Classification: Physical Science: Environmental Science

Wind farms change the ground-level climate

Alona Armstrong1, 2, Ralph R. Burton3, Susan E. Lee3+, Stephen Mobbs3, Nicholas Ostle1, 4,

Victoria Smith3, Susan Waldron2 and Jeanette Whitaker4

1School of Geographical and Earth Sciences, University of Glasgow, Glasgow G12 8QQ, UK

2Lancaster Environment Centre, Lancaster University, Lancaster, LA1 4YQ, UK

3National Centre for Atmospheric Science, School of Earth and Environment, University of

Leeds, Leeds, LS2 9JT, UK

4Centre for Ecology and Hydrology, Lancaster Environment Centre, Library Avenue,

Bailrigg, LA1 4AP, UK

+Now at School of Civil Engineering, University of Birmingham, Edgbaston, Birmingham,

B15 2TT, UK

Corresponding author: a.armstrong@lancaster.ac.uk, +44 (0)1524 510243

Keywords: wind turbines, micro-climate, carbon cycling

1

Abstract

The global drive to produce low-carbon energy has resulted in an unprecedented deployment

of onshore wind turbines: there has been a 25 % increase in wind turbine production capacity

over the last decade which is predicted to grow annually by 6 % between 2011 and 2035 (1).

This represents a significant land use change for wind energy generation, from 75,200 km2 in

2012 to 301,333 km2 in 2035, with uncertain consequences for local climatic conditions and

the regulation of ecosystem carbon (C) cycling. There is a paucity of data on the effects of

wind farms on soil and atmospheric climates (2-4), limiting our ability to determine the true

C balance of this renewable energy technology (5). Here, we present high-resolution data

from a wind farm collected during operational and idle periods that shows that wind farms

affect several measures of ground-level climate. Specifically, we discovered that operational

wind turbines raised air temperature (TA) by 0.22 °C and absolute humidity (AH) by 0.03 g

m-3 during the night, and increased the variability in air, surface and soil temperature

throughout the diurnal cycle. Further, the microclimatic influence of turbines on TA and AH

decreased logarithmically with distance from the nearest turbine. These effects on local

microclimate have uncertain implications for ecosystem C cycling and understanding needs

to be improved to determine the overall C balance of wind energy.

2

Significance statement (120 word lay person stating significance)

The nascent increase in wind turbine deployment across the world, promoted by the drive to

produce low-carbon energy, represents a significant land-use change, with uncertain

consequences for the hosting ecosystems. We present the first field data showing that wind

turbines can affect several measures of the ground-level microclimate known to influence

carbon cycling. We demonstrate warming and moistening at night-time, and increases in the

variability in air, surface and soil temperatures, and absolute humidity. Given the centrality of

climate, and especially temperature, as a driver of terrestrial C cycling, these findings

demonstrate the potential for the functioning of the host landscape to be affected by wind

farm operation. We estimate that carbon sequestration may be reduced by 4% per annum.

3

Introduction

The electricity generating capacity of wind turbines is predicted to increase more than any

other renewable, amounting to over 7 % of global production by 2035 (6). Deployment of

this magnitude will result in substantial land use change, an additional 226,133 km2, given the

relatively low energy density of wind power (7) and the anticipation that 80 % of wind

capacity will be onshore (6). Effects of this land use change on human populations, non-

volant wildlife, avian and bat communities have received consideration (8-11). However, the

effect on ground-level microclimates has not been resolved, despite implications for

processes and properties, including C cycling (5), of the hosting ecosystem.

Wind farms have been postulated to affect climatic conditions from the within-farm to global

scale through modification of the vertical distribution of energy and moisture within the

atmosphere and their exchange between the land surface and atmosphere (12). Previous

studies have modelled effects (12-16), measured air temperature differences upwind and

downwind of wind turbines (2-4), and used satellite data to examine temperature effects over

a 10,000 km2 area (17). To fully resolve the implications of such temperature differentials on

hosting ecosystems, high resolution spatially explicit field data are needed when turbines are

operational and idle.

