This dataset contains turbine- and plant-level power outputs for 252,500 cases of diverse wind plant layouts operating under a wide range of yawing and atmospheric conditions. The power outputs were computed using the Gaussian wake model in NREL's FLOw Redirection and Induction in Steady State (FLORIS) model, version 2.3.0. The 252,500 cases include 500 unique wind plants generated randomly by a specialized Plant Layout Generator (PLayGen) that samples randomized realizations of wind plant layouts from one of four canonical configurations: (i) cluster, (ii) single string, (iii) multiple string, (iv) parallel string. Other wind plant layout parameters were also randomly sampled, including the number of turbines (25-200) and the mean turbine spacing (3D-10D, where D denotes the turbine rotor diameter). For each layout, 500 different sets of atmospheric conditions were randomly sampled. These include wind speed in 0-25 m/s, wind direction in 0 deg.-360 deg., and turbulence intensity chosen from low (6%), medium (8%), and high (10%). For each atmospheric inflow scenario, the individual turbine yaw angles were randomly sampled from a one-sided truncated Gaussian on the interval 0 deg.-30 deg. oriented relative to wind inflow direction. This random data is supplemented with a collection of yaw-optimized samples where FLORIS was used to determine turbine yaw angles that maximize power production for the entire plant. To generate this data, a subset of cases were selected (50 atmospheric conditions from 50 layouts each for a total of additional 2,500 cases) for which FLORIS was re-run with wake steering control optimization. The IEA onshore reference turbine, which has a 130 m rotor diameter, a 110 m hub height, and a rated power capacity of 3.4 MW was used as the turbine for all simulations. The simulations were performed using NREL's Eagle high performance computing system in February 2021 as part of the Spatial Analysis for Wind Technology Development project funded by the U.S. Department of Energy Wind Energy Technologies Office. The data was collected, reformatted, and preprocessed for this OEDI submission in May 2023 under the Foundational AI for Wind Energy project funded by the U.S. Department of Energy Wind Energy Technologies Office. This dataset is intended to serve as a benchmark against which new artificial intelligence (AI) or machine learning (ML) tools may be tested. Baseline AI/ML methods for analyzing this dataset have been implemented, and a link to their repository containing those models has been provided. The .h5 data file structure can be found in the GitHub repository under explore_wind_plant_data_h5.ipynb.

**Overview** Wind and lidar turbulence profiles from 40 m to 220 m above the surface. **Data Quality** These two-minute-averaged data files consider the 1 Hz line-of-sight measurements that pass the -22 dB CNR quality control threshold. **Uncertainty** Line-of-sight measurements are converted to horizontal wind speed by assuming horizontal homogeneity in the measurement volume (as discussed in several publications, including Rhodes and Lundquist 2013 and Lundquist et al. 2015). In inhomogeneous flow, this assumption may not be valid. These instruments were sited to avoid breaking the horizontal inhomogeneity assumption, although the Gordon's Ridge sight may be problematic in this respect.

No description found

**Overview** **Winds.** A radar wind profiler measures the Doppler shift of electromagnetic energy scattered back from atmospheric turbulence and hydrometeors along 3-5 vertical and off-vertical point beam directions. Back-scattered signal strength and radial-component velocities are remotely sensed along all beam directions and are combined to derive the horizontal wind field over the radar. These data typically are sampled and averaged hourly and usually have 6-m and/or 100-m vertical resolutions up to 4 km for the 915 MHz and 8 km for the 449 MHz systems. **Temperature.** To measure atmospheric temperature, a radio acoustic sounding system (RASS) is used in conjunction with the wind profile. These data typically are sampled and averaged for five minutes each hour and have a 60-m vertical resolution up to 1.5 km for the 915 MHz and 60 m up to 3.5 km for the 449 MHz. **Moments and Spectra.** The raw spectra and moments data are available for all dwells along each beam and are stored in daily files. For each day, there are files labeled "header" and "data." These files are generated by the radar data acquisition system (LAP-XM) and are encoded in a proprietary binary format. Values of spectral density at each Doppler velocity (FFT point), as well as the radial velocity, signal-to-noise ratio, and spectra width for the selected signal peak are included in these files. Attached zip files, *449mhz-spectra-data-extraction.zip* and *449mhz-moment-data-extraction.zip*, include executables to unpack the spectra, (GetSpectra32.exe) and moments (GetMomSp32.exe), respectively. Documentation on usage and output file formats also are included in the zip files. **Data Details** Note, the b0 data is identical to 00 data but a netcdf extraction of the b0 data was also created for the duration of the WFIP2 campaign. **Data Quality** Various quality control (QC) algorithms developed over the years process data in real time on the radar software layer. These algorithms, which run in real time, act on time-series, spectra, moment, and consensus data layers that are persisted in different forms.

