- - - -

8.5

SCENARIOS FOR CLIMATE CHANGE IMPACT?

ASSESSMENT& IN THE GREAT lAt<ES REGIO~

Kenneth E..Kunker.and Stanley A. Changnon

.Mid~estern ClimateCenterIllinois.State Water Survey

. Champaign, Illinois

Thomas E. Croley IIand Frank H. Quinn

Great lakes Enyi~onmental Research laboratory, AnnArbor,Michigan

1. INTRODUCTION, ,

The purpose of this study was to assessthe effect of changed climate conditionssimilartothose predicted by some GCMs on the temporalbehavior of various components of the hydrologiccycleof the Great Lakes basin, as measured overa series of seasons, years, and decades. Wefurther sought a senSitivity-type Investigation of the ...

effects of temporal fluctUationsasSociated withrather extreme climate conditions such as onesmuch warmer and drier, or much warmer andwetter, than the basin's current climate and withlarger changes than predicted by GCMs' To thisend, 'we developed a series of four climatescenarios each coveringa 42-year period of time.These were used to study the primary watermanagement problem Inthe Great Lakes withItsmany competing.demands for water which Is theInterannual.fluctuations in net basin supplies andlakelevels. "

The calculation of the hydrologicsystemeffects Includingthose to the net basin supply andlake levels, was done u~ifl9 a ~te-of-the-artGreat lakes hydrologic model (Croley, 1994).Operation.ofthisconipl~xmodel requires the inputof temporally-detailed,mete6rolo.glcafdata for'asmany statioris ,as possible d~tributed across. the800,000 'km2 'basin. To meet the model'soperational demands, and those Imposed by thisstudy's requirement to have 40-year tests ofeffects, long-term (40-year) measures of thehydrologically-relevantatmospheric conditions(i.e.,daily values of temperatures, precipitation,cloud

. .Corresponding author address: Kenneth E.

Kunkel, Midwestem Cfimate Center, Illinois State WaterSurvey, 2204 Griffith Dr., Champaign, IL 61820-7495.

212 AMERICANMETEOROLOGICAL SOCIETY--- ----

-.- -- .------- " -..-

cover, humidity,and winds) and from a densenetwork'ofstationS(orgrid points) had to exist"overthe entirebasin. To furtherachieve the desired testof differentand more extreme climate conditions,severai long-termsets of detailed meteo'rological

.data embracing areas the -sizeof the Great lakesbasin had to be found, or developed, for widelydifferentclimaticzones; or climate scenarios.

2. REQUIREMENTSFOR CLIMATESCENARios

The climate scenarios needed to achievethe objectives of the study had to possess thefollowingcharacteristics:"

. Dailyprecipitationand dallymaximum andminimumtemperature data were requiredat a highspatialdensity of approximately1stationper 1,000.~2 overthe Great Lakesregionof800,000km2. .. , Hourly observations of wind, humidity,clou~ cover, and temperature .wererequired at a spatial density ofapproximately1 station per 10,000 km2.. The hourly' ani:t daily,data time seriesneeded to be of considerable length.inorderto studyvariability.At least 30 yearsofdata, and .ldeally40 years or more; wereconsideredessential. .

. The data must possess spatial andtemporalcoherencethat Is realisticand

t}I

consistentwiththe physicallaws governingatmospheric behavior.

. At least four widely different climatescenarios were required. These mustbracket the range of values estimatedunder doubled C02 conditions derivedfrom widely used GCMs.

t

II

t

Itwas not considered scientificallyvalid toattempt to create 40 years of synthetichourly anddaily meteorological data for 1,000 locations andacross four different climatic zones the size of theGreat LakesBasin. Anotherapproach, usingdirectoutputof currentGCMs,was also rejected becausea singlegrid box represents an area of20,000 km2or more which is far too crude spacing for input tothe hydrologymodel. A thirdpossibilityis the use ofregional climate models that are embedded inGCMs. However, the massive computerrequirements of this approach and theirdevelopmerit have thus far limited the length oftime series to 1 to 2 years in length (Bates et al.1994). Other approaches were needed.