We used TA and AH data during a meteorologically ‘normal’ period (see SI) from 101

locations across an area of peatland at Black Law Wind Farm, Scotland, plus surface and soil

temperature (TSU and TSO, respectively) at 36 locations, clustered at four sites, across the

whole farm (Fig. S1) to assess whether there was wind turbine-induced changes to the

microclimate. We compared the data during periods when the wind farm was operational

(ON) and idle (OFF), and sites downwind (D) or not downwind (ND) of wind turbines (Fig.

4

S1), where appropriate. Given the trend for stable boundary layers at night (warm air above

cold air) and neutral (well-mixed) or unstable (cold air above warm air) boundary layers

during the day we examined the effects during the day and night, and diurnal patterns.

Results and discussion

Integrated effects of distance to turbine affects TA and AH. Firstly, to determine the

integrated effect of the whole wind farm across the measurement period, we analysed the

mean day and night-time differences in TA and AH departure for each site from the site-wide

mean between ON and OFF periods, using all data (i.e. data were not categorised as D or

ND) (see Fig. S1). To assess the spatial extent of effects we related the departures to distance

from the nearest turbine.

During the night, air closer to a wind turbine was warmer and moister, with TA departures

reaching 0.25 °C and AH departures 0.1 g m-3 and positive departures evident up to 200 m

away (Fig. 1a). The night-time warming and moistening was caused by downward mixing of

warmer moister air by the turbines during stable conditions (18). Analysis of midnight

soundings from the two nearest upper-air stations reveals that during the most stable

conditions the lapse rate is positive for both temperature and moisture (see SI) (13). During

the day, air closer to a wind turbine was cooler, with departures up to 0.05 °C, but AH was

not influenced (Fig. 1b). This weaker day-time effect is attributable to a convectively-driven,

well mixed boundary layer (18). Importantly, the trends observed for TA during the day and

night and AH at night can be approximated by a logarithmic function (Fig. 1), demonstrating

the effect of wind turbines can be quantified and thus represented in models of Earth surface

energy balance (e.g. numerical weather prediction models).

5

Temporal-spatial variation in TA, TSU, TSO, and AH. To quantify the temporal-spatial

effects of wind turbine operation we examined the variation in TA, TSU, TSO, and AH

throughout the diurnal cycle (irrespective of wind direction) using the coefficient of variation

of each of the measurements across the whole site for each measurement interval averaged

for both the ON and OFF periods as a function of time of day.

The spatial variations in the TA, TSU, TSO and AH data were significantly greater during the

ON period compared with the OFF period (Fig. 2), suggesting turbines increased vertical

mixing and turbulence. This has not been evident in other studies (2-4, 12, 13, 17, 18). The

differences in temperature variation between ON and OFF periods were greatest for TSU data

(Fig. 2b) and smallest for TA data (Fig. 2a), reflecting relatively well-mixed air at 2 m and

increased variability at the surface arising from peatland micro-topography and vegetation

shading.

The difference in TA and AH variability between the ON and OFF periods was greater during

night than day (Figs 2a, d), whereas it was approximately equal for TSU and TSO, with smaller

differences during transition periods around sunrise and sunset (Figs 2b, c). This

demonstrates that night-time TA and AH are most sensitive to turbines, due to downward

mixing of warm air. Further, the mixing down of warm air during the ON period appears to

have affected the diurnal trend in TSO and AH: the peaks in variability in TSO and AH were

later in the day during the OFF period compared with the ON period (Figs 2c, d). This

suggests that the night-time background gradient of TSO and AH are eroded by turbulence

earlier in the day during the ON period.

Directional effects of turbines on TA and AH. Our final analysis examined the magnitude

6

of effect of wind turbines on TA and AH during the night when the relationship between

departure and distance from the turbine was stronger (Fig. 1, see SI for day-time results). We

excluded all data from one hour before sunrise to one hour after sunset and filtered for wind

directions between 220° and 240° (aligned along the main axis of the wind farm) (Fig. S1).

Temperature departures for all the data that fit these criteria were calculated for each

sampling location, (as for Fig. 1), thus allowing the relative temperature and AH of sites

downwind and not downwind of turbines to be compared.

