Gdańsk during the period 1958–2001 and link them with extreme significant wave heights hindcast for five grid points in this gulf. The long-term statistics of atmospheric pressure systems over central and northern Europe, and the north-eastern Atlantic Ocean are derived from a 44-year REMO reanalysis database. A link between the mean, minimum and variability range of atmospheric pressure has been quantified. In general, the higher the mean pressure the smaller its variability and vice versa. Long-term characteristic features of winds over the Baltic Sea have been estimated from the REMO database. Strong winds directions vary from W, WSW to SW in the southern Baltic to more southerly SSW directions in the northern part of the Baltic Sea. The Empirical Orthogonal Functions (EOF) analysis shows that more than 50% of the variability in the atmospheric pressure in the Baltic Sea can be explained by the first EOF mode. The first four EOF modes can reproduce above 90% variability of the hindcast pressure time series. Statistical properties of the hindcast significant wave height over the Gulf of Gdańsk are computed based on the 44-year HIPOCAS database. All the computed statistics of wave heights reveal a very strong sheltering effect caused by the Hel Peninsula.

The Variation of Ionospheric O+ and H+ Outflow on Storm Timescales

AbstractGeomagnetic storms are primarily driven by stream interaction regions (SIRs) and coronal mass ejections (CMEs). Since SIR and CME storms have different solar wind and magnetic field characteristics, the magnetospheric response may vary accordingly. Using FAST/TEAMS data, we investigate the variation of ionospheric O+ and H+ outflow as a function of geomagnetic storm phase during SIR and CME magnetic storms. The effects of storm size and solar EUV flux, including solar cycle and seasonal effects, on storm time ionospheric outflow, are also investigated. The results show that for both CME and SIR storms, the O+ and H+ fluences peak during the main phase, and then declines in the recovery phase. However, for CME storms, there is also significant increase during the initial phase. Because the outflow starts during the initial phase in CME storms, there is time for the O+ to reach the plasma sheet before the start of the main phase. Since plasma is convected into the ring current from the plasma sheet during the main phase, this may explain why more O+ is observed in the ring current during CME storms than during SIR storms. We also find that outflow fluence is higher for intense storms than moderate storms and is higher during solar maximum than solar minimum.

Optimal capacity allocation and economic evaluation of hybrid energy storage in a wind–photovoltaic power system

During the global energy crisis, a significant influx of renewable energy sources was connected to the power grid, resulting in adverse fluctuations. To address this challenge and simultaneously reduce environmental pollution, a hybrid energy storage system containing hydrogen energy storage (HES) and compressed air energy storage (CAES) are proposed. The system aims to reconfigure the energy storage devices by an economical means and effectively alleviate the volatility challenges by the large amount of renewable energy accessing. First, according to the behavioral characteristics of wind, photovoltaics, and the energy storage, the hybrid energy storage capacity optimization allocation model is established, and its economy is nearly 17% and 4.7% better than that of single HES and single CAES, respectively. Then, considering the difficulty of solving the complexity dimension, a carnivorous plant algorithm (CPA) is adopted to solve the model and accurately obtain the strategy of hybrid energy storage configuration in this paper. The running time of a CPA algorithm is 33.6%, 36%, and 55% shorter than particle swarm optimization, whale optimization algorithm, and firefly algorithm, respectively. Finally, the simulation analysis is performed by IEEE 33 node arithmetic. The results show that the network loss with hybrid energy storage is reduced by about 40% compared with that without hybrid energy storage. However, improving voltage stability and the economy is optimal by using configured hybrid energy storage.

МНОГОЛЕТНИЕ ИЗМЕНЕНИЯ ВЕТРОВОГО РЕЖИМА В СЕВЕРО-ЗАПАДНОЙ ЧАСТИ КАСПИЯ

In recent years, much attention has been paid to climate change issues. Regional climate change can be manifested in the changes in the main climate-forming factors, including wind. Studying the climatic characteristics of wind in the Caspian Sea is considered as the most important task, since wind affects the variability of currents, the formation of hydrological and hydrochemical processes, and the information about the wind regime is necessary for the operation of marine organizations, the specifics of which is the dependence of all stationary and mobile objects in the sea on hydrometeorological conditions. The paper provides an analysis of the frequency of wind directions and wind speeds for individual months, seasons, years, different gradations of wind speed. For the climatic characteristics of wind speed for consecutive 30-year periods, a comparative analysis of changes in climatological normals is carried out, and deviations of wind characteristics relative to the standard reference period are revealed.

