1

CLIMATE VARIABILITY AND CHANGE: THEECONOMIC VULNERABILITY OF THE

SKIING INDUSTRY AND SURROUNDINGCOMMUNITIES IN ARIZONA, USA

(DRAFT DO NOT CITE WITHOUTPERMISSION OF AUTHORS)

Rosalind H. Bark-Hodgins1 and Bonnie G.Colby

University of Arizona

ABSTRACTClimate change models predict declining snowpack, shorter and more variable snowseasons, warmer winter temperatures withincreased incidence of winter snowpack meltand sublimation loss, earlier spring snowmelt,and an increase in the elevation at whichseasonal snowpack can be maintained. Theimplications are significant for the ski industrybecause beginner skiers tend to learn at lowerelevation ‘local’ ski areas [1] and these sameskiers are more likely to quit if ski seasons arepoor [2]. Adaptation strategies such assnowmaking can reduce climate changevulnerability [3] by increasing snow packdepth, durability and season reliability.However, snowmaking is expensive; costs,excluding capital costs, range between $500and $4,000 per acre foot of snow dependingon system efficiency [4], and crucially for thesouthwest snowmaking also requires largevolumes of water2 [5]. Previous research hasfocused on low elevation ski resorts in Europeand Canada, this paper adds to the literatureby investigating the impacts of climatevariability and change on low latitude, highelevation ski resorts in Arizona, USA. Arizonaski areas experience high inter-annualvariability in snow reliability in terms of totalsnow pack, season length, and season timing.Severe winter drought conditions are common,for example in the 1983/84 and 2001/02seasons. Such seasons, may preview longer 1 Corresponding author address: Rm 319, AREC,

University of Arizona,

PO Box 210023, Tucson, AZ 85712-0023. email:

rbark@email.arizona.edu2 1 acre foot of snow = 139,222 gallons or 527,011

liters of water.

term climate change impacts. Two casestudies in Arizona, Sunrise Park andSnowbowl, highlight the opportunities andchallenges of manmade snowmaking as anadaptation strategy to climate change and alsoto more general climate variability.

INTRODUCTIONThe predictions of climate change models:less snow, reduced snow pack, and shorterand more erratic snow seasons, with morepronounced effects at lower elevation, areworrisome for ski3 industries around the world.Less research has been done on the impactsof climate change on low latitude, highelevation ski resorts, such as those thatpredominate in the southwest USA. However,this research does inform this study about thelikely threats and how mitigation andadaptation strategies might reduce theeconomic vulnerability of two Arizona skiresorts, Sunrise and Snowbowl, and theirsurrounding communities to climate variabilityand change. Of particular economic concern isthe possible impact climate change will haveon Sunrise which is a more marginal alpinecommunity that is heavily dependent on winterrecreation.

Assessments of climate change impacts on skiresorts have been completed in Australia [7,2], Austria [8], Canada [9, 10], Scotland [11],Switzerland [12, 13] and the United States[14]. All these studies predict that climatechange will have negative consequences forlow elevation ski resorts: without adaptationstrategies ski resorts will have to make a profitduring a shorter season. Warming is predictedto reduce snow cover and thereby fewer daysin the season will meet minimum operationalsnow base for winter sports. These thresholdsvary with the sport: 10 cm for cross countryskiing, 30 cm for downhill skiing andsnowboarding, and >30 cm above the tree linein rocky terrain [1, 3]. Mountain managers maydecide not to open their specific resort untilnatural and manmade snow base is deeperthan these minimums. In fact Arizona’s skiresorts have higher snowpack minimums than

3 We use ‘ski industry/resort’ as shorthand for ski

and snowboard industry and ‘skiers’ for skiers and

snowboarders.

2

those above: Snowbowl’s snowmaking baserequirement is 64 cm [15] whilst Sunrise’s twosnowmaking expansion plans quote a low of36 cm to a high of 61 cm.4 Reasons for thesehigher thresholds are the high elevation of theski resorts both of which are above the treeline: elevation at Sunrise ranges from 2,836mto 3,354m and at Snowbowl from 2,805m atthe base to a peak 3,506m, consequentsteeper slopes, and losses in the dry interiorclimate [16]. Additionally it is difficult toimagine that many skiers would be satisfiedwith the experience of skiing on a 30 cm snowbase. Possible reasons for the differencebetween the two sites are the steeper slopes,high wind conditions, and the suboptimalsouthwest-west aspect of the Humphrey’s podof ski slopes at Snowbowl. Average snowconditions are similar at both resorts. Theseexamples illustrate that the impact of climatevariability and change needs to be addressedon a site-by-site basis that takes account ofsnow conditions, snow requirements, and theeconomics of the ski enterprise.

Breiling and Charamza [1] investigated theprobable impacts of climate change on the skiindustry in Austria. This study is relevant forSunrise, although Austria is located at a highlatitude (47º-48º), many of its ski areas arelocated in small alpine communities that aredependent on winter tourism. Sunrise is justsuch a resort; it is located in the somewhatremote White Mountains of central easternArizona. The Austrian study predicts that a2°C warming will reduce snow cover bybetween 47% and 79%, from a 1965-1995baseline. The forecast reduction in skiabledays is greater because of the minimum snowdepth required for winter recreation. They notethat even with the warming there will still bearound one year in two where snow depth iswithin the range of the baseline data. Suchincreased unpredictability is highly damagingto this infrastructure-intensive industry. Theyposit that a 2°C warming will necessitatestrong adaptation, and that the costs ofincreased snowmaking may make all lowelevation ski resorts uneconomic therebyrestricting skiing to higher elevations. Theyconclude that a minimum elevation of 400 m is

4 Personal communication.

necessary for profitable winter tourism inAustria. A similar study in the northwest USAsuggests a 75 cm to 125 cm reduction inaverage snow depth and the movement of themean altitude ski lift (masl in meters) from 900masl to 1,250 masl [6].

Not only is the shift in masl a concern forindividual ski resorts but it has implication forthe wider ski industry. For example, skiing inmany parts of the world is currentlyconcentrated at low elevations with easyaccess to population centers. Winterrecreation research has shown that thesesmall, ‘community’, opportunistic ski areas areessential to the growth of the industry as theycater to novices, families, and skiers getting inshape before a longer vacation at a larger skiresort. For example, Scott, McBoyle and Mills[3] report that 45% of skiers in Toronto, thelargest single market of active skiers inCanada, travel less than one hour to ski. InArizona, Snowbowl operators estimate thatday visitors account for 65.5% of total visitswhilst destination (overnight) skiers make up36.4% of total visits [15, p3-87], however, afterthe expansion and with more consistentoperation they anticipate that the proportion ofdestination skiers will increase to 42% of thetotal [15, p3-92] . These examples confirmMcBoyle and Wall’s [9] insight that smallregional ski areas are important in developingski industry. Sunrise has a program that isdirectly growing the sport in Arizona: localunder-12 year olds receive free seasonpasses.

The timing of snowfall is crucially important tothe economics of the ski industry [1]. Forexample, in Austria adequate snow pack overthe Christmas and New Year holidays and inFebruary is essential for a good season. In theWhite Mountains and Snowbowl timing is alsocrucial5 and can sometimes be more importantthan snow abundance. For instance the2000/01 ski season was good from a revenueperspective at Sunrise, even though snowfallwas lower than average, because snow 5 The peak days in the Arizona ski season are

Christmas-New Year, Martin Luther King

weekend, Tucson’s Rodeo week, college and

school Spring breaks, every weekend during the

season, and Thanksgiving, given early snowfalls.

3

accumulated in October and was present overthe decisive Thanksgiving and Christmas-NewYear holiday season [17]. The timing ofsnowfall also in part determines thecompetitive positions of Snowbowl vis-à-visSunrise. For example in the 2004/05 seasonSnowbowl received early snow opening onNovember 26, whilst Sunrise was played catchup as it’s season got a slow start, only pickingup after large snowfalls early in 2005.

An important concept is “snow reliability” [18].In Switzerland the ‘reliability’ threshold isassumed to be 7 out of 10 good winterseasons, with a snow cover depth of 30-50cm, for a minimum 100 days betweenDecember 1 and April 15. Currently, just 85%of Swiss ski resorts meet this snow reliabilitytest. However, if, as per one climate changescenario, snow reliability were to rise to 1,500m in the period 2030-2050, only 63% of allresorts would meet this test. The authorsconclude that “climate change will lead to anew pattern of favoured and disadvantagedski tourism regions. If all other influencingfactors remain the same, ski tourism willconcentrate in the high-altitude areas that aresnow-reliable.” Such an outcome is not only aconcern for the resorts and surroundingcommunities in the lower elevation preAlps butalso for the ecologically sensitive Alps regionwhere pressures to expand skiing are likely toincrease.