**Overview** **Winds** A radar wind profiler measures the Doppler shift of electromagnetic energy scattered back from atmospheric turbulence and hydrometeors along 3-5 vertical and off-vertical point beam directions. Back-scattered signal strength and radial-component velocities are remotely sensed along all beam directions and combined to derive the horizontal wind field over the radar. These data typically are sampled and averaged hourly and usually have 6-m and/or 100-m vertical resolutions up to 4 km for the 915 MHz and 8 km for the 449 MHz systems. **Temperature** To measure atmospheric temperature, a radio acoustic sound system (RASS) is used in conjunction with the wind profile. These data typically are sampled and averaged for five minutes each hour and have a 60-m vertical resolution up to 1.5 km for the 915 MHz and 60-m up to 3.5k m for the 449 MHz. **Data Details** Spectra data are stored in two daily files, a header (file names contain "H") and a data (file names contain "D") file. The (H)eader files are made up of binary data records containing information about the operational parameters of the profiler, while (D)ata files, also composed of binary data records, contain the spectra data collected by the profiler, i.e. spectral values for each spectral bin for every range gate. **Data Quality** Various quality control (QC) algorithms developed over the years process data in real time on the radar software layer. These algorithms, which run in real time, act on time-series, spectra, moment, and consensus data layers that are persisted in various forms. For a detailed description, refer to the attached QC document: *915 and 449 MHz Radar Wind Profilers and RASS QC*. **Uncertainty** The uncertainty is defined by the spacing of the spectral bin.

**Overview** Wind and lidar turbulence profiles from 40 m to 220 m above the surface. **Data Quality** These two-minute-averaged data files consider the 1 Hz line-of-sight measurements that pass the -22 dB CNR quality control threshold. **Uncertainty** Line-of-sight measurements are converted to horizontal wind speed by assuming horizontal homogeneity in the measurement volume (as discussed in several publications, including Rhodes and Lundquist 2013 and Lundquist et al. 2015). In inhomogeneous flow, this assumption may not be valid. These instruments were sited to avoid breaking the horizontal inhomogeneity assumption, although the Gordon's Ridge sight may be problematic in this respect.

**Overview** The WindCube v2 is a pulsed LIDAR and uses four beams sent in succession in the four cardinal directions along a 28°scanning cone angle to measure horizontal velocity and wind direction. A fifth beam is sent in the vertical direction to measure vertical velocity. Measurement heights are user-programmed and range from 40 m to 200 m. During HilFlowS, the WindCube v2 was programmed to measure from 40 m to 150 m at 10-m intervals. The data sampling rate was 1 s, wind speed accuracy was 0.1 m/s, and direction accuracy was 1.5° for the instrument. The data were averaged across 10-min. averaging intervals. **Data Details** Data Begins: 2019-07-03\ Data Ends: 2019-09-23 Additional Information, files begin with:\ Timestamp (end of interval)\ Wiper count Available at each level (40 m - 150 m):\ Wind Speed (m/s)\ Wind Speed Dispersion or Standard Devation (m/s) \ Wind Speed min (m/s)\ Wind Speed max (m/s)\ Wind Direction (deg)\ Z-wind or Vertical Wind Speed (m/s)\ Z-wind Dispersion or Standard Deviation (m/s)\ carrier-to-noise ratio (CNR), specified in decibels (dB)\ CNR min (dB)\ Dopp Spect Broad (m/s)\ Data Availability (%)

**Overview** **Winds** A radar wind profiler measures the Doppler shift of electromagnetic energy scattered back from atmospheric turbulence and hydrometeors along 3-5 vertical and off-vertical point beam directions. Back-scattered signal strength and radial-component velocities are remotely sensed along all beam directions and combined to derive the horizontal wind field over the radar. These data typically are sampled and averaged hourly and usually have 6-m and/or 100-m vertical resolutions up to 4 km for the 915 MHz and 8 km for the 449 MHz systems. **Temperature** To measure atmospheric temperature, a radio acoustic sound system (RASS) is used in conjunction with the wind profile. These data typically are sampled and averaged for five minutes each hour and have a 60-m vertical resolution up to 1.5 km for the 915 MHz and 60-m up to 3.5k m for the 449 MHz. **Data Details** Spectra data are stored in two daily files, a header (file names contain "H") and a data (file names contain "D") file. The (H)eader files are made up of binary data records containing information about the operational parameters of the profiler, while (D)ata files, also composed of binary data records, contain the spectra data collected by the profiler, i.e. spectral values for each spectral bin for every range gate. **Data Quality** Various quality control (QC) algorithms developed over the years process data in real time on the radar software layer. These algorithms, which run in real time, act on time-series, spectra, moment, and consensus data layers that are persisted in various forms. For a detailed description, refer to the attached QC document: *915 and 449 MHz Radar Wind Profilers and RASS QC*. **Uncertainty** The uncertainty is defined by the spacing of the spectral bin.