Effect of built form on local wind characteristics, a case of tropical savanna climate in India

Passive techniques play a substantial role in regulating the indoor environmental quality of a building, especially in the tropics. Proper air movement is necessary for warm and humid indoors by utilising the local wind pattern. Many factors influence and alter the wind character of a region. Any built form has a significant influence on the wind and vice versa. This research extensively studies the effects of wind when it encounters different built and unbuilt spaces of an office building in India’s tropical savanna climate. Applications such as Building Information Modelling (Autodesk Revit) and Computational Fluid Dynamics simulation (Autodesk CFD) are used for analysis. The impacts on the local wind character at chosen wind path timings are studied in detail to arrive at the conclusions. Various effects like wind acceleration, corner jetting, channelling, funnelling, windbreak and cumulative effects are observed as occurring due to the influence of the building’s form on the local wind. These effects increased the wind velocity from 10kmph up to 19kmph and decreased up to 6kmph. The study concludes by arriving at several considerations while developing built forms that help to reduce negative impacts and improve positive impacts on aspects like aerodynamics, user comfort, impact on surroundings, the microclimate and structural wind loads.

Parallel ribbon vortex: A phenomenological flow feature in an atmospheric boundary layer near sloped terrain

In this study, flow behaviors and vortex morphology around the sloping terrain that is covered by an atmospheric boundary layer are investigated using a combination of numerical and experimental methods. An array of wind tunnel experiments and high-fidelity large eddies simulations have been conducted, considering different slope angles θ = 90°, 45°, 26.5°, and 18.2°. Systematic analysis of turbulence phenomena pertaining to the flow field's topography-driven distortions and accelerations was performed quantitatively via time-averaged velocity field, terrain-induced turbulence intensity, velocity spectrum, and acceleration ratio. Two recirculation zones, one originating from the upstream boundary layer's gradual separation bubble (GSB) and the other from the forced separation bubble (FSB) behind the downstream sharp leading-edge, were confirmed by detailed visualization of the flow field. An acceleration zone of considerable magnitude is also observed, connecting the GSB and FSB. Subsequently, a new vortex structure—the parallel ribbon vortex (PRV)—was disclosed. The PRVs originate from the disordered turbulence structures inside the GSB, spiraling into coherent structures. The accelerated mainstream up the step then stretches and elongates the coherence structures into ribbon-like vortices parallel to the mainstream, completing the formation process. Finally, the PRVs transport fluid up the step before encountering the strong spanwise vortices and dissipating thereafter. The investigation into the slope flow characteristics and turbulence structure conducted in this study has the potential to enhance the prediction of wind characteristics and wind load specifications in mountainous regions.

Evaluation of meteorological characteristics and the influence of the Southern Annular Mode at newly established Bharati Station, East Antarctica

The Indian Antarctic program, a multi-disciplinary and multi-institutional program initiated in 1981 with the first expedition to Antarctica, has been instrumental in advancing our understanding of Antarctic meteorology and its global climate implications. This study focuses on the latest meteorological characteristics of newly established research base Bharati aiming to unravel the climate dynamics of Bharati station in East Antarctica. The meteorological data for the period December 2014 to December 2022 have been analysed in present study. We observed the highest maximum temperature of 9.9 °C on January 5, 2018, and the lowest minimum temperature of −40.3 °C on September 6, 2016. The monthly mean temperature exhibits pronounced seasonal variation, with a summer maximum of approximately 0 °C in December and January, and a winter minimum of about −19 °C from May to September. While the eight-year observations are not sufficient for studying long-term trends, we identify a slight increasing tendency in temperature at Bharati. In addition to temperature, we examined the characteristics of pressure, relative humidity, precipitation, wind, and blizzards. The analysis reveals a negative correlation between the Southern Annular Mode (SAM) and temperature and pressure at Bharati. Station experiences predominantly easterly and northeasterly winds. The upper air data were obtained using radiosonde and ozonesonde ascents, enhancing our understanding of the atmospheric conditions in the region. These results provide important knowledge about the weather in Bharati and how it relates to the Southern Annular Mode. Understanding this helps us comprehend how East Antarctica weather influences global climate patterns.

Field measurement study of intense thunderstorm outflows characteristics based on 356 m high meteorological tower

Based on the wind speed records obtained by 356-m-high Shenzhen meteorological gradient tower (SZMGT) during 2019–2020, 10 sets of typical intense thunderstorm outflow records were obtained by screening. This paper suggests that moving average period T=60s is a suitable choice based on the correlation between time-varying mean wind and fluctuating wind. In order to discuss the wind characteristics of thunderstorm flow more precisely, it is divided into ramp-up stage, peak stage, and dissipation stage, and the average wind characteristics and fluctuating wind characteristics are analyzed in detail. Analysis revealed the change of average wind angle is usually caused by the movement of the downburst. The shape of the downburst flow profile is time-varying and has certain differences from several empirical profiles. The dissipation stage is closer to synoptic wind in terms of the wind profile than the ramp-up and peak stages. The turbulence intensity of downbursts is not significantly related to the height, but it is more significantly related to the average wind speed as well as the gust factor.