3. CLIMATESCENARIOSELECTION:SPATIALTRANSFERS

The technique selected to define theweather conditions required for this study was"spatial transfers" or "climate transposition." Were-Iocated the Great lakes basin to four otherclimatic zones in the central and southern UnitedStatesto samplethe desired climatic differences influctuations over time. Robinson and Finkelstein(1989) assessed means of developing climatescenarios for impact studies like this, and theysuggested, as one useful empirical method fordeveloping climate scenarios, climate analogues.Climateanaloguesuse historicaldata to representa changed and often extreme climate conditionperceivedto existat some Mure time. One form ofclimate analogue is labeled as "spatial transfers"(Changnon 1991). The underlyingrationale for thisapproach is the expectationthat future changes inbasin climate conditions may approximatelatitudinal and/or longitudinalshifts.

The major advantage of this approach isthat the transposed data represent an actualclimate time series (Robinson and Finkelstein1989). They exist and thus conceivably couldhappen again somewhere else in the future. Allkey features of the climate are realistic, includingthe temporal variability and the frequency andmagnitude of extremes. Further, the spatial

--:) j\2)

relationships are obviously realistic. One canquestion whether the exact climate conditionscurrentlyexistingin the southwestern United Statescould ever exist in the Great lakes region, but forthe sakeof achievingthe desiredsensitivityanalysisfrom drastically different climates, we haveassumed this possibilityexists.

This technique takes advantage of theexistingdetailed climatic data record in the UnitedStatesandCanada. In this study we used data forthe period1949-1990. Data in digital form for bothhourly and daily weather elements were readilyavailable. By using. the existing long-termobservations,we were able to create a wide rangeof surface climate conditions. Also, by carefullychoosing our latitudinal and longitudinal shifts,wewere able to realize climates that approximatetemperature and precipitationoutcomes predictedby GCMs or that-are more extreme. Finally, ourpastexperiencewith thoseattemptingto assesstheimpactof the climatechangehas revealed that useof "actual"historicaldata led to improved credencein understanding and accepting outcomes.

There aredisadvantagesto this approach.One isthatthe data do not maintain current majorgeographic controls on climate. A major examplein thisstudyisthe effectof the Great lakes on localsnowfall,temperature,and otherclimatic elements.Datain the GreatPlains (to take one possibleshift)will notexhibitlocal gradients created by the lakes.Thus, some adjustments in the transposed datawere required,negating some of the simplicity thatmakes this method attractive. larger scalegeographicfeaturesalso playa role. For instance,the locations of the Rocky Mountains, the Gulf ofMexico,the Atlantic Ocean, and Hudson Bay haveinfluenceson the climate of North America. Theseinfluencesin all likelihoodwill not shift in a changedclimate in a straightforward fashion. For instance,it is highly improbable that the effect of the RockyMountainson the current climate of Kansaswill besimilar in a future climate of the southern Greatlakes. Thus, to the degree (unknown) that thiseffect is important, transposed data will beunrealistic. .

4. ESTIMATIONOF LAKE EFFECTS

PastWater Survey research based on thelake Michigan basin used a climatologicaltechnique for defining the extent ofthe lake effecton upwind and downwind sides of the lake onmonthly, seasonal, and annual precipitation,terriperatures, and other weather conditions(Changnon, 1968). This technique was used to

9TH CONF.ON APPLIEOCLIMATOLOGY 213

~ n ________ - - -. - -- - ---

derive measures of lake effects around each lakefor the four majorseasons(winter,spring, summer,and fall), and for seven wea~er parametersincluding precipitation, maximum temperature,minimumtemperature,averagedally temperature,cloud cover, wind speed, and atmospheric watervapor pressure.

This empirical three-step techniqueinvolved first the deletion of existingstationswithin80 km of the lee shores, assuming they arepotentiallylake influenced. Then, pattems of eachweather conditionwere constructed based on theremaining data. Third, the interpolated values ateachstationwerecompared with the actual valuesto determine a percentage change due to lakeeffects. The use of atmospheric models tosimulatethe effectsof the Great Lakes on regionalclimate conditionshas begun but to date is limitedto examiningshort discrete periods of time (Bateset aI., 1994).

