<<

®

Reprinted with permission from the July 2008 issue Change Effects On Wind Predicted changes in wind due to global warming are expected to be modest, but are large enough to affect the profitability of wind projects.

BY SCOTT EICHELBERGER, JAMES MCCAA, BART NIJSSEN & ANDREW WOOD oncern about the effects of summer resources of models (i.e., where do the models climate change has been the Northwest U.S. agree on the sign of the change and Cone of the motivating forces According to the study, regional where do they differ?). behind the rapid development of differences in changes wind projects. However, the predicted by different climate mod- Methodology Intergovernmental Panel on Cli- els make it difficult to draw mean- The IPCC has defined a series of mate Change states that “there is ingful conclusions based on the emission scenarios that have been used evidence for long-term changes results from any single GCM simu- as the basis for climate change model- in the large-scale atmospheric cir- lation. Therefore, the projected wind ing studies. These scenarios represent culation, such as a poleward shift speed changes from a large number story lines that provide alternative fu- and strengthening of the wester- of GCMs and for two different emis- ture scenarios. The scenarios do not ly winds” and that these observed sion scenarios are examined herein. have an assigned probability. For this changes likely will continue. Of interest is not just the mean pre- study, model simulations are based on A recent review of historic cli- dictedPercentage change, Of but Global also the Climate degree theModels A2 and Showing B1 families Increased of scenarios. mate reanalyses in Geophysical of consensus between the different These scenarios were selected because Research Letters concludes that Annual-Mean Wind Speed Values In 2050. “jet streams have risen in alti- 180˚ 120˚W 60˚W 0˚ 60˚E 120˚E tude and moved poleward in both hemispheres.” These changes in 60˚N 60˚N circulation may directly affect the energy production of existing and 30˚N planned wind projects. The large 30˚N

infrastructure investments asso- 0˚ ciated with today’s utility-scale 0˚ projects motivate an evaluation 30˚S of the possible impacts of climate 30˚S change on wind speed. A recent study found significant 60˚S 60˚S differences between the changes predicted by four global climate 180˚ 120˚W 60˚W 0˚ 60˚E 120˚E

models (GCMs) that used two % IPCC emission scenarios, both in 20 30 40 50 60 70 80 sign and magnitude, but nonethe- less concluded that a warmed cli- Figure 1: Percentage Of Global Climate Models Showing Increased Annual-Mean Wind mate may reduce the spring and Speed Values In 2050.

You may subscribe to North American Windpower online at www.nawindpower.com. Copyright © 2008 Zackin Publications Inc. All Rights Reserved. Figure 2A: Annual-Mean Difference resentative of the expected large- 180˚ 140˚W 100˚W 60˚W 20˚W scale changes, since small-scale and local effects are not resolved by the GCMs. Another limitation is the use 50˚N of daily-mean west-to-east (zonal or u) and south-to-north (meridional or v) components to compute scalar

50˚N daily-mean wind speeds. Because these components can be positive as well as negative, this meth- od can lead to a significant underesti- mation of the scalar wind speed in ar- eas that routinely experience a reversal

30˚N of the during the day (i.e., locations in which the dominant 30˚N winds are thermally driven). To the extent that such winds are not repre- sented well – due to the coarse hori- zontal resolution of the GCMs – this discrepancy may not represent a large inaccuracy of the analysis. 10˚N 10˚N Results: global Figure 1 shows a global map of m/s the percentage of GCM simulations predicting increased annual-mean −0.2 −0.1 0.0 0.1 0.2 near-surface wind speed values in their associated climate change effects current and future conditions are 2050 using the A2 emission scenar- typically span the range of projected derived from GCM output. Because io. The map is created by counting changes. The A2 scenario can be de- each GCM has its own biases, results the number of model simulations scribed as “business as usual,” while for each model are processed indi- that predict increased wind speed the B1 scenario focuses on sustainable vidually by comparing wind speed values and dividing by the total development and has half the carbon values from a historic 20th century number of simulations (i.e., 14). The map provides no information dioxide emissions of A2 in 2100. control simulation with that of a concerning the magnitude of the The surface wind fields from a 14- simulation based on one of the fu- expected change, but it does pro- member ensemble of GCM simula- ture emission scenarios. vide information on the agreement tions were evaluated to analyze the Mean wind speed values repre- between the model simulations. impact of global climate change on senting the present-day climate are Dark red areas correspond to near-surface wind speeds across the computed using daily-mean GCM regions in which most of the mod- globe. Data from these GCM simu- output over the period 1991-2000. els predict stronger surface wind lations are publicly available as part Mean wind speed values in the future speeds. Conversely, areas of dark of the World Climate Research Pro- are computed using daily-mean GCM blue imply that most of the models gramme’s (WCRP) Coupled Model data from the period 2046-2055. predict weaker surface wind speeds Intercomparison Project phase 3 Wind speed differences are reported (i.e., only a very small percent- (CMIP3) multi-model data set. The at 10 meters above ground level. At age of the models predict stronger CMIP3 data set contains GCM turbine hub heights, these differences surface wind speeds). Areas of pale output for historic periods and are expected to be slightly greater, but color indicate regions in which projections of the future climate should retain the same spatial patterns a clear consensus does not exist under different emission scenarios. and relative magnitudes. among the GCMs. Since the spa- Although the CMIP3 data set con- The horizontal resolution of the tial pattern of the predicted wind tains additional GCM simulations, GCMs used in this study ranges speed changes is very similar for only 14 simulations contain suffi- from approximately 200 kilometers both the A2 and B1 emission sce- cient data to perform the analysis. to 450 kilometers. Therefore, the narios, results will only be shown Wind speed values representing results of the analysis are only rep- for the A2 emission scenario.