A novel wind comfort evaluation method for different airflows by considering dynamic characteristics of wind direction and velocity

The dynamic velocity characteristics of wind have been studied over the past several decades. However, there was a lack of consideration for the incoming flow directions. Besides, the comfort assessment studies mainly focused on thermal comfort from the mechanism of heat transfer instead of wind comfort from the mechanism of tactile stimuli. This study carried out a mixed indoor and outdoor experiment. The indoor experiment was conducted with a room temperature of 29 °C and a relative humidity of 50%, which closely resembled the outdoor environmental conditions. Totally 17 healthy subjects dressed at 0.36 clo and being sedentary were exposed to five kinds of airflow during the experiment. According to the results, outdoor true natural wind was more preferred than directly supplied wind. This paper also proposed a novel wind comfort (WC) model by considering various air parameters. The model relied on tactile stimuli, which were the reflections of wind pressure fluctuations and could be affected by changes in velocity and direction of airflows. Based on these two dynamic characteristics, four wind perceptions, including wind intensity vote (WIV), velocity variation sense (VVS), direction variation sense (DVS) and overall enveloping sense (OES) were introduced and linearly regressed. The findings revealed that not only velocity but also the direction could affect the wind comfort evaluation. This paper will hopefully provide a better understanding of impacts of incoming flow directions on comfort and further rethink a novel route of dynamic wind environment imitation for future indoor environments.

Profiling wind LiDAR measurements to quantify blockage for onshore wind turbines

AbstractFlow modifications induced by wind turbine rotors on the incoming atmospheric boundary layer (ABL), such as blockage and speedups, can be important factors affecting the power performance and annual energy production (AEP) of a wind farm. Further, these rotor‐induced effects on the incoming ABL can vary significantly with the characteristics of the incoming wind, such as wind shear, veer, and turbulence intensity, and turbine operative conditions. To better characterize the complex flow physics underpinning the interaction between turbine rotors and the ABL, a field campaign was performed by deploying profiling wind LiDARs both before and after the construction of an onshore wind turbine array. Considering that the magnitude of these rotor‐induced flow modifications represents a small percentage of the incoming wind speed ( ), high accuracy needs to be achieved for the analysis of the experimental data and generation of flow predictions. Further, flow distortions induced by the site topography and effects of the local climatology need to be quantified and differentiated from those induced by wind turbine rotors. To this aim, a suite of statistical and machine learning models, such as k‐means cluster analysis coupled with random forest predictions, are used to quantify and predict flow modifications for different wind and atmospheric conditions. The experimental results show that wind velocity reductions of up to 3% can be observed at an upstream distance of 1.5 rotor diameter from the leading wind turbine rotor, with more significant effects occurring for larger positive wind shear. For more complex wind conditions, such as negative shear and low‐level jet, the rotor induction becomes highly complex entailing either velocity reductions (down to 9%) below hub height and velocity increases (up to 3%) above hub height. The effects of the rotor induction on the incoming wind velocity field seem to be already roughly negligible at an upstream distance of three rotor diameters. The results from this field experiment will inform models to simulate wind‐turbine and wind‐farm operations with improved accuracy for flow predictions in the proximity of the rotor area, which will be instrumental for more accurate quantification of wind farm blockage and relative effects on AEP.

Prediction of wind fields in mountains at multiple elevations using deep learning models

To optimize the site selection of wind turbines, it is essential to understand the variability of wind speeds at different elevations on mountains and their wind characteristics. The erection of anemometer at a large altitude in mountainous regions poses significant financial and safety challenges. Therefore, it is of utmost importance to design a method that facilitates the estimation of multiple high-elevation wind speeds utilizing low-elevation measurements. This study built a graph data structure by assimilating prior knowledge, including the mountain topography and relative position of the anemometer. A custom-designed graph neural network model was designed to predict wind speed data at high-elevation regions based on wind speed data at a low-elevation region. Moreover, an enhanced UNet model was employed to predict wind field data at higher elevations utilizing predicted wind speed data. To improve the stability and performance of the model, the flux conservation equation was integrated into the loss function of the graph neural network. The results suggest that integrating mountain information and physical loss in the model can significantly improve the accuracy of wind speed data predictions. The graph neural network has remarkably enhanced the performance of the UNet model.