The concept of snow reliability is a good one.However, we would add one additionalrequirement for our study areas: the number ofconsecutive poor snowfall seasons that can bewithstood. The reason for this addition is that astring of, for example three poor seasons inten, is likely to have a greater impact on thefinancial viability of Sunrise and Snowbowlthan three poor seasons spaced equally overa ten year period. In fact Snowbowlexperienced just such a scenario and wishesto avoid a repeat of such year-on-year lossesby investing in snowmaking. Before thebanner 2004/05 season, Snowbowl recordedfour unprofitable operating seasons (1996,1999, 2000 and 2002) in the last eleven. Thecombination of poor seasons and the capitalintensive nature of the industry meant that allnet cumulative profits over the period were

reinvested in ongoing maintenance and capitalimprovements [15]. Other factors are alsoimportant for reliability, such as thresholdmaximum temperatures and rainfall during theski season which may hasten early ski resortclosure [19]. Specific reliability or profitabilitythresholds are established by the mountainmangers for Sunrise and Snowbowl. However,if managers rely exclusively on naturalsnowfall some seasons will fall short of thesethresholds. Snowmaking would enable theresorts to more consistently meet minimumoperational days during the season: Snowbowlhas set a goal to consistently operate 125days per season plus or minus 15% [15, 3-112] whilst Sunrise usually operates 122 daysduring the season, but, in the relatively snowpoor 2003/04 season only operated 100 days.Consistent operation is important both forconsumers and suppliers of snow-basedrecreation. Many skiers would prefer to plantheir skiing activity and vacations withcertainty, certainty that may be absent ifresorts rely on natural snow, whilstconsistency allows managers to fully utilize lift,lodge, and snowmaking infrastructure andstaff throughout the season.

Climate change is a long-term threat to the skiindustry. Scott, McBoyle and Mills [3] arguethat climate change is a “catalyst that willreinforce and accelerate the pace of structuralchange” in the ski industry. They note that theski industry is increasingly two-tiered withlarge four-season high elevation resorts andsmall less profitable and less attractive lowelevation resorts. However, they also arguethat gradual warming will give the industrytime to adapt and those with fewer constraintswill, whilst those resorts that exit the industry,will do it in an actively planned manner. But,for smaller, lower elevation resorts, such asWilliams and Mt. Lemmon, Arizona a couple ofconsecutive poor seasons combined with landownership and water resource constraints mayusher more rapid restructuring which in turnwill impact local economies.

Arizona’s ski resortsArizona has four ski resorts, two of which arevery small and therefore not discussed further,the larger two are Snowbowl which is nearFlagstaff, and the largest resort, Sunrise Park

4

near Greer. Map 1 shows the locations of allfour ski resorts in the state. Table 1summarizes ski resort statistics from Arizona’sand nearby, competitor resorts. It also recordsmean values and the Snowbowl expansionplan data. In-state competitors are importantbecause none of Arizona’s four ski resorts areworld class; they mainly cater to in-stateresidents. Two-thirds of Snowbowl’s visitorsare day trippers whilst Sunrise estimates thateighty percent of its around 200,000 annualskier visits come from in-state, with theremainder drawn from New Mexico, SouthernCalifornia, and Mexico [20]. However, it isimportant to take a regional assessmentapproach [3] because a proportion of Arizonaskiers can, and do travel, to other nearbyresorts, particularly if snow conditions are poorin the state. Additionally, skiers from the other

Four Corner states travel to Arizona if the skiconditions are good relative to their own, forexample Sunrise was one of the fewsouthwestern resorts with good snow duringthe 1998/99 season and benefited from largenumbers of out-of-state visitors. Severalfactors will determine how any single ski resortfares with respect to climate variability andchange: the relative impact of climate changeon the resort and its competitors (a function ofelevation, aspect, humidity, snow patterns,etc) and any resultant changes in intra andinter-regional skiing market share, the costs ofadditional snowmaking, how adaptation byskiers could alter skiing demand, and theimpact of other adaptation strategies such asbusiness diversification and weatherderivatives and insurance [3].

Map 1: Arizona’s ski resorts

Key: 1 = Arizona Snowbowl Ski Area, 2 = Bill Williams Ski Area,3 = Sunrise Park Ski Resort, and 4 = Mt. Lemmon Ski Area.Source: http://southwest-vacation-travel.net/arizona/ski-area-map.htm

5

Table 1: Arizona, New Mexico, and southern Colorado Ski AreasARIZONA NEW MEXICO

SunrisePark

Snowbowl Snowbowlafterexpansion

WilliamsSki Area

MountLemmonSkiValley

AngelFireResort

PajaritoMountain

RedRiverSkiArea

SandiaPeakSkiArea

SipapuSkiArea

SkiApache

SkiSantaFeResort

TaosSkiValley

TrailTotal trails 65 32 >32 7 18 71 37 58 30 31 55 44 110 Runs (%) Beginner Intermediate Advanced

404020

374221

323216

305020

582022

314821

205030

323830

355510

205030

203545

204040

242551

Longest run(km)

4.43 3.22 ~4.13 1.21 2.59 5.15 1.93 4.02 4.02 2.14 3.70 4.83 9.25

Mountain Skiable area(ha) 324 55 83 12 28 180 113 117 81 28 304 267 524 Elevationpeak (m) 3,353 3,505 3,505 2,484 2,789 3,254 3,173 3,155 3,163 2,821 3,505 3,674 3,602 Elevationbase (m) 2,835 2,804 2,804 2,286 2,499 2,621 2,804 2,667 2,645 2,499 2,926 3,156 2,806 Annualsnowfall (cm) 635 635 635 381 508 533 381 554 318 330 470 572 775 Snowmaking % ofterrain

10 -- 100 -- -- 52 -- 87 15 33 45 100

Night skiing Yes No No No No Yes No No No No No No No

Lift Total lifts 10 5 9 2 3 6 6 7 7 4 11 5 12 Lift capacity(person/hr)

16,000 4,960 ~7,500 850 2,000 5,770 6,500 6,720 4,500 2,900 16,500 7,800 15,500

Other Snowboard Cross-country Tubing

Yes16 km

Yes

Yes40 km

Yes40km

Yes

YesNo

Yes

YesNo

Yes22 km

YesNo

YesNo

Yes

Yes47 km

No

Yes3 km

No

YesNoNo

YesNoNo

NoNo

Yes

Source: www.ski-guide.com

6

Table 1 continued: Arizona, New Mexico and southwestern Colorado Ski AreasCOLORADO

Durango

Mountain

Resort

Hesperus

Ski Area

Telluri

de Ski

Resort

Wolf

Creek

Ski

Area

Silverton

Mountain

Crested

Butte

Mountain

Resort

Monarch Ski

and

Snowboard

Area

Powderhorn

Resort

Aspen

Highlands

Aspen

Mountain

Ajax

Buttermilk Snowmass

TrailTotal trails 75 13 84 77 85 54 29 130 76 42 Runs (%) Beginner Intermediate Advanced Expert

235126

30203020

243838

20352520

5050

10258

57

213742

20501515

18301636

482626

353926

6501232

Longest run (km) 3.22 1.68 7.40 3.22 4.18 1.61 3.54 5.63 4.83 4.83 8.53

Mountain Skiable area (ha)

486 32 688 647 - 428 271 206 320 272 174 1,255 Elevation p(m) 3,299 2,707 3,737 3,628 4,054 3,707 3,646 3,002 3,559 3,417 3,018 3,813 Elevation b(m) 2,680 2,469 2,659 3,139 3,170 2,858 3,289 2,499 2,451 2,422 2,399 2,470 Annual snowf(cm) 660 381 785 1181 1016 757 889 635 762 762 508 762 Snow making %of terrain

21 -- 15 -- 57 15 14 31 26 17

Night skiing No Yes No No No No No No No No No No

Lift Total lifts 11 2 16 6 1 14 5 4 4 8 9 21 L i f t c(person/hr)

