Assessing Wind Resource Variability Using the New York State Mesonet Data

A major issue in wind farm resource determination is the lack of measurements at wind turbine hub-heights with which wind resources and mechanical loads can be quantified. Lack of measured data also limits the incorporation of temporal variability into wind resource assessments, which is necessary for the integration of wind-generated electricity into the grid system. The New York State Mesonet (NYSM) (www.nysmesonet.org) presents a unique opportunity to address some of these issues. The NYSM comprises:

• a Profiler Network of 17 sites with Doppler lidars that retrieve vertical profiles of atmospheric variables including wind speed and direction
• a network of 126 surface weather stations that measures temperature, humidity, wind speed and direction, pressure, solar radiation, snow depth, and soil information
• a Flux Network of 17 stations that monitor the surface energy budget
• a Snow Network of 20 sites measuring snow water content

Here we focus on the 10-minute wind speed and direction data from the NYSM Profiler Network as measured with Leosphere WindCube WLS-100 series Doppler LiDAR (Light Detection and Ranging). There is good spatial coverage through the state but here we focus on the seven sites with greater than 55% data availability during the data period January 2019 to December 2022. Wind speeds and directions in 25 m increments centered on 150 m height reported at 10-minute resolution are used to describe the wind resource at each NYSM site and are integrated with those from two New York State Energy Research and Development Agency (NYSERDA) floating lidar buoys in New York Bight to evaluate the relative magnitude of wind resources and likely electrical power quality on- and off-shore. To determine the power production at each site, the wind speed time series was applied to the power curve of the International Energy Agency 15 MW reference turbine that has a rotor diameter of 240 m and a hub-height of 150 m.

Data from the NYSM lidars show the expected seasonal variation with higher wind speeds during the cold season (October-April) than during the warm season. Using a 40-year record of wind speeds from the ERA5 reanalysis, it is shown that the higher wind speed data availability in summer at the NYSM sites leads to a negative bias of about 1.5-4.5% in wind speeds at 150 m. In addition, data were selected from the time series to represent the most complete (‘best’) year (September 2019 to September 2020) from the full 2019-2022 period for further analysis. The probability of wind speeds below 3 ms^-1 (nominal wind turbine cut-in speed) is twice as high at the NYSM vs the buoy sites. Calculations of the wind energy density at 150 m height from floating NYSERDA lidar buoys show the wind resource is 40% higher offshore than those calculated from the seven NYSM sites. Nonetheless, the wind energy density is close to or exceeds 300 Wm^-2 at all sites and is greater than 800 Wm^-2 at the best NYSM site.

NYSM lidar data exhibits higher variability in potential electrical power production than at the offshore sites. For example, the temporal autocorrelation is higher at the buoy lidar sites and there is a lower probability of ramp events (changes in wind speed of >±20%) from one 10-minute time step to the next compared to the NYSM lidar sites.

Wind speed measurements at multiple heights can be used to determine wind shear, which is a key determinant of wind turbine structural loading. Typically, a value of 0.2 is assumed for the power law shear exponent from onshore wind turbine design standards. Fitting the power law to wind speeds between 100 m and 250 m height, the number of data periods with negative shear (shear exponent<0) or higher positive shear (shear exponent>0.3) can be determined. In general, the sites have negative shear between 10 and 20% of the time, while high positive shear occurs 13-24% of the time. At all 17 of the NYSM sites, 5% of shear exponent values during wind turbine operation (i.e. when wind speeds are between 3-25 ms^-1) lie above 0.39 and a further 5% of values fall below -0.09. The site with the lowest occurrence of either negative or highly positive shear is the most inland. This may in part be linked to the frequency of low-level jets (LLJ) although for most of the NYSM locations the mean LLJ core heights are above 300 m and thus above the swept area even of the IEA 15 MW reference wind turbine. The highest frequency of occurrence (14% of all 10-minute periods) of LLJ at heights from 100 to 250 m occurs during June at a site on the coast of Long Island.

New York state is likely to require expansion of the wind turbine fleet both on- and off-shore to meet renewable electricity targets and grid integration of the resulting electricity will be critically contingent on understanding co-variability of power production across these fleets. Thus, an analysis is undertaken to quantify the spatial coherence of power production as a function of separation distance between lidars. The correlation of time-series of estimated power production at the different lidar locations decays exponentially with increasing separation distance (0-600 km). For the effective sample size of the lidar data sets, the power production time series would be considered fully de-correlated at Spearman correlation coefficients < 0.2 which occurs at distances of greater than 500 km and to below 0.4 at distances of about 350 km.

As shown the NYSM data have high temporal resolution and are distributed across the state of NY. Further, the heights of measurement are well-suited to wind energy applications. As indicated, such data can be used to improve wind resource estimates, in spatial and temporal forecasting, and for studies aiding integration of renewable energy into the grid.