We found that temperatures were significantly warmer (Fig. 3a) in areas downwind of

turbines during the ON period, with an average relative warming of 0.22 °C. Although

sensors recorded lower relative humidity (RH) close to the turbines (Fig. S5), the AH of air

downwind of the turbines was, on average, 0.03 g m-3 moister (Fig. 3b). This was due to the

exponential variation of saturated water vapour pressure with temperature. Analysis of the

data shows that the relative increase in saturated water vapour pressure had more effect on

AH than the combined increase in temperature, and lowering of RH (see SI), consistent with

turbine-induced mixing of warmer and moister air downwards (12). During the OFF period,

temperature and AH departures were variable with no statistically significant difference

between D and ND sites (Figs 3c, d).

Conclusion

Together, these findings provide the first field evidence that operational wind turbines can

have a measureable effect on ground-level climate: increased TA and AH during the night and

greater variability in TA, TSU, TSO and AH throughout the diurnal cycle. Although the effects

on TA and AH were statistically significant, the observed differences were smaller than the

background variation recorded across the site (Fig. 3), attributable to differences in

7

topography, hydrology, soil properties and vegetation type. Importantly, we demonstrate that

the effects on both TA and AH can be described by a logarithmic function of distance from

nearest turbine, a generic approach showing for the first time how the integrated effect of a

wind farm may be estimated.

 

Given the centrality of climate, and especially temperature, as a driver of terrestrial C

cycling, these findings demonstrate the potential for the biogeochemical functioning of the

host landscape to be affected by wind farm operation. Night-time warming was most

influenced, suggesting that microclimate effects on night-time biological processes that

govern C cycling, i.e. decomposition and plant-soil respiration, may be important. However,

effects on respiration could also be offset by plant physiological responses to warming (19).

Our results indicate that wind farm operation had greater effects on spatial and diurnal

variability in TSU and TSO than TA. TSU and TSO are recognised as stronger regulators of plant-

soil C dynamics (20, 21), consequently, the effects on the net C balance of the hosting

ecosystem may be stronger than potentially inferred from previous studies which only

measured TA (2-4).

 

Although, the scale of the observed effects on TA and AH was small relative to typical

seasonal and diurnal variances, changes of a similar magnitude have been postulated to

impact on ecosystem C gains and losses at the global scale given the temperature sensitivity

of productivity and decomposition processes (22). Our results, calibrated against C flux data

from a peatland in northern England (23), suggest an estimated reduction in C sink capacity

of 4 % attributable to night-time warming caused by wind turbines. Together, these findings

advance our scientific understanding of the effect of operational wind turbines on ground-

level climate. Given the importance of temperature controls on plant-soil C cycling, improved

8

estimates of wind turbine-induced microclimatic effects on soil C stocks and greenhouse gas

emissions is needed.

Materials and methods

Site: Black Law Wind Farm, Scotland (55°46′01″N 03°44′20″W, altitude 250-320 m),

comprises 54 turbines over 18.6 km2. The turbine hub heights are approximately 70 m and the

total capacity is 124 MW.

Measurements: Air temperature (TA) and relative humidity (RH) were measured every 15

minutes using HOBO U23 loggers (Onset, USA) at 2 m above the land surface at 101

locations across a 2.6 by 1.4 km area of Black Law Wind Farm (Fig. S1). Absolute humidity

(AH) was derived from RH and TA. Surface and soil (-5 cm) temperatures (TSU and TSO,

respectively) were recorded every 30 minutes using Hobo Pendant data loggers (Onset, USA)

at 36 locations, clustered at four sites (Fig. S1). At the same sites soil moisture (-10 cm) was

measured every 30 minutes using Campbell CS625 water content reflectometers (using a site

specific calibration) connected to Campbell CR200 loggers (Campbell Scientific Limited,

UK). Wind direction was measured with a Gill 2D sonic anemometer (Gill Instruments, UK)

at 2 m above the land surface every 10 seconds and averaged over 10 minute intervals (Fig.

S1).

The wind farm was operational (hereafter referred to as ON) from 24th May 2012 to 7th June

2012, idle (hereafter OFF) from 12th June 2012 to 25th July 2012 and operational (ON) from

28th July to 15th November 2012. The switching on and off of the turbines was a phased

operation lasting several days; data from this period were excluded.