15,050 21,168

8,280 18,160 6,100 4,370 5,400 10,755 7,500 27,984

Other Snowboarding Cross-country Tubing

YesNo

Yes

YesNo

Yes

Yes30 km

Yes

Yes6 km

No

YesNoNo

Yes70 km

No

YesNoNo

YesNoNo

YesNoNo

Yes65 km

Yes

YesNoNo

Yes65 km

No

Source: www.ski-guide.com

7

Table 1 continued: Arizona, New Mexico and southwestern Colorado Ski AreasCOLORADO

SunlightMountain

Resort

SkiCooper

BeaverCreekResort

BreckinridgeSki Resort

KeystoneResort

ArapahoeBasin Ski

Area

CopperMountain

Resort

MEAN

TrailTotal trails 67 26 146 146 116 69 125 Runs (%) Beginner Intermediate Advanced Expert

2055205

304030

343927

153352

12295

54

15452020

21253618

Longest run (km) 4.02 2.26 4.43 5.63 4.83 2.41 4.51

Mountain Skiable area (ha) 190 162 658 894 753 198 985 341 Elevation peak (m) 3,016 3,566 3,487 3,962 3,719 3,978 3,753 3,379 Elevation base (m) 2,403 3,200 2,256 2,926 2,835 3,261 2,960 2,710 Annua l sn(cm) 635 635 787 775 584 932 711 675 Snow making %of terrain

12 -- 37 25 49 25 16 32

Night skiing No No No No Yes No No

Lift Total lifts 4 5 16 24 21 6 22 8 L i f t (person/hr)

4,600 3,300 30,739 36,680 35,175 8,700 30,63010,610

Other Snowboarding Cross-country Tubing

Yes29 km

No

Yes25 km

No

Yes32 km

No

Yes23 km

No

Yes18 km

No

YesNoNo

YesNo

YesSource: www.ski-guide.com

Table 1 shows that Sunrise and Snowbowl aresmall resorts compared to many in the region,are average elevation, but low compared tothe largest Colorado resorts, and have limitedsnowmaking capability compared to theaverage, and in particular compared to theworld class Colorado and New Mexico resorts.In summary Snowbowl and Sunrise haverelatively good conditions for ‘local’ ski resortsbut have lower snow reliability as a result oflower elevation and limited snowmakingcapabilities compared to the larger FourCorners resorts.

The White Mountain Apache Reservation islocated in central east Arizona. It covers6,480km2 with elevation ranging from 1,846 mto 3,538 m. The US Department ofCommerce’s Economic DevelopmentAdministration funded the development ofskiing in the White Mountains in the 1970s.Prior to the development of the ski industry theWhite Mountain Apache tribe relied onsummer-based businesses, such as forestrywork and wildfire defense, camping, hikingand hunting. One consequence of suchseasonality was that infrastructure andemployees were underutilized out-of-season[17]. However, these investments have alsomade the tribe and surrounding communitiesmore dependent on snow-based recreationand therefore more vulnerable to winter-basedclimate variability and change. Meanwhile,Snowbowl is the state’s northern most skiresort. It has a good location lying 25 km northof the third largest city in the state, Flagstaff,and two hours drive from the largest city in thestate, Phoenix and four hours drive from thesecond largest city in the state, Tucson. Incontrast Sunrise is four and a half hours drivefrom both Phoenix and Tucson and issurrounded by small towns with smallpopulations. However, Snowbowl is a muchsmaller resort than Sunrise and has somecompetitive disadvantages vis-à-vis Sunrise.

There are a number of competitive differencesbetween the Sunrise and Snowbowl that offersome opportunities for Sunrise. However, it isimportant to note that management at Sunrisebelieves that a good season for Snowbowl is agood season for Sunrise because it increasesinterest in the sport and because manyArizona skiers ski at both resorts for variety ofexperience. Sunrise is located on the WhiteMountain Apache reservation whereas theother three resorts in the state are constrainedby their location on USFS land; any plans toexpand facilities must pass an environmentalimpact statement (EIS). Sunrise is the largestresort in Arizona and the only resort to offernight skiing, limited snowmaking, and acasino. It is also the only resort to offer on-sitelodging at the 100 room Sunrise Lodge. It alsohas a nearby RV park. The other resorts arerestricted by US Forest Service regulationsthat prohibit such on-site lodging.

Skier visitation data for Sunrise is notavailable, but, we have data on skier visitationfor the state for the period 1995/96 through2004/05 and for Snowbowl from 1981/82through 2004/05. Using both sets of data wecan calculate skier visitation at Sunrise andthe two smaller resorts in the state, Williamsand Mt. Lemmon. Skier visitation at thesesmaller resorts is minimal and therefore theremainder accounts well for Sunrise visitation.We can see from Table 2 that Snowbowlexperiences large fluctuations in visitornumbers whilst visitor numbers are steadier atSunrise. This is in part due to limitedsnowmaking capabi l i ty at Sunrise.Nevertheless, it is interesting to note thatalthough Sunrise is almost six times largerthan Snowbowl that in good snow years skiervisits are the same at both sites. This reflectsthe superior location of Snowbowl which isnearer large population centers.

Table 2: Arizona ski visitation and Snowbowl dataSnowbowl Arizona Sunrise (Mt Lemmon,

Wiliams)Season Snowfall,

cmDays open Skier visitation

1981-1982 673.1 123 63,0001982-1983 701.04 135 99,6261983-1984 193.04 64 28,9131984-1985 675.64 118 114,7071985-1986 533.4 124 105,2521986-1987 736.6 112 125,0261987-1988 462.28 92 119,2591988-1989 431.8 79 120,1321989-1990 609.6 74 99,2801990-1991 591.82 112 106,0001991-1992 914.4 134 173,0001992-1993 1168.4 130 181,0001993-1994 558.8 114 116,3881994-1995 657.86 122 176,7781995-1996 287.02 25 20,312 102,575 82,2631996-1997 685.8 109 153,176 365,787 212,6111997-1998 838.2 115 173,962 384,665 204,5831998-1999 381 60 35,205 246,941 211,7361999-2000 457.2 45 66,152 243,685 177,5332000-2001 690.88 138 162,175 355,780 193,6052001-2002 220.98 4 2,875 214,135 211,2782002-2003 523.24 96 87,354 277,361 190,0072003-2004 368.3 120 72,000 238,420 164,4202004-2005* 1168.4 133 190,000 370,000 180,000Source: [15, Arizona data from Kottke Survey]*Snow and visitation data based on news reports. Season length was estimatedusing season start and end dates (Nov 26-April 10) minus an estimated 3days for blizzard condition closings.

Arizona’s winter precipitation variabilityArizona’s climate is subject to several climateoscillations that singularly and in combinationproduce high inter-annual and decadal winterprecipitation variability that in turn impactssnow season consistency. On an annual timescale the Pacific/North America (PNA)teleconnection and the North AtlanticOscillation (NAO) contribute to high inter-annual precipitation variability. The PNAteleconnection relates to the position of theridges and troughs in the polar jet stream inthe upper troposphere. The PNAteleconnection describes shifts in the jet

stream path in terms of positive and negativevalues. A positive PNA index describes the‘usual’ position of the polar jet stream, with apronounced ridge over the western US and astrong trough over the eastern US. During thistypical pattern the southwest receives verylittle winter precipitation and this phase hasbeen linked with winter droughts for exampleduring 1976-1977. The strength of the positivePNA phase is also affected by the NAO. Anegative NAO phase reinforces positive PNAconditions, bringing dry winters to thesouthwest. During a PNA negative phaseArizona receives above average winter

precipitation. The important climatic oscillationthat influences winter precipitation in Arizonaon a decadal time scale is the El NiñoSouthern Oscillation (ENSO). Warm (positive)phase, El Niño conditions bring above normalwinter precipitation in Arizona, for example theabundant snowfall 2004/05 season, whilst cold(negative) phase, La Niña conditions bringdrier conditions. This phase is more reliablydry in Arizona than the warm phase is wet.

Longer term climate cycles, such as thePacific Decadal Oscillation (PDO), also have adistinct impact on winter precipitationvariability in Arizona. The PDO, like ENSO,has two phases, negative and positive, butunlike ENSO events which are typically limitedto a year or two, the duration of the PDO cycleis around 30 years. The positive phase PDO issimilar to the warm phase (El Niño) of theENSO, whilst the negative phase PDO hassimilar conditions to the cool phase (La Niña)of the ENSO cycle [21]. Negative PDOconditions are associated with regionaldrought in the Southwest. Significantly,research has found that PDO interacts withthe shorter-term ENSO cycle, enhancing orweakening La Niña and El Niño climateconditions [22]. For example, a positive PDOenhances typical El Niño conditions andweakens La Niña conditions whilst a negativePDO enhances typical La Niña conditions andweakens El Niño conditions. Theseinteractions may be crucial to understandingwinter precipitation variability in Arizona;concurrent positive PDO and warm phaseENSO cycles result in above normal winterprecipitation and positive PDO cyclesmodulate the typical dry winter effect of thecold phase ENSO cycle.