. Mapsforeachelementweredevelopedtoshow the spatial pattern with and without lakeeffects, and for the lake-effect differences. Themaps of lake effect differences were theninterpreted and digitized on a 0.5° latitude-longitudegrid. To obtaina measure of lake effectsunder more extreme weather conditions, theprocess for estimating lake effects was repeatedfor the five wettest and fIVedriest years of eachseason. These represent natural changes thathaveoccurredunder more extreme conditionsandwhichconceivablyportraythe magnitudeof changethat could be associated with future climatevariationswhich could be wetter or drier.

For each meteorological elementevaluated,two.sets of maps were generated. Thefirstincludedall observations. This map was usedprimarilyto establishthe spatial distributionpatternwithin the Great Lakes basin and the surroundingareas. The spatialpatteminthe basin incorporatesboth lake-induoedchanges and those produced bythe broad-scalecfimaticconditionsof the region. Arelatively large region surrounding the basin wasusedto providean adequatemeasureof the spatialpattern characteristicswell beyondthe lake-effectregions.

A second map was then generated, aftereliminating all stations in the 80-km lake effectband. The pattern existingin the no-effect regionsurrounding the basin was used as the primaryguide in establishing the climatological patternassumedto existif no lakeeffect was present. Themapswereoveriaidandthe lake effect in the basinwas derived by determining differences betweenthe two sets of values. These differences valueswere usedto generatea digitizedmap of computed

214 AMERICANMETEOROLOGICALSOCIETY

differences within the basin.After lake-effect maps had been drawn for

each conadian, the analyzed data were digitized ongrid squares for input to the hydrologic model.

5. CLIMATE SCENARIOS

We created four scenarios by movingtheGreat lakes basin to the south and west of itscurrent position. In all of these, the relativespatialrelationshipsof the geography of the Great lakeswere preserved with the outline of the basin laidover the existing climate network. Table 1 givesbasic information about these scenarios. Figures1-4showsthe locationof the Great Lakes basinforthese scenarios.

Croleyet at (1995)describedthe changesin climate for these scenarios and the associatedchanges in net basin supplies and lake levels.Scenarios1 and2 areassociatedwith temperaturechangesof5-7°C,the upperend of estimates fromdoubled-COz GCM simulations (lPCC,. 1992).Scenarios 3 and 4 represent more extremeconditions, with temperature changes of 8-11°C.With the exceptionof Lake Superior for scenarios1 and 3, the scenarios are associated withsubstantial precipitation increases. However,accompanyingincreasesin landevapotranspirationand lake evaporation negate these precipitationincreases. Net basin supplies decreasesubstantially in scenarios 1 and 3, but areapproximatelyequal to the present climate valuesin scenarios2 and 4.

These scenarios are characterized byincreased interannual variability in precipitation,overiandevapotranspiration,and lake evaporation.As a result, net basin supplies exhibit substantialincreasesIn interannual variability(Table 2). Evenfor scenarios 2 and 4 where mean values do notchange substantially, the large increases invariability would create severe problems for lakewater management.

x

)

.-- _._------------

Table 1. Basic Information for climatetranspositionscenarios.

Number Latitude Shift longitude Shift

1 5° South 10° West2 5° South 0°3 10° South 11° West4 10° South 5° West

"

..~,

t,«.\c'o.

l~-f:.~~<\\-

Ji

1-.".r

j',:\;

i~~

~J

I~~

~

Li"

,\.

The adjustment of the climate data toinclude lake effects has mixed impacts on theresults. In the case of scenario 1 (Table 3), lakeeffect adjustments decrease net basin supplies inSuperior,Michigan,and Erie,with increases notedin Huron and Ontario. The changes in variabilityare also mixed with increases in Superior andOntario,decreasesin Michigan,and liWechange inHuron and Erie.

REFERENCES

Bates, G.T., Hostetler, S.W., and F. Giorgi, 1994:Two-year simulation of the Great lakesRegion with a coupled modeling system.Mon.Wea.Rev.,122, In press.