9

Data processing: ‘Day’ was classified as one hour after sunrise to one hour before sunset, as

defined by the National Oceanic and Atmospheric Administration algorithm (24), and ‘night’

as one hour after sunset to one hour before sunrise to avoid transition periods. Fraction of

day, between 0 and 1, was calculated, with sunrise as 0.0 (or 1.0), sunset as 0.5 and time

linearly scaled between given sunrise and sunset are the meteorologically relevant temporal

controls on boundary layer development (25) (using time-of-day is potentially misleading

since the sun rose and set at significantly different times during the period of measurement at

Black Law). The data were binned into 24 pseudo-hourly bins for analysis.

Absolute humidity (AH) was derived from relative humidity (RH) by first calculating the

water vapour saturation pressure (SWVP - temperature-dependant), and then calculating the

vapour pressure (temperature- and RH-dependant) then finally the absolute humidity based

upon temperature and vapour pressure. The SWVP was found to be the dominant control on

AH as the former varies exponentially with temperature, as explained more fully in the SI.

Departures were used in the analysis of the TA, AH and RH data to remove diurnal and

seasonal signals (the greatest controls) from the data, (Figs 1, 3, S5). Essentially, the site-

wide instantaneous mean was calculated for all measurement locations, and subtracted from

each individual measurement and then averaged over the ON and OFF periods and for D and

ND groups where appropriate. Consider first the vector of N hobo temperatures at time t:

T (t) = [T1(t), T2(t), …., TN(t)] for t = t1, t = t2, …, t = tM (i.e. M times in the sample set)

then calculating the site-wide scalar mean temperature at time t = tj gives

TAV (t j) = 1N ∑

i=1

N

T i(t j)

The vector of departures from the mean at time t jis

10

T ' (t j) = [T1(t j) - TAV(t j), T2(t j) - TAV(t j), ..., TN(t j) - TAV(t j)]

allowing the time-averaged mean departure vector to be constructed:

T ' =[ 1M ∑

j=1

M

T 1' ( t j ) ,

1M ∑

j=1

M

T 2' (t j ) , …, 1

M ∑j=1

M

T N' (t j)]

This vector is the one used in the paper, and when departures are mentioned they refer to this

vector, or elements of it.

Variation in TA, TSU, TSO, and AH was assessed by calculating the coefficient of variation

across the site for each measurement interval and averaging in pseudo hourly bins (based on

fraction of the day) for ON and OFF periods. The coefficient of variation cv was calculated

using the same steps as above for TA, AH and RH departures: the site-wide mean cv was

calculated and the time-averaged cv vector constructed (the vector of departures step was

omitted). Given the cv of a population is defined as the standard deviation divided by the

mean of the population: cv = σ / μ, the temperatures were first converted to degrees Kelvin. It

is inappropriate to use degrees Celsius to calculate cv, as negative or zero values of μ would

then be allowed, giving meaningless values of cv .

To examine the magnitude of effect of the turbines on TA, RH and AH, only data when the

wind direction was from between 220° and 240° (i.e. aligned along the main axis of the wind

farm) was considered (Fig. S1). Firstly, any cumulative effect of the turning turbines is likely

to be greatest for this directional spread as the wind direction is approximately aligned with

the main axis of the wind farm. Secondly, this directional filtering allows the measurement

instruments to be divided into three groups by location: (i) those that are always downwind of

the turbines – the downwind group, D; (ii) those that are never downwind of the turbines –

the not downwind group, ND; (iii) those that may or may not be downwind of the turbines,

11

the exclusion group, which were excluded from this directional analysis (Fig. S1). There was

sufficient data satisfying this filtering to test for statistical significance. Also, for the data to

be included the wind had to be originating from between 220° and 240° at the measurement

time and the previous 30 minutes, and also satisfy the day-night criteria for the entire 30

minutes period. This allowed the development of any underlying signal to be captured, while

still providing a significant amount of data to be retained. Departures for all the data that fit

these criteria were calculated for each sampling location, (as for Fig. 1, see above), thus

allowing the relative temperature and AH of sites downwind and not downwind of turbines to

be compared.

Statistics: Differences in coefficient of variation between ON and OFF periods for each

pseudo hour (Fig. 2) and temperature departures between D and ND groups were tested (Fig.

3) using a t-test with unequal variances using Stata13 (StataCorp, Texas).