The climate may be entering a new cold phaseof the PDO cycle. Shifts in the past haveoccurred about every twenty to thirty years, forexample there was a shift from a cold PDO toa warm PDO in around 1925, then to a coldPDO in 1947, and then to a warm PDO in1977. The importance of such a shift is thatthe negative PDO weakens El Niño conditions,conditions that bring above normal winterprecipitation to Arizona, and strengthens LaNiña conditions, conditions that bring lowerthan normal winter precipitation. Furthermore,

such a shift is also significant because longerterm climate interactions between the positivephase of the Atlantic Multidecadal Oscillation(AMO) and the negative phase of the PDOhave in the past signaled mega droughts inthe southwest, such as in the 1950s [23].Al though droughts remain largelyunpredictable, there is concern that the currentdrought in the southwest could persist due toNorth Atlantic warming and concurrent PDOcooling. This same combination could alsosignal poor snowfall conditions in thesouthwest.

Using data from Sunrise we tested whetherthe climate signals of ENSO, and the PDOmodulation of ENSO on winter snowfall aredetectable. The data used in the statisticalanalysis6 was obtained from a snow coursedata station at Mt. Baldy.7 This is the closestmeasuring station to Sunrise with consistentdata. The snow course data recorded snowdepth and SWE8 six times per season in theperiod 1983-2004.

Model 1 used the 132 snow courseobservations and tested for the ENSO signal:

Snow depth = b0 intercept + b1ENSO + e [1]

The results of the model are: Adjusted R-squared = 0.22,b0 = 50.88, t = 19.94 and pr>|t| = <.0001, and

b1= 14.30, t = 6.12 and pr>|t| = <.0001

The results show that ENSO is a positive andsignificant variable in explaining the variationin snow depth at the Mt. Baldy site.Specifically, a one unit increase in the ENSO

6 All the modeling was done in SAS. We used an

ordinary least squares regression method and

simple linear functions.7 The Baldy Mountain SnoTel site is located at

latitude 33.98, longitude 109.50 and elevation

2,782 m. This compares to Sunrise Mountain

latitude 33.84, longitude 109.91 and elevation

2,836 m to 3,354 m.8 The amount of water in snow varies and is

measured by the snow water equivalent (SWE).

For example, a foot of fresh, heavy, wet snow may

produce 38 mm of water whereas light, powdery

snow may be equivalent to just 13 mm of water.

index raises snow depth by 14.3 cm. That is atthis site there is a positive correlation betweenEl Niño and snow pack that we expected apriori. For a visual representation of thisrelationship Chart 1 plots snowfall data against

ENSO data. The two lines move in concertthat is La Niña years are correlated with lowersnowfall and El Niño years with aboveaverage snowfall.

Chart 1: Mt. Baldy snowfall vs. ENSO

-2.00

-1.50

-1.00

-0.50

0.00

0.50

1.00

1.50

2.00

2.50

1982

-198

3

1983

-198

4

1984

-198

5

1985

-198

6

1986

-198

7

1987

-198

8

1988

-198

9

1989

-199

0

1990

-199

1

1991

-199

2

1992

-199

3

1993

-199

4

1994

-199

5

1995

-199

6

1996

-199

7

1997

-199

8

1998

-199

9

1999

-200

0

2000

-200

1

2001

-200

2

2002

-200

3

2003

-200

4

Season

EN

SO

0

50

100

150

200

250

300

350

400

Maxim

um

snow

depth

, cm

ENSO Index

Snow

Next we test whether a PDO-ENSO signalcould be detected in the dataset. A binaryvariable was created that equaled one if PDOand ENSO were simultaneously positive, andzero otherwise.

Snow depth (cm) = b 0 intercept + b 1PDO-

ENSOeffect + e [2]

The results of the model are:

adjusted R-squared = 0.12,b0 = 43.19, t = 12.06 and pr>|t| = <.0001

b1= 23.03, t = 4.26 and pr>|t| = <.0001

Model 2 shows that the PDO-ENSO effect is apositive and significant determinant of snowdepth variation at the Mt. Baldy site. Theresults show that the combination of a positivePDO and ENSO increases snow depth by 23

cm compared to the case when this interactionis absent. This data seems to confirm that thePDO-ENSO effect is indeed an importantpredictor of snowfall at our site. Note also thatthe PDO-ENSO effect is larger than the ENSOeffect alone.

The modeling was repeated using the snowcourse data and SWE as the dependentvariable. Note that 92% of the variation inSWE in our dataset is determined by snowdepth alone. These results reported thatENSO is a significant and positive determinantof SWE at the thousandth percent level. A oneunit increase in the ENSO index raises SWEby 4.4 cm (t=6.19, pr>|t|=<.0001). In additionthe PDO-ENSO binary variable was a positiveand significant variable at the thousandth percent level. The combination of a positive PDOand positive ENSO raised predicted SWE by7.83 cm (t=4.84, pr>|t|=<.0001).

Snowfall variability and skier visitationThe next models ran tested whether climatevariability is a significant determinant of annualsnowfall, skier visitation and season length atSnowbowl (for data see Table 2). We testedmodels with ENSO and the PDO modulated

ENSO, separately, the effects were notsignificant when modeled in combination. Thefirst model tested:

SNOWFALL, cm = b0 intercept + b1ENSO or

ENSO-PDO + e, n=22 [3a/b]

Table 3: Snowbowl snowfall and climate oscillationsParameter estimate t-value pr>|t| Adjusted R-squared

INTERCEPT 580.74 11.57 <.0001 0.15ENSO 114.03 2.23 0.0372

orINTERCEPT 493.54 7.78 <.0001 0.21ENSO-PDO 249.15 2.59 0.0171

The results show that ENSO and themodulated ENSO are positive and significantdeterminants of annual snowfall at Snowbowlwith a five per cent confidence level. Forexample, in a ENSO-PDO season snowfall is249 cm greater than when this combination isabsent. The ENSO results show that a 1 unitincrease in ENSO raises annual snowfall by114 cm. These results tell us that mountain

managers should take full advantage of suchnatural snowfall years perhaps with targetedENSO marketing. The next model tested theeffect of climate on skier visitation.

SKIER VISITS = b0 intercept + b1 ENSO or

ENSO-PDO + e, n=22 [4a/b]

Table 4: Snowbowl visitation and climate oscillationsParameter estimate t-value pr>|t| Adjusted R-squared

INTERCEPT 105,609 9.94 <.0001 0.15ENSO 23,829 2.20 0.0395

orINTERCEPT 91,612 6.48 <.0001 0.12ENSO-PDO 42,349 1.98 0.0615

Table 4 results show that ENSO and themodulated ENSO are positive and significantdeterminants of skier visitation at Snowbowl.For example for each one unit increase inENSO (warming) annual visitation rises byalmost 24,000, that is El Niño years aresignificantly better than La Niña seasons. The

last model tests whether these climateoscillations help us to explain variation inseason length at Snowbowl.

DAYS_OPEN = b0 intercept + b1 ENSO or

ENSO-PDO + e, n=22 [5a/b]

Table 5: Snowbowl season length and climate oscillationParameter estimate t-value pr>|t| Adjusted R-squared

INTERCEPT 92.82 13.09 <.0001 0.19ENSO 17.85 2.47 0.0223

orINTERCEPT 82.62 8.99 <.0001 0.14ENSO-PDO 31.08 2.15 0.0430

The results from this model show us thatseason length at Snowbowl is positivelyaffected by El Niño conditions and the ENSO-

PDO effect. For example, in an El Niño-PDOyear the season is 31 days longer than withoutthis effect. All these models confirm that

climate variability is indeed an importantdeterminant of ski season outcomes atSnowbowl.