Changnon, SA, 1991: Development and use ofclimate analogues. Proceedings,Symposium on Climate Scenarios,Universityof Waterloo, Waterloo, Ontario,49-58.

Croley, T.E. II, 1994: Hydrological impacts ofclimate change on the laurentian Greatlakes. Trends in Hydrology, 1, 1-25.

Croley, T.E., II, F.H. Quinn, K.E. Kunkel, and SAChangnon, 1995: Potential Great Lakeshydrologyand lake level Impacts resultingfrom global warming. Preprints, SixthSymposium on Global Change Studies,Dallas, TX, January 15-20.

Intergovemmental Panel on Climate Change,1992: The Supplementary Report to theIPCC Scientific Assessment Ed. J.T.Houghton, BA Callander, and S.K.Vamey, Cambridge UniversityPress, 200pp.

Robinson, P.J., and P. Finkelstein, 1989:Strategiesfor the Developmentof ClimateScenarios for Impact Assessment. U. S.Environmental Protection Agency,ResearchTriangle Park, NC, 83 pp.

Figure 1. location of Great lakes basin forscenario 1.

9THCONF.ONAPPLIEDCUMATOLOGY215

Table 2. Standard Deviation(mm) of Annual Net 8asinSupplies for Current Ctimate("8ase") and

Percent Change for ClimateScenarios

Scenario

Lake 8ase 1 2 3 4

Superior 164 +21% +32% +39% +78%Michigan 203 +62% +53% +38% +100%Huron 165 +103% +47% +108% +133%Erie 268 +87% +51% +86% +97%Ontario 304 +102% +57% +141% +74%

Table 3. Comparison of simulationresultsforannual total net basinsupplies (mm)with and

without adjusting station data for estimatedlakeeffects. These are resultsfor scenario 1.

StandardDeviationof

Annual Total Annual Total

Without With Without WithLake Lake Lake Lake

Lake Effect Effect Effect Effect

Superior -24 -220 198 231Michigan 419 249 329 281Huron 468 554 335 337Erie 933 834 501 500Ontario 1941 2163 616 734

Figure 2. loca.tionof Great lakes basin forscenario2.

Figure 3. location of Great lakes basin forscenario 3.

216 AMERICANMETEOROLOGICALSOCIETY, .1

-- --- - .-.-

Figure4. location of Great lakes basin for. scenario4.

<,M~.;:~~)i'

.&-~'

...-..0;.

.-

NINTH CONFERENCEON

APPLIED CLIMATOLOGY

January 15-20, 1995 Dallas, Texas

Sponsored by

American Meteorological Society

Cosponsored by

World Meteorological Organization

Front Cover: The two figures, prepared from Landsat MSS data, depict the expansion of urban land cover in the Dallas-

Fort Worth, Texas region between 1974 and 1989. The urban-related land cover features are represented by light blue andgrey, while the arable land is represented by brown, green, and red. The displacement of arable land by urban expansionis evident around and between the two cities. The increased urbanization around the Dallas-Fort Worth International Airportis clearly visible (slightly right of center of figures). During this time interval the population of the Dallas-Fort WorthMetropolitan area increased from 2,467,000 to greater than 3,766,000.

The influence of land uselland cover on climatological values of the diurnal temperature range is presented in paper 6.4included in this preprint volume on page 163 by Kevin P. Gallo, NOAAINESDIS,Washington,DCandT. C. PetersonandD. R. Easterling.

Coverfigures courtesy of NOAAINESDISlNational Climatic Data Center and USGSlEROS Data Center.

All Rights Reserved. No part of this publication may be reproduced or copied In any form or by any means - graphic, electronic, or mechanical,including photocopying, taping, or Information storage and retrieval systems - without the prior written permission of the publisher. Contact AMSfor permission pertaining to the overall collection. Authors retain their individualrightsand should be contacted directly for permission to usetheir materialseparately. The manuscripts reproduced herein are unrefereed papers presentedat the Ninth Conference on Applied Climatology.Their appearance in this collectiondoes not constiMe formal publication.

AMERICAN METEOROLOGICAL SOCIETY45 Beacon Street, Boston, Massachusetts USA02108-3693

..