**Overview** **Winds** A radar wind profiler measures the Doppler shift of electromagnetic energy scattered back from atmospheric turbulence and hydrometeors along 3-5 vertical and off-vertical point beam directions. Back-scattered signal strength and radial-component velocities are remotely sensed along all beam directions and combined to derive the horizontal wind field over the radar. These data typically are sampled and averaged hourly and usually have 6-m and/or 100-m vertical resolutions up to 4 km for the 915 MHz and 8 km for the 449 MHz systems. **Temperature** To measure atmospheric temperature, a radio acoustic sound system (RASS) is used in conjunction with the wind profile. These data typically are sampled and averaged for five minutes each hour and have a 60-m vertical resolution up to 1.5 km for the 915 MHz and 60-m up to 3.5k m for the 449 MHz. **Data Quality** Various quality control (QC) algorithms developed over the years process data in real time on the radar software layer. These algorithms, which run in real time, act on time-series, spectra, moment, and consensus data layers that are persisted in various forms. For a detailed description, refer to the attached QC document: *915 and 449 MHz Radar Wind Profilers and RASS QC*.

**Overview** This dataset contains measurements from eight different Vaisala Triton Wind Profiler instruments. The Triton Wind Profiler is a sodar wind profiler that measures wind speed, direction, and turbulence intensity at heights from 30 m to 200 m above ground every 10 minutes. The eight Tritons are located at various sites around the WFIP2 study area. **Data Details** Regarding the minimum requirements for the site description, a Keyhole Markup Language (KML) file is attached with all of the AON Triton locations. Unfortunately, there are no photos of the sites. The layout of each site is simple. At all locations, the Triton Wind Profiler is placed on the ground with the solar panel facing due south. Each unit is solar powered and communicates its data via satellite, so there are no cables of any kind. Also, the specified start and end dates are for the entire AON network. Some individual units start later or end earlier. All start/end dates for the individual units are given as follows: AON1 (z17): 10/1/2015 -- 7/31/2017 AON2 (z14): 10/1/2015 -- 7/31/2017 AON3 (z18): 10/1/2015 -- 7/31/2017 AON4 (z12): 12/5/2015 -- 7/31/2017 AON5 (z06): 10/1/2015 -- 7/31/2017 AON6 (z05): 10/1/2015 -- 7/31/2017 (w/gap 2016-08-01 -- 2016-09-28) AON7 (z02): 10/1/2015 -- 7/31/2017 (w/gap 2016-07-01 -- 2016-11-18) AON8 (z01): 12/7/2015 -- 4/9/2016 AON9 (z20): 11/19/2016 -- 7/31/2017 **Data Quality** The Triton firmware has a quality assessment algorithm that assigns a quality factor (“quality”) to each time/height measurement of wind, expressed as a percent value in the range 0-100. In addition, the upward Doppler velocity (“vert”) is measured and can be used as an indicator of falling precipitation, which negatively affects data quality. Two automated procedures are applied in real time as these data (level 00) are sent to the DAP. Data are set to a missing value (null in the CSV file) when either “quality” < 90% or “vert” < -1.5 m/s. Because the data level is from a real-time feed, no manual quality control (QC) has been performed. Periods of missing data are likely due to real-time glitches, many of which will be filled in once the final data retrieval and QC process are performed (level b0). Finally, for the first few months of the study, the variables provided in the real-time (level "00") files differ slightly than what is specified on this metadata web page. The header line in the earlier CSV files is fairly self-explanatory in defining the variables and units. **Uncertainty** When compared to nearby towers instrumented with cup anemometers and wind vanes, the root mean square (RMS) difference in 10-minute wind speed between the Triton and met tower typically is around 0.5 m s-1. When tested at 30 different sites in a recent validation study, the RMS difference in long-term mean wind speed between the Triton and met tower is 1.3%. **Constraints** Various meteorological and environmental conditions can lead to either weaker returns or enhanced noise, resulting in a poor measurement. The higher the target point, the more difficult it is to retrieve a strong signal. Hence, a common situation is that good data will be obtained up to some height yet not above it. The percentage of time that good data are recovered at a particular height is the data recovery rate. In a recent validation study, data recovery rates were around 98% at lower heights, slowly dropping off to 96% at 100 m, 83% at 160 m, and 70% at 200 m.