The timing of snowfall and the length of the skiseason are crucial to the financial success ofArizona’s ski resorts. Using Snowbowl datafrom the 1981/82 through 2004/05 seasons[15 and news reports for the most recent

season] we modeled season days open as afunction of total seasonal snowfall. The meanseason length during the 23 seasons modeledwas 96 days (minimum 4 days and maximum138 days). The model tested was:

DAYS_OPEN = b 0 intercept +

b1SNOWFALL(cm) + e, n=23 [6]

Table 6: Modeling Snowbowl season length, Adjusted R-squared = 0.58Parameter estimate t-value pr>|t| Mean

INTERCEPT 29.02 2.23 0.0370SNOWFALL 0.11151 5.58 <.0001 601.87

The model results show that total seasonalsnowfall (cm) is a positive and significantdeterminant of the variation in season length.Readers should remember that Snowbowldoes not have snowmaking capability andtherefore this model models the importance ofnatural snowfall. For each 9 cm increase insnowfall during the season the season lengthincreases by 1 day. Total snowfall is animportant determinant of season length butthere are other factors that were not modeledthat are also important (the model onlyexplained 58 per cent of the variation in

season length), for example the number ofdays that meet the resorts minimum snowdepth.

Using this same data we also modeled skivisitation as a function of the season lengthand total snowfall. Mean visitation in the 23seasons was 110,025 (minimum 2,875 andmaximum 190,000). The model tested was:

Visits = b0 intercept + b1SNOWFALL(cm) + b2

DAYS_OPEN + e, n=23 [7]

Table 7: Modeling Snowbowl visitation, Adjusted R-squared = 0.824Parameter estimate t-value pr>|t| Mean

INTERCEPT -19,942 -1.46 0.1608SNOWFALL 109.92 3.67 0.0015 601.87

DAYS_OPEN 668.17 3.24 0.0041 96

The model results show that total seasonalsnowfall (cm) and the number of days openexplains 82 percent of the variation inseasonal visitors. The results show that foreach additional cm of snow visitationincreases by 110 visitors and each additionalday open increases visitation by 668, both aresignificant at the one per cent level. Otherfactors that may be important which were notmodeled are the timing of the snowfall inrelation to peak visitation periods, number ofdays with new powder, the day of the weekChristmas falls, and macro factors such as thestate of the economy. We could imagine thatan additional opening day that fell on theThanksgiving weekend would raise visitationby more than the 668 skiers modeled:

Snowbowl often hits peak capacity of 3,400skiers per day on such vacation weekends.However, we do not have data on seasonopening dates and there is no SNOTEL site atSnowbowl to assess whether snow conditionswould support early opening. The nearestSNOTEL sites are at much lower elevationand therefore not good proxies for this site.Snow course data for Snowbowl for the period1996-2005 does exist but the data collectionperiod is December 30 through March 31 thatis it does not cover the Thanksgiving period.

Models 4 through 7 formalize what mountainmanagers already know: poor snowfall and ashort season result in poor financial seasons.For example, low snowfall in the 1998/99 and

1999/00 seasons heralded poor revenues atSunrise, whereas the 1997/98 and 2004/05seasons had abundant snow and were verygood revenue years. Meanwhile, at Snowbowlthe 2004/05 season was snow abundant andalmost record breaking whereas 2001/02 wasa poor season for snowfall and ski visitation.Without snowmaking, the profitability andviability of either resort may be threatened bya series of winter drought seasons.

SNOWMAKINGSnowmaking reduces a ski operations’vulnerability to poor natural snowfall enablingresorts to operate for a longer season than ifthey relied on natural snowfall alone. The firststudy to model snowmaking as a climateadaptation strategy was Scott, McBoyle andMills [3]. Their study in central Ontario,Canada found that climate change scenariosthat double CO2 concentrations would reduceski seasons by a smaller 7% to 32%compared with a 40% to 100% loss predictedby McBoyle and Wall [24]. The differencebetween these studies is a measure of theeffectiveness of snowmaking in extendingseasons. Furthermore, improved snowmakingcapabilities9 would reduce season losses tojust 1% to 21% [3]. Nevertheless, these resultsrely on large increases in snowmaking by theyear 2080 of between 191% and 380%. Suchlevels of snowmaking may not be financially orenvironmentally viable.

Snowmaking is not viable for thesnowmobiling, Nordic and cross country skiingsectors because of the huge economic andenvironmental barriers to snowmaking for tensor hundreds of kilometers of trails [3]. Sunriseand Snowbowl both have vulnerable crosscountry skiing sectors. However, withoutdetailed information on the size of this specificmarket relative to the downhill skiing market,the impact of this vulnerability is unknown.Both resorts have however identifiedsnowmaking as an adaptation strategy toclimate variability for their downhill skibusiness. 9 Newer snowmaking equipment can efficiently

make 15 cm of snow base per day at -2ºC

compared to older equipment that makes just 10

cm snow base per day and requires lower

temperatures of at most -5ºC.

Snowmaking can extend seasons byfacilitating an earlier start and later finish thannatural conditions would allow and by buildingup snowpack after warm or rain conditions. Inthese ways snowmaking enables mountainmanagers to achieve more consistency intra-and inter-ski seasons. The manager can thendecide on the optimal season length based oncosts, including projected snowmaking costs,demand, and profit margins. An example inthe southwest is Beaver Mountain, Utah.Management here requires season length tobe 100-105 days to realize typical industryprofit margin of 6.5%-7% [6]. The assessmentnotes that a mere 2°C increase in temperaturecould reduce this resort’s revenue by 20% andconcludes that “location can be crucial tosuccess. Resorts at higher elevations mightnot be negatively affected by projectedchanges in climate, and some resorts couldeven benefit from wet snows that help build abetter snow base.” Sunrise and Snowbowl aresuch higher elevation ski resorts, elevationthat may buffer some of the negative effects ofclimate change.

Arizona’s ski resorts are small relative to thelarge, commercial resorts in nearby Coloradoand Utah. Sunrise currently has snowmakingcapabilities for just 10% of skiable terrain. Toput this in context, climate change vulnerablelow elevation-high latitude ski resorts in centralOntario, Canada have snowmakingcapabilities for 100% of skiable terrain [3]whilst snowmaking capabilities of 50% andhigher are the norm for the top Four Corner skiresorts. Snowbowl currently has nosnowmaking capability. However, its operatingcompany has recently had its ambitiousexpansion plans approved; these plansincorporate snowmaking for 100% of itsexpanded terrain [15]. Sunrise began theprocess to expand its snowmaking capabilitiesin 2000 but the effort faded, now in 2005 it isagain seeking funding to increasesnowmaking capacity to improve seasonconsistency. Water rights are a major restraintto further expansion of snowmaking beyondthat approved at Snowbowl, but this constraintis not binding on the White Mountain ApacheTribe which likely has sufficient water

supplies.10 Sunrise would however seekEconomic Development Administrationassistance to help fund the upgrade.

For illustrative purposes we graphed actualsnow depth and minimum snow depththresholds to visualize when and how muchsnowmaking (shaded pink) might be requiredin a good and bad year. The graphs belowshow the timing and volumes of snowmakingrequired to meet Sunrise’s 36 cm minimumsnow depth over a season starting onNovember 20th and ending April 10th duringan El Niño episode (1982/83) and a La Niñaepisode (1998/99). There are very limitedhistoric climate records for Sunrise andtherefore for this exercise we used daily datafrom the Mt. Baldy SNOTEL site. This sitedoes not record snow depth and therefore weestimated snow depth at Sunrise usingcorrelations between SWE data recorded atMt. Baldy and Sunrise snow depth at the basefor the 2004/05 season, for which we havesome data. The graphs are thereforeillustrative of snowmaking requirements ratherthan exact measurements of theserequirements.

Chart 2 shows that during this El Niño episodesnowmaking requirements would have beenlimited to early in this arbitrarily chosenseason whereas in this La Niña episode largesnowmaking requirements would have beennecessary at the start and near the end of theseason to maintain adequate snow depths forskiing. During a severe winter drought it maybe unprofitable to make snow particularly ifwinter temperatures are warm causing snowlosses. However, to fully address this issuethe actual costs of increased snowmakingshould be weighed against the expectedrevenue gains from increased visitation.

10 The actual water rights of the tribe are as yet

undefined. However, the White Mountain Apache

Tribe has recently entered into adjudication of its

water rights. This process once complete will fully

define the tribes water rights.