Calculating the change in ecosystem carbon loss from Scottish wind farms: We made a

crude estimation of the effects of the temperature changes observed in this study on the C

balance of wind farm hosting ecosystems in Scotland, to provide an indication of the

potential magnitude of impact. We used the respiration rates from a peatland warming

experiment in northern England to derive the potential change in the C balance. Given there

were no significant differences in temperature between the D and ND groups during the day

when the turbines were OFF (see SI), we assumed day time fluxes were as for the control

vegetation plot at the northern England site. We assumed the 0.22 °C temperature increase we

observed at night occurred during the growing and non-growing season. We used the ambient

and warmed respiration rates from the control vegetation plots in the northern England study

and to calculate the effect of a 0.22 °C temperature rise by linearly scaling. We assumed an

12

average day length of 12 hours 20 minutes and that the growing season occurred from May to

September (as per the northern England study(24)) to calculated the fluxes over the entire

year with and without wind turbine-induced warming and calculated the percentage

difference.

Acknowledgements

This study was supported by the UK Natural Environment Research Council

(NE/H01036X/1, NE/H010351/1, NE/H010335/1). AA acknowledges financial support from

an Energy Lancaster fellowship during which all data analysis and manuscript preparation

was done. We thank Scottish Power Renewables and the land owners for allowing site access.

We thank Martin Coleman, Hemanth Pasumarthi, Salvatore Peppe, Harriet Richardson,

Kenny Roberts, Fraser Russell, Gavin Thompson and Scott Wylie for assistance in the field,

and Barbara Brooks, James Groves, Salvatore Peppe and Felicity Perry for assistance

calibrating the loggers.

Author Contributions

A.A., R.R.B, S.E.L, S.M., N.O., S.W. and J.W. conceived the research. A.A., S.L., V.S. and

S.W. undertook the field research. A.A. and R.R.B. analysed the data. A.A. and R.R.B.

drafted the manuscript with feedback and contributions from S.E.L, S.M., N.O., S.W. and

J.W.

References

1. IEA (2013) World Energy Outlook 2013 (International Energy Association, Paris, France).2. Baidya Roy S & Traiteur JJ (2010) Impacts of wind farms on surface air temperatures.

Proceedings of the National Academy of Sciences 107(42):15679-15684.3. Rajewski DA, et al. (2013) CWEX: Crop/Wind-energy EXperiment: Observations of surface-

13

layer, boundary-layer and mesoscale interactions with a wind farm. Bulletin of the American Meteorological Society 94(5):655-672.

4. Smith CM, Barthelmie RJ, & Pryor SC (2013) In situ observations of the influence of a large onshore wind farm on near-surface temperature, turbulence intensity and wind speed profiles. Environmental Research Letters 8(3):034006.

5. Armstrong A, Waldron S, Whitaker J, & Ostle NJ (in press) Wind farm and solar park effects on plant-soil carbon cycling: uncertain impacts of changes in ground-level microclimate. Global Change Biology.

6. IEA (2012) World Energy Outlook 2012: Renewable Energy Outlook (International Energy Association, Paris, France).

7. Denholm P, Hand M, Jackson M, & Ong S (2009) Land-Use Requirements of Modern Wind Power Plants in the United States. (U.S. Department of Energy, National Renewable Energy Laboratory, Colorado).

8. Lovich JE & Ennen JR (2013) Assessing the state of knowledge of utility-scale wind energy development and operation on non-volant terrestrial and marine wildlife. Applied Energy 103(0):52-60.

9. Northrup JM & Wittemyer G (2013) Characterising the impacts of emerging energy development on wildlife, with an eye towards mitigation. Ecology Letters 16(1):112-125.

10. Pearce-Higgins JW, Stephen L, Douse A, & Langston RHW (2012) Greater impacts of wind farms on bird populations during construction than subsequent operation: results of a multi-site and multi-species analysis. Journal of Applied Ecology 49(2):386-394.

11. Knopper L & Ollson C (2011) Health effects and wind turbines: A review of the literature. Environmental Health 10(1):78.

12. Baidya Roy S, Pacala SW, & Walko RL (2004) Can large wind farms affect local meteorology? Journal of Geophyscial Research 109(D19):D19101.

13. Baidya Roy S (2011) Simulating impacts of wind farms on local hydrometeorology. Journal of Wind Engineering and Industrial Aerodynamics 99:491-498.

14. Fiedler BH & Bukovsky MS (2011) The effect of a giant wind farm on precipitation in a regional climate model. Environmental Research Letters 6:045101.

