Surface Wind Speed Distributions: Implications for Climate and Wind Power Scott B. Capps and Charles S. Zender Department of Earth System Science University of California, Irvine Thanks: W. Liu (JPL), A. Monahan (UVic), P. Rasch (NCAR), D. Shea (NCAR), M. McPhaden (PMEL), D. Wang (UCI) Source: NASA
Sub-Gridscale (SGS) Winds In a typical atmospheric model: U(x,y,t) ~104 km2 10-30 minutes Finer spatial and temporal fluctuations are considered SGS
Sub-Gridscale (SGS) Winds Average spatial and temporal wind speed distribution. (Frontal Air-Sea Interaction Experiment (FASINEX): 15minute periods, 5 buoys over ~40x40km, ~1 day) QuikSCAT (0.25 x0.25 resolution) 10m winds near Japan with superimposed T85 grid (1.4 x1.4 resolution). Missing values outside swath, over land, and near precipitation. Level 3 QuikSCAT data from http://podaac.jpl.nasa.gov/
Motivation: SGS Winds and Dust Mobilization Sandblasting (the bombardment by saltating particles) is thought to be the ultimate source of most fine dust emissions (C. S. Zender, personal communication). Smaller particle sizes require higher winds to overcome adhesive forces. Threshold Saltating particle size distribution and wind friction speed (Grini, Zender and Colarco, 2002, GRL).
Motivation: SGS Winds and Dust Mobilization How do we represent this relationship with a single mean wind speed? Saltating particle size distribution and wind friction speed (Grini, Zender and Colarco, 2002, GRL).
Motivation: SGS Winds and Non-linear Surface Fluxes Non-linear fluxes computed from a mean wind speed are not equal to those integrated over the entire wind speed distribution: 2 2 = C Dz U z C Dz pw U U du 0 The two-parameter Weibull PDF has been shown to closely represent wind speed distributions (Justus et al., 1979; Pavia and O'Brien, 1986; Cakmur et al., 2004; Monahan, 2006b): where k = shape and c = scale.
Storm Track Region PDF: Mean = 10.0 m/s Shape = 2.67
Storm Track Region PDF: Mean = 10.0 m/s Shape = 2.67
Storm Track Region PDF: Mean = 10.0 m/s Shape = 2.67 ~30% increase in momentum flux Tau (bin avg) = 0.18 N/m2 Tau (mean wind) = 0.14 N/m2
Trade Wind Region PDF: Mean = 7.0 m/s Shape = 6.7
Trade Wind Region PDF: Mean = 7.0 m/s Shape = 6.7 Tau (bin avg) = 0.066 N/m2 Tau (mean wind) = 0.063 N/m2
Representing SGS Winds: Weibull PDF Shape can be calculated given standard deviation and mean wind speed: Surface wind Characterization and Comparison:
Sea Surface Wind Speed Distribution Characterization and Comparison Top row: QuikSCAT 2000-05 mean 10m ocean surface wind speed, standard deviation, Weibull shape, and 90th percentile. Rows two and three: Differences NCEPII-QuikSCAT and CAM3-QuikSCAT. Top left corner: Mean absolute bias/mean bias. Top right corner: Max/Min/RMSE. Stippling indicates 5% level of significance (Capps and Zender, 2008, J. Climate).
Representing SGS Winds Within a GCM A Weibull PDF is fitted using the prognostic gridcell mean wind speed (Justus et al., 1978): Shape: where Ckis 1.05. Scale: 1) The PDF is discretized into equal-probability wind speed bins. 2) Surface fluxes are calculated for each bin. Experiment: Fluxes averaged over 4 bins. Control: Fluxes computed from 1 bin = mean wind speed.
Representing SGS Winds Within a GCM A Weibull PDF is fitted using the prognostic gridcell mean wind speed (Justus et al., 1978): Shape/Scale Mean Wind Speed (m s-1)
Instantaneous Response to SGS Winds Mean momentum flux (CAM3 2000-05 June). Average instantaneous momentum flux response (4bin 1bin) to sub-gridscale winds (CAM3 2000-05 June).
Instantaneous Response to SGS Winds Unstable surface layer Average instantaneous SHFLX and LHFLX response (W m-2, 4bin - 1bin) to sub-gridscale winds (CAM3 2000-05 January). Ocean surface energy and momentum fluxes vs. wind speed (CAM parameterizations).
Instantaneous Response to SGS Winds Stable surface layer Average instantaneous SHFLX and LHFLX response (W m-2, 4bin - 1bin) to sub-gridscale winds (CAM3 2000-05 June). Ocean Surface Energy and Momentum fluxes vs. Wind Speed (CAM parameterizations).
Climate Response to SGS Winds Increased precipitation in central Africa and America boost LHFLX and cool surface temperatures. Reduced precipitation in western Australia lead to a reduction of LHFLX and warmer surface temperatures.
Climate Response to SGS Winds SHFLX reduced where precipitation increased. Reduced precipitation results in a SHFLX increase.
Climate Response to SGS Winds The non-linear momentum flux response has slowed near surface winds.