Chart 2: Snowmaking requirements in an El Niño season, 1982/83

0

50

100

150

200

250

300

350

400

11/2

0/19

82

11/2

7/19

82

12/4

/198

2

12/1

1/19

82

12/1

8/19

82

12/2

5/19

82

1/1/

1983

1/8/

1983

1/15

/198

3

1/22

/198

3

1/29

/198

3

2/5/

1983

2/12

/198

3

2/19

/198

3

2/26

/198

3

3/5/

1983

3/12

/198

3

3/19

/198

3

3/26

/198

3

4/2/

1983

4/9/

1983

Date

Sno

w d

epth

(cm

)

Chart 3: Snowmaking requirements in a La Niña season, 1998/99

0

50

100

150

200

250

300

350

400

11/2

0/19

98

11/2

7/19

98

12/4

/199

8

12/1

1/19

98

12/1

8/19

98

12/2

5/19

98

1/1/

1999

1/8/

1999

1/15

/199

9

1/22

/199

9

1/29

/199

9

2/5/

1999

2/12

/199

9

2/19

/199

9

2/26

/199

9

3/5/

1999

3/12

/199

9

3/19

/199

9

3/26

/199

9

4/2/

1999

4/9/

1999

Date

Sno

w d

epth

(cm

)

Scott, McBoyle and Mills [3] resultsdemonstrated the “importance of snowmakingas a climate adaptation and that the value ofinvestments in snowmaking systems will onlyincrease under climate change”. However,they also note that ski resorts will haveincreased snowmaking requirements at

warmer temperatures, which will increaseenergy requirements and costs. The authorsdid not address the costs and investments inequipment necessary for such extensivesnowmaking and therefore it is unclearwhether such increases in snowmaking wouldbe financially viable. Investment decisions are

complex: each resort needs to weigh the costsagainst the benefits given the likely responsesof consumers, and the snow conditions atnearby competing resorts that might besuperior as a result of aspect, altitude, orsome other reason.

The politics and economics of snowmakingThirteen tribes are opposed to using reclaimedwater for snowmaking on the sacred SanFrancisco Peaks where Snowbowl is locatedand environmentalists are opposed to theproposed use of scarce water resources.However, these objections were overruled inthe recent EIS approval of Snowbowl’sexpansion and 83 ha (205 acres) ofsnowmaking opt ion. Paradoxica l ly ,snowmaking investments at Snowbowl mayprevent desecration of other sacred sites inthe state, sites that might offer superior snowconditions than at Snowbowl and Sunrise. Theproposed Snowbowl snowmaking operationrequires 5.7 mn liters of water per day for 119days between November 1 and February 28each ski season. Snowbowl has no waterrights and currently trucks in all potable water.Therefore to support snowmaking Snowbowlhas signed an agreement with the City ofFlagstaff that would allow for a maximumtransfer of 673.8 mn liters of Grade Areclaimed wastewater per season for 5 yearswith possible renewal three more five yearperiods. The plan cites three scenarios withassociated snowmaking water requirements: adry year for the maximum 486 af (673.8 mnliters), an average year of 364 af and a wetyear 243 af. To put these numbers in contextan acre foot11 of water supports on average afamily of four for a year, thus, Flagstaff isgiving up water that could support nearly 486families, or other alternative economic growthactivities.

Adaptation strategies such as snowmakingcan reduce climate change vulnerability byincreasing snow pack, durability, and seasonreliability. However, snowmaking is expensive;costs, excluding capital costs, range between$500 and $4,000 per acre foot (af) of snowdependent on system efficiency [4],accounting for approximately 15%-25% of total

11 An acre foot = 325,851 gallons.

operating costs. Total costs at Snowmass,Colorado, including capital costs are as highas $5,000 af [25]. Furthermore, it may becomemore difficult to finance such investments.Bürki, Elasser and Abegg [18] found evidencethat Swiss banks are becoming more wary offunding infrastructure investments at skiresorts lower than 1,500 m whilst ski areamanagers in southern Ontario, Canada nowhave to assess and address climate changeimpacts in financing negotiations with lenders[3]. Crucially for the southwest snowmakingalso requires large volumes of water12 [5]. Toillustrate, a typical sized ski resort producesaround 500-1,000 af of snow per season at acost between $0.25mn-$4mn and usesbetween 214 af and 427 af of water perseason.

Snowbowl’s plans anticipate applying a baseof 64 cm over the 83 ha terrain at the seasonstart during a wet season to ensure goodskiing conditions over the Thanksgiving break,one and half times in an average season andtwo times in a dry season, equivalent to 427af, 640 af and 854 af of snow per season,respectively. They estimate total waterdemand of 243 af in a dry season, 364 af in anaverage season and 486 af in a wet season.Included in this total is approximately 6 afannually for toilet flushing.13 Calculating wateruse per acre foot of snow from these totalwater requirement estimates gives usapproximately 185,401 gallons of water peracre foot of snow which is one third higherthan an industry estimate (see footnote 9).This difference might reflect a preference forsnow with a higher snow water equivalent(wetter snow) or an estimate for losses in thelow humidity environment in Arizona. If weassume a 2% inflation rate then the costs,excluding capital costs, of this snowmaking inten years time when the investments come on-line range from $598 af to $4,780 per af ofsnow. Total snowmaking costs would rangefrom $255,346 to $2.041 mn in a wet year to

12 1 acre foot of snow = 139,222 gallons of water.13 Snowbowl currently trucks in 5.7 mn liters ofpotable water annually of which 60% is used for

toilets. If we assume growth in flushing use

equivalent to expected growth in visitors from an

annual average 98,000 to 215,000 we arrive at 6 af

for flushing annually.

$510,692 to $4.082 mn in a dry year. To putthese costs in perspective Snowbowl’s annualgross revenues averaged $4.525 mn in theeleven seasons from 1993/94 to 2003/04. Thesnowmaking proposal assumes that visits willincrease to 215,000 skiers a year from theaverage 98,000 now, at current ticket prices14

and assuming a 2% inflation rate, revenues in2013/14 would be $12.1 mn (this does notinclude revenues from snowtubing visitors).Therefore snowmaking costs could account forbetween 2.1% to 16.9% of total skier revenuesin a wet year and between 4.2% to 33.7% in adry year. If we assume an industry profit of10% snowmaking costs rise to between 2.3%to 18.7% in a wet year to between 4.7% and37.5% in a dry year, this compares to theindustry norm of between 15%-25% of totaloperating costs. However, all these costs arejust illustrative, at temperatures higher than -2°C snowmaking is not technically viable, andeven at temperatures below this thresholdsnowmaking may not be financially viable, asin the above example during a dry year withhigher assumed snowmaking costs.

Snowmaking expansion plans for Sunrise aremore modest. Management has two plans;both are restricted to increasing snow makingcapability for Sunrise Peak only not the twoother peaks in the resort. These plans are notpublic and therefore approximations are used;the first more comprehensive plan anticipatesincreasing snow making from the current 32.4ha to around 55 ha to a depth of 61 cm. Thisoption would mean 100% snowmakingcapability for Sunrise Peak. A second scaleddown option reduces the depth of thesnowpack to 41 cm and the number of trailswith snowmaking, reducing the total area toaround 45 ha. The timing of snowmaking inboth cases is expected to be in the preseason,November through Christmas. The initialcapital investments are estimated between $4mn and $9 mn. Operating costs will depend onthe efficiency of the system, the number ofsnowmaking appl icat ions, and thetemperatures at which snow is made. To

14 In the 2004/05 season adult full-day lift ticket

prices were $42, juniors $24 and seniors $22,

respectively.

illustrate these on-going costs we assume oneearly season base application that wouldenable skiing over Thanksgiving. If we assumea five year time horizon for the snowmaking tocome on-line and a 2% inflation rate thenoperating costs might range from $552 to$4,416 per af of snow. Therefore operatingcosts for the full scale plan would rangebetween $166,000 to $1.3 mn and for thereduced plan between $80,000 to $640,000.Doubling the applications would double theseestimated costs. Sunrise current revenues arebetween $6 mn and $8 mn a year with a profitmargin of around 10 per cent. This financialpicture falls short of financing the total capitalcosts but it would seem that the ski resort cancover the on-going costs of snow makingparticularly if a longer, more consistent seasonincreases skier visits to the resort, specificallyat peak times such as Thanksgiving. Sunrisecan accommodate 8,000 visitors per day, if weassume an average ticket price of $30/day15, apeak day during the season could bring inrevenues of $240,000.