You may subscribe to North American Windpower online at www.nawindpower.com. Copyright © 2008 Zackin Publications Inc. All Rights Reserved. Figure 2B: Annual-Mean Percent Increase dicted change in annual-mean wind speed (2A) as well as the percentage 140˚W 100˚W 60˚W 180˚ 20˚W of models that predict an increase in the annual-mean (2B), winter-mean

50˚N (2C) and summer-mean (2D). The predicted changes from the indi- vidual models vary greatly from the ensemble average. 50˚N However, most models predict an increase in annual mean wind speed values across the northern part of the continent and along a broad swath from Hudson southward into Texas and parts

30˚N of Mexico, including the Yucatan peninsula (Figure 2B). There is 30˚N less agreement among the models about decreases in mean annual wind speed values. The greatest agreement is found in the moun- tainous West, the Southwest and along the Atlantic seaboard, but 10˚N much uncertainty remains. 10˚N The areas that show the greatest agreement among models also show % the largest predicted annual mean wind speed differences (Figure 2A). 20 30 40 50 60 70 80 Averaged across all the models, On an annual basis, climate addition, a separate analysis of the the predicted changes in the mean change is predicted to cause stron- CMIP3 data set showed that annual wind speed are relatively ger surface wind speed values across tracks are expected to continue to small. Changes in the annual mean the boreal regions of the Northern shift poleward in response to climate are predicted to be less than +/- 0.2 Hemisphere, including much of change forces. meters per second (m/s) in general Canada, Siberia and northern Eu- In some areas, the surface wind and less than +/- 0.1 m/s for much rope, and in tropical and subtropical speed changes shown in Figure 1 ap- of North America. regions in Africa, and Central and pear to be associated with the predict- Since the ensemble average is South America. However, Green- ed changes in the mean position of the composed of many dissimilar fields, land, southern Europe, China, In- storm tracks. For example, the pre- the magnitude of the predicted dia, southern Australia and much of dicted stronger surface wind speeds change is small. However, individual the west coast of South America are across northern Europe and weaker models show predicted wind speed expected to experience decreasing surface wind speeds across southern differences of +/- 0.5 m/s, which are wind speed values. Europe are consistent with a poleward large enough to affect the profitabil- Inter-model contrasts are larg- shifting storm track across the conti- ity of existing and future wind proj- est in western North America, sub- nent – great for Germany, but sad for ects. For the wind-rich area of west Saharan Africa, broad swaths of Spain. Similarly, the predicted changes Texas, the models show a strong Eurasia, Brazil and the Andean re- between 20 degrees South consensus of an increase in the an- gion of South America. The annual and 40 degrees South in South Ameri- nual mean wind speed. analysis masks seasonal variation in ca and Australia correspond well with In addition, the relatively small the predicted wind speed changes; a poleward shift in the position of the predicted changes in the annual in some areas, these changes differ storm tracks. mean wind speed values mask a in sign across . strong seasonal signal. Most models Recent research has found that Results: North America agree that wind speed values likely the jet streams and associated storm Figures 2A, 2B, 2C and 2D focus will increase over much of North tracks have been shifting poleward specifically on North America and America during the winter months during the period 1979-2001. In include the ensemble-mean pre- (i.e., December, January, February),

You may subscribe to North American Windpower online at www.nawindpower.com. Copyright © 2008 Zackin Publications Inc. All Rights Reserved. Figure 2C: Winter-Mean Percent Increase with the strongest consensus be- tween the models from Hudson Bay 180˚ 140˚W 100˚W 60˚W 20˚W south into the central U.S. The predicted increase in the win- tertime mean across the Upper Mid- 50˚N west of the U.S. will bolster the many wind projects in the region that are predominantly wintertime peaking. 50˚N Most models agree on a decrease in winter wind speed values in Baja and around the Gulf of California. Dur- ing the summer months (i.e., June, July, August), model agreement shows nearly the opposite signal compared to 30˚N the winter months, with most models

30˚N agreeing on a decreased wind speed in the northern half of the continent, part of Mexico and the Caribbean, and an increase in Alaska and along the Gulfs of Mexico and California. The results of this evaluation of 10˚N predicted changes in mean wind 10˚N speed show that the predicted ensemble-mean changes in annual % mean wind speeds are expected 20 30 40 50 60 70 80 to be modest. However, seasonal Figure 2D: Summer-Mean Percent Increase changes and changes predicted by individual models are large enough 100˚W 180˚ 140˚W 60˚W 20˚W to affect the profitability of existing and future wind projects. l

50˚N

At 3TIER, Scott Eichelberger is the director of resource assessment, James 50˚N McCaa is director of high performance computing, Bart Nijssen is chief techni- cal officer and Andrew Wood is senior scientist for hydrology. For more infor- mation, e-mail [email protected]. The authors acknowledge modeling 30˚N groups the Program for Climate Mod-

30˚N el Diagnosis and Intercomparison and the World Climate Research Pro- grammer’s (WCRP) working group on coupled modeling for their roles in making available the WCRP CMIP3 multi-model data set. Support of this 10˚N data set is provided by the Office of 10˚N Science, U.S. Department of Energy. In addition, the authors would like to % thank Geoff Dutaillis of Babcock & 20 30 40 50 60 70 80 Brown for helping to fund this study.

You may subscribe to North American Windpower online at www.nawindpower.com. Copyright © 2008 Zackin Publications Inc. All Rights Reserved.