Climate Response to SGS Winds Top row: QuikSCAT 2000-05 mean 10m ocean surface wind speed, standard deviation, Weibull shape, and 90th percentile. Rows two and three: Differences CAM3-QuikSCAT and CAM3/SGS-QuikSCAT. Top left corner: Mean absolute bias/mean bias. Top right corner: Max/Min/RMSE. Stippling indicates 5% level of significance (Capps and Zender, 2008, J. Climate).
Climate Response to SGS Winds SLP (hpa) 500-hPa Height 2000-2005 mean DJF sea level pressure (left, hpa) and 500-hPa geopotential height differences (right, gpm). CAM3-NCEPII (top row) and CAM3 four-bin Wind Speed PDF-NCEPII (bottom row). Top left corner of each plot: Mean absolute bias/mean bias. Top right corner: Max/Min/RMSE. Stippling indicates 5% level of significance (Capps and Zender, 2008, J. Climate).
Climate Response to SGS Winds 2000-05 Global Mean Absolute Biases and Percent Changes (Capps and Zender, 2008, J. Climate).
SGS Winds and Implications for Ocean Transport CCSM3 climatological wind stress bias (N m-2). (Large and Danabasoglu, 2006, J. Climate).
Stability Dependent Wind Speed PDF Shape has been predicted using just the mean wind: Can we formulate the shape or standard deviation as a function of atmospheric stability?
Stability Dependent Wind Speed PDF Shape has been predicted using just the mean wind: Can we estimate the standard deviation as a function of atmospheric stability? Wind component variances calculated following Holtslag and Moeng (1991), Nieuwstadt, F.T.M. (1984) and Panofsky and Dutton (1984). Three day TKE, wind speed standard deviation, friction velocity and surface kinematic buoyancy flux timeseries for a CAM3 desert gridcell.
Wind Power Distribution Over the Global Ocean Liu et al., 2008, GRL
Wind Power Distribution Over the Global Ocean Extending Liu et al., we ask the following: 1) How much power exists at typical modern wind turbine hub heights (~80m)? 2) How is this power influenced due to surface layer stability? 3) How much of this power is extractable (usable)?
Wind Power Distribution Over the Global Ocean Extending Liu et al., we ask the following: 1) How much power exists at typical modern wind turbine hub heights (~80m)? 2) How is this power influenced due to surface layer stability? 3) How much of this power is extractable (usable)? Wind profiles are determined using: 1) Surface stress computed from QuikSCAT measurements (Large et al., 1994, Rev. Geophys.). 2) OAFLUX SST, 2m air temperature and humidity, and SHFLX. 3) Monin-Obukhov Similarity Theory (MOST).
80m Wind Power Over the Global Ocean Extending Liu et al., we ask the following: 1) How much power exists at typical modern wind turbine hub heights (~80m)? 2) How is this power influenced due to surface layer stability? 3) How much of this power is extractable (usable)? Wind profiles are determined using: 1) Surface stress computed from QuikSCAT measurements (Large et al., 1994, Rev. Geophys.). 2) OAFLUX SST, 2m air temperature and humidity, and SHFLX. 3) Monin-Obukhov Similarity Theory (MOST). Wind speed profiles given 6 and 14 ms-1 10m neutral stability wind speeds for an unstable (20Wm-2<shflx<40Wm-2), neutral (0Wm-2) and stable (-40Wm-2<shflx<-20Wm-2) surface layer (Capps and Zender, 2009a, submitted GRL).
80m Wind Power Over the Global Ocean 2000-06 DJF (top) and JJA (bottom) MOST 80m wind speed minus logarithmic 80m wind speed (ms-1). Positive T2m minus SST contours in magenta, negative in blue and zero in black. Capps and Zender, 2009a, Submitted GRL.
80m Wind Power Over the Global Ocean Wind power is determined using discrete 2x daily QuikSCAT measurements: and, a fitted Weibull PDF (Assume constant air density): E w / A=1 /2 c 3 1 3/ k Third moment 3 p U U du w 0
80m Wind Power Over the Global Ocean Global mean 80M power is ~1.6x 10m power. Capps and Zender, 2009a, Submitted GRL. 2000-06 DJF (top) and JJA (bottom) 80m wind power from full Weibull PDF (Wm-2). T2m minus SST contoured with positive (negative) regions in magenta (white) and zero in black.
80m Wind Power Over the Global Ocean DJF JJA 80-10m Wind Speed Difference (m s-1). 80m MOST Wind Speed minus 80m Log Wind Speed (m s-1). 80m Wind Power Density from Full Weibull PDF (W m-2). 80m Multiple of 10m Wind Power. Capps and Zender, 2009a, Submitted GRL.
80m Wind Power Over the Global Ocean DJF JJA 80-10m Wind Speed Difference (m s-1). 80m MOST Wind Speed minus 80m Log Wind Speed (m s-1). 80m Wind Power Density from Full Weibull PDF (W m-2). 80m wind power >6x 10m power east of continents during summer. Capps and Zender, 2009a, Submitted GRL.
Usable Wind Power 2000-06 QuikSCAT Wind Histograms with Fitted Weibull PDFs. Usable Winds and Power in Green. 2000-06 QuikSCAT Wind Speed Standard Deviation (m s-1). Capps and Zender, 2009b, In Prep.
Usable Wind Power Capps and Zender, 2009b, In Prep. DJF JJA 2000-06 Usable Power Percent of Full Power.