Although Scott, McBoyle and Mills [3] make aclear argument for the expansion ofsnowmaking when profitable, they do note thatthe success of such supply-oriented changesis dependent on the responses of consumers.A question that both Snowbowl and Sunriseneed to answer before investing further in theirresorts is does the demand support it? Thedemand for skiing is not only influenced bysnow pack, but also relative snow conditions,prices at competing resorts, and theavailability of other recreation activities. TheSnowbowl EIS attempts to answer thisquestion by calculating the utilization rate atthe resort. This is found by dividing skier visitsby total capacity. This rate averaged 64% inthe period 1990-2004 peaking at 83% in 1998.They comment that US ski areas measurestrong demand as a utilization rate above40%: on this basis they conclude that there issufficient demand in the region to supportexpanded ski facilities. However, research hasshown that the demand for spring skiingtypically “wanes before the snow pack isexhausted” [3]. A similar phenomenon has

15 In the 2004/05 season adult full-day lift ticket

prices were $41, juniors $24 and seniors $20.

been recorded at Sunrise and Snowbowl. Forexample, in the good snowfall 2004/05 seasonSunrise closed the season on April 3, 2005and Snowbowl on April 10, 2005, though bothstill had sufficient snow to remain open.Demand for skiing drops off in springregardless of the snow situation and mountainmanagers often decide to close resorts asthese last weeks are often unprofitable.Additionally, it can be difficult for resorts tokeep staff so late in the season, as many areseasonal workers and move to theirspring/summer employment. Marketing toskiers in the shoulder seasons is one way thatSunrise and Snowbowl can capitalize on theirsnowmaking investments and good latenatural snowfall. In fact Sunrise resort doestarget this season by offering reduced lift ticketprices, $25 for adults and $15 for juniors, inthe last weeks of the season. It also modifiesits supply-side by closing Apache and Cyclonepeaks; only Sunrise peak remains open.

These same authors note that there is littlecurrent understanding “how recreational usersand tourists respond to climate variability(whether or not to participate or purchaseequipment, activity substitution, use patterns,destination choice).” Some researchers haveattempted to answer the question as to howdifferent types of skiers might respond toclimate change. Skier survey results fromSwitzerland found that during a poor season49% of skiers would switch to a more snow-reliable resort, but more worrying for theindustry, 32% of respondents said they wouldski less often [18]. Only 4% of the surveyrespondents said that they would abandonskiing. The broader issue is the differentialimpact of such demand changes. It is likelythat smaller ski resorts would suffer most as

young skiers, day skiers, and novice skiers,their predominant clientele, are the most likelyto respond negatively to poor seasons. In factKönig’s [18] study of skiers in Australiasupports this conclusion. She found that halfof all advanced skiers would travel overseasfor quality snow conditions, whilst only 18% ofnovice skiers would, and 16% would give upskiing altogether. Therefore, a largeuncertainty in any modeling of the impact ofclimate change on winter recreation is theresponse of local skiers to changes in skiseason length, timing, and relative quality.Scott, McBoyle and Mills [3] argue that onescenario that would leave resorts as well off, isif skiers adjust their recreation behavior byskiing more frequently in the shorter season. Asensitivity analysis could address thisuncertainty in consumer response andestimate future ski recreation demand.

Climate change and snowmakingIn order to investigate the impact of humanfossil fuel use on the climate system climateresearchers have developed GeneralCirculation Models (GCM). These modelsreplicate the complexity of climate systemsand can be used to model ‘what if’ scenarios,such as greenhouse gas forcing, or climatechange. All these models suffer from coarseresolution and as such cannot capture theinfluence of varied topography in thesouthwest nor ENSO and monsoon effects[26]. To overcome some of these limitationsUSGS climatologists have developed aregional simulation model, called RegCMwhich is nested in a GCM. The results fromthis regional model, which simulates adoubling of CO2 atmospheric concentrations,are shown in the table below.

Table 8: Southwest climate change: RegCM simulation_ Temperature, °C _ Precipitation, cm

Season RegCM RegCMWinter +4.2 -3.05Spring +4.2 0 to +1.52Summer +5 -1.52Fall +4.2 -1.52 to +1.52 Source: [27]

The RegCM model predicts a decline in winterprecipitation in concert with higher winter

temperatures [28]. Reduced snowpack andsnow season length and higher spring

temperatures will in turn bring forwardspringtime runoff, a situation likely to reducewater supplies in the snowpack-water supplydependent southwest. The simulations seemto show that climate change will herald moredrought-like conditions in the southwest.However, there is a big caveat to thismodeling; although the RegCM model is animprovement over GCMs it is still relativelycoarse resolution, with grids 15km2 (Sunriseresort is smaller than this grid at 3.24 km2).Significantly, it does not account well forelevation. In the paper the researchers usedRegCM to recreate current temperature andprecipitation maps in the southwest to validatethe model. These simulations are notparticularly good at reproducing actual climateconditions in either the White Mountains

region or the San Francisco Peaks; thereforewe have less confidence about the magnitudeof the changes reported in Table 8. However,this does not mean that warming has notalready happened or that it will not happen inthe future. Arizona mountain managers mustprepare for warmer winter temperatures,wetter snow, and perhaps less overall or moreerratic snow, and rainfall instead of snowfallon warmer winter days.

In the absence of a RegCM designed toforecast changes in snow depth and seasonlength we instead investigate temperaturedata from Snowbowl [15, p3-227]. In Table 9we arbitrarily choose a 1°C warming for eachmonth during the ski season.

Table 9: Snowbowl winter temperatures, °C, and a climate change scenario1960-1990 data 1°C winter warming

Month Average Maximum Minimum Average Maximum MinimumNovember -7.2 3.6 -14.4 -6.2 4.6 -13.4December -10.4 -0.4 -16.9 -9.4 0.6 -15.9January -10.2 -0.1 -17.1 -9.2 0.9 -16.1February -9.1 1.2 -16.1 -8.1 2.2 -15.1March -6.5 4.0 -14.4 -5.5 5.0 -13.4April -2.8 7.8 -11.6 -1.8 8.8 -10.6

The table shows that minimum temperatureseven with a 1°C winter warming would still becold enough to make snow even with lessefficient snowmaking systems that requiretemperatures of at least -5°C compared tonewer systems that operate at -2°C. However,these averages hide those days when warmsystems might move through the ski resortscausing snow melt. Chart 4 shows how snowdepth modeled for Mt. Baldy16 varies with a 3-day average of the maximum temperature.The chart shows that there is somerelationship between snow depth andmaximum temperature: new snowfall isassociated with a dip in the 3-day averagewhilst snow melt occurs after temperaturesconsistently exceed 8°C. Using thisobservation and the data in Table 9 snow maymelt earlier in the season and higher Apriltemperatures could shorten the season. Itshould be noted that the Mt. Baldy SNOTEL

16 Snow depth is approximated by using the

following formula, snow depth = SWE*7.7.

site is at a lower elevation than the base ofboth Sunrise and Snowbowl. Temperatures atboth ski bases would thus be cooler as per thedry adiabatic lapse rate. Using this rate thebase at Sunrise and at Snowbowl are 0.54°Cand 0.23°C cooler, respectively. We areinterested in the base elevations because it ishere where snow tends to accumulate last andmelt first. It is also an area of heavy use.

Chart 4: Mt. Baldy snow depth and 3-day average of maximum temperatures, °C

0

10

20

30

40

50

60

70

80

11/2

3/19

98

12/3

/199

8

12/1

3/19

98

12/2

3/19

98

1/2/

1999

1/12

/199

9

1/22

/199

9

2/1/

1999

2/11

/199

9

2/21

/199

9

3/3/

1999

3/13

/199

9

3/23

/199

9

4/2/

1999

Date

Snow

depth

, cm

-5

0

5

10

15

20

25

3-d

ay a

vera

ge m

axim

um

tem

p,

C

The final impact simulated by climate changemodels is reduced winter precipitation. If suchprojections are realized Sunrise andSnowbowl may experience more winterdrought-like conditions, such as currently in aLa Niña year. However, other models forecastwetter snow that may actually help the skiareas as such snow makes a good base. Inthe absence of a model for our study areasour best approximation is that the current mixof wet, average, and dry years is likely tobecome more erratic and this will in turn makemanaging the resorts for profit morechallenging, particularly, if managers relyexclusively on natural snowfall.

Note that if we had a climate change scenariofor the Snowbowl region we could estimate theimpact of reduced snowfall on the number ofdays open using Model 6 and then using theresults from Model 7 we could estimate theimpact on skier visitation. For example, ifsnowpack were forecast to decline by 45 cm(pre-snowmaking investments) Model 6estimates that the season would be cut shortby 4 days. In turn reducing season length by 4days would reduce skier visits by an estimated2,672. If these skiers spend $100/dayeconomic output would decline by $267,200.