15. Keith DW, et al. (2004) The influence of large-scale wind power on global climate. Proceedings of the National Academy of Sciences 101(46):16115-16120.

16. Wang C & Prinn RG (2010) Potential climatic impacts and reliability of very large-scale wind farms. Atmospheric Chemistry and Physics 10:2053-2061.

17. Zhou L, et al. (2012) Impacts of wind farms on land surface temperature. Nature Clim. Change 2(7):539-543.

18. Zhou L, Tian Y, Baidya Roy S, Dai Y, & Chen H (2013) Diurnal and seasonal variations of wind farm impacts on land surface temperature over western Texas. Clim Dyn 41(2):307-326.

19. Peng S, et al. (2013) Asymmetric effects of daytime and night-time warming on Northern Hemisphere vegetation. Nature 501(7465):88-92.

20. De Frenne P, et al. (2013) Microclimate moderates plant responses to macroclimate warming. Proceedings of the National Academy of Sciences.

21. Graae BJ, et al. (2012) On the use of weather data in ecological studies along altitudinal and latitudinal gradients. Oikos 121(1):3-19.

22. Xia J, et al. (2014) Terrestrial carbon cycle affected by non-uniform climate warming. Nature Geosci 7(3):173-180.

23. Ward SE, et al. (2013) Warming effects on greenhouse gas fluxes in peatlands are modulated by vegetation composition. Ecology Letters 16:1285-1293.

24. ESRL (Solar Calculation Details. (Earth System Research Laboratory ).25. Stull RB (1991) An Introduction to Boundary Layer Meterology (Kluwer, Dordrecht).

14

15

Figure Captions

Figure 1. Turbine proximity influences observed effects of wind farm operation on TA and

AH. The effect of distance (x) from the nearest turbine on the temperature and AH departure

during the night (a) and day (b). Positive departures occur up to 200 m from the nearest

turbine. Blue dots represent the temperature departure difference for ON-OFF periods and red

triangles the AH departure. These data can be approximated by ΔT = 0.62 - 0.12 ln (x), r =

0.75, ΔAH = 0.15 - 0.03 ln (x), r = 0.61 during the night and ΔT = -0.11 + 0.02 ln (x), r =

0.53 during the day, with dotted blue lines representing temperature and solid red lines AH.

The daytime AH cannot be approximated by a logarithmic function (r < 0.2). Black dotted

lines represent the zero departure and 200 m-distance baselines.

Figure 2. Diurnal variations in TA, TSU and TSO and AH differ during wind farm operational

and idle periods. Data points represent the average coefficient of variation (CV) ± standard

error (SE) of the TA (a), TSU (b), TSO (c) and AH (d) during the ON (blue dashed lines) and

OFF (solid red lines) periods. Fraction of the day was calculated based on sunrise and sunset

times (see Methods), with 0 representing sunrise and 0.5 sunset and 0.04 approximately 1

hour, with sunset to sunrise shaded grey. There were significant differences in the variation

during ON and OFF periods for each hour (p<0.05 for pseudo 10:00 and 11:00 for the TSO

data and pseudo 07:00 for the AH data, p<0.01 for all other data and hour intervals).

Differences in soil moisture were also investigated but no significant effects found (see SI).

Figure 3. Night-time TA and AH departures during wind farm operational and idle periods.

Grey lines represent the Black Law Wind Farm road network, black squares turbines and

coloured circles the mean TA (a) and AH (b) departures for the ON period; and mean TA (c)

and AH (d) departures for the OFF period. During the ON period the TA was warmer and AH

16

moister in the D group compared with the ND group, indicated by the concentration of

yellow to red circles on plot (a) and darker blue circles on plot (b) compared with plots (c)

and (d) respectively. During the ON period, TA (a) and AH (b) departures in D and ND areas

were significantly different (p < 0.01 for TA and p < 0.05 for AH), whereas there were no

significant differences during the OFF period, plots (c) and (d) (p > 0.10). See Fig. S1 for

illustration of D and ND groups. Examination of day-time effects

17

Fig. 1a.

Fig.1b

18

Fig. 2a

Fig. 2b

Fig. 2c

19

Fig. 2d

Fig. 3a

20

Fig. 3b

21

Fig. 3c

22

Fig. 3d

23