Local economic impacts

We saw that the economics of snowmakingsupport snowmaking at both Sunrise andSnowbowl, particularly, if the most efficientsystems are installed. However, theinfrastructure costs of snowmaking are highand include the building of large waterreservoirs, pipelines, and purchasing snowguns, etc, and in the case of Snowbowlbuilding a 24 km pipeline to conveywastewater to the resort. Total costs of theexpansion and snowmaking are estimated at$19.77 mn; the snowmaking component isestimated to cost $8.2 mn. Sunrise estimatesthat its more modest snowmaking plans willcost between $4 mn and $9 mn to construct.These costs need to be weighed against thebenefits arising from more consistent skiseasons. The benefits of snowmaking at eithersite are not confined to increased profits at theski resort but include the local economicimpacts accruing to surrounding areas as aresult of skier spending. These impacts havebeen modeled for both resorts. Ski resorts dobenefit local communities and in turn thesesame communities might support snowmakinginvestments directly or efforts to raiseinvestment funds from the EconomicDevelopment Administration.

Snowbowl est imates that in theseasons1996/97 to 2002/03 an average 22.1persons were employed by the ski resort on a

full-time equivalent (FTE) basis, 272.4 personsyear-round on a full-time seasonal basis and204.3 persons on a part-time seasonal basis.These different types of workers add up to 172FTE jobs. The EIS report quotes averagedirect expenditures of $9.79 mn per seasonduring this same period in Coconino County.This total was calculated by multiplying theaverage ski visitation of 97,900 by expenditureof $100 per skier. This $100 figure is based onsurveys in Colorado, Utah and at Snowbowl. Itmay be somewhat high considering that two-thirds of all Snowbowl’s visitors are day-trippers and probably many are renting skiequipment and buying gas in Phoenix. Thereport goes on to estimate that total direct andindirect economic impacts in Coconino Countyof 232 FTE jobs and $12.08 mn in economicoutput (in $2003, inflation adjusted). From thisdata we can calculate that the multiplier usedis 1.24. The expansion of Snowbowl will inturn create an estimated 232 FTE constructionjobs and $21.24 mn economic output inCoconino County. At the end of the 10 yearplanning period when Snowbowl anticipates itwill attract 215,000 visitors a year, the skiresort will directly support 332 FTE jobs and atotal 564 FTE jobs in Coconino County.Economic output attributable to Snowbowl isforecast at $23.7 mn. Snowbowl alsocontributes to the US Forest Service throughservice fees and its community throughproperty and sales taxes. These fees andtaxes are forecast to rise from their baselineaverages of $90,000, $36,000 and $257,000to $193,000, $455,833 and $669,000,respectively by the 2013/14 season. The EIScalculates that winter tourism accounts for8.6% of Flagstaff area’s overall economy [15,3-188].

In a previous study the revenues from skiing atSunrise were estimated at between $5.1 mn to$5.95 mn [17]. Updated figures for the verygood 2004/05 season are ski revenues of $8.4mn.17 It should be noted that these ski revenuedata include ski ticket prices only. That is otherspending by skiers on accommodation, food,ski equipment hire or purchase, and gas arenot included in this figure. In the SnowbowlEIS [15] they estimate total expenditures of

17 Personal communication.

$100 per skier, if we use this figure, then directexpenditures by visitors at Sunrise and insurrounding areas in the 2004/05 season wasaround $20 mn. This figure does not includeindirect expenditures resulting from increasedincome in the region. If we use the samemultiplier as in the Snowbowl study we gettotal economic output of $24.7 mn. Note thatthe economic multiplier is likely to be higherfor Sunrise than Snowbowl because of thelarger proportion of overnight skiers.

Sunrise is one of the largest non-governmental employers in the region:employment peaks in winter months at around518 employees, but, the FTE payroll is a muchlower 198, as many jobs are seasonal. Gibsonand Evans use public information on taxrevenue and survey data to determine arecreation dependency ratio for the entireWhite Mountains region.18 A recreationdependency ratio is the percentage of totalincome in the region generated by therecreation sector. The White Mountains areais not as ‘developed’ as Snowbowl and thusSunrise is proportionately more important tothe local community. In the White Mountainsthis ratio is 44%. However, this overall statistichides variation between communities:Sunrise’s dependency is 90% and nearbyGreer’s is 79%. They also calculate a specificwinter dependency ratio. The winterdependency ratio is the percent of annualsales occurring in the winter season, definedin their study as January to March. Thisdefinition underestimates winter dependency,as in a good year a large proportion of skivisits occur during Thanksgiving and over theChristmas-New Year period. Although winterrecreation is important to some communities inthe region, overall this season generates just27% of annual sales compared to the summerseason’s 33%, Sunrise is highly dependent onwinter recreation, with a winter dependencyratio of 60%. Note that the White Mountainswinter recreation dependency ratio is morethan three times higher than at Snowbowl.

18 This includes the off-reservation towns of Show

Low, Pinetop-Lakeside, Greer, and Springerville-

Eager and Sunrise and Whiteriver on the

reservation.

These data show that although the ski industryin Arizona is small it is economically importantto the White Mountain Apaches and also tothe surrounding communities in the WhiteMountains and in the Flagstaff area.Importantly the ski resorts bring winter tourismand revenues that balance out peak summer-time visitation patterns.

DISCUSSIONClimate change and the increased probabilityof poor snowfall is likely to acceleraterestructuring in the ski industry, favoringresorts at higher elevation and latitudes, andalso more geographically diversifiedcompanies, and those able to affordsnowmaking investments. Snowmaking notonly allows resorts to adapt to shorter snowseasons predicted by climate changemodeling but also provide more consistentconditions to even out the impact of moregeneral climate variability. The economicsseem to support the financial viability of suchinvestments at Sunrise and Snowbowl,particularly when we account for localeconomic impacts. We found that Sunrise andits surrounding areas are more than threetimes more dependent than Snowbowl onwinter-recreation. Nevertheless, the threat ofclimate change a century from now shouldencourage policymakers to develop acomprehensive regional development plan toease future structural adjustment resultingfrom a climate that may be warmer and drier.In turn such planning may encouragemitigation efforts. Scott, McBoyle and Mills’s[3] analysis of four different climate changescenarios lead them to argue that the skiindustry should pursue climate mitigation inorder to avoid the worst case scenario. Theypraise the National Ski Areas Association’sSustainable Slopes charter and the NationalSki Areas Association’s ‘Keep Winter Cool’campaign. The industry could also be apolitical force for a more comprehensivenational policy to reduce greenhouse gasemissions.

We have narrowly focused on the economicsof snowmaking at Sunrise and Snowbowl,however, the overall impact of changingclimate on these two resorts should beassessed in a larger framework that takes

account of climate impacts elsewhere in theregion. One possible outcome is even if skiseasons are shorter at Sunrise/Arizona, ifsnowfall declines are worse at Snowbowl/NewMexico, then Sunrise/Arizona might have afuture, as skiers switch to Sunrise/Arizona.Such a substitution effect would be the bestcase scenario; although individual skierswould be displaced from their usual recreationarea. However, another likely scenario is thatskiing declines overall in the state and region.

The negative effects of climate variability andchange on the ski industries at Sunrise andSnowbowl are still unknown. However, longer-term climate change may boost non-winterrecreation and revenues. Both ski resorts havestrong non-winter recreation programs.Snowbowl anticipates that its improvedfacilities will attract 30,000 summer visitors ayear. Whilst, Sunrise generates around $1 mnin non-winter revenues from Sunrise LakeMarina, the hotel, general store, and non-skierusers of chair lifts [17]. In Canada researchersinvestigated the impact of climate change onthe length and quality of summer tourism inCanada’s western mountain parks [29]. Theyfound that tourist numbers in Calgary, Albertawould improve in every month of spring andsummer under climate change scenarios.Another study investigated the changingseasonality for park visitation in the RockyMountain National Park, Colorado [30]. Itpredicted that warmer spring to fall weatherwould result in 193,000 to 333,540 additionalvisitors. In turn increased visitation wouldboost local economic output by between 6%and 10% and increase local employment bybetween 7% and 13%. Data from the WhiteMountains [31] records that summerrecreation is the biggest contributor to theregional economy, followed second by winterrecreation. A change in climate could furthershift recreation visitation to the summermonths, as valley residents seek respite fromeven higher temperatures, and also to theshoulder seasons, spring and fall, whichcurrently account for just 20% each of annualrevenue. However, Scott and McBoyle [32]caution that a change in recreation seasonalitycould have other implications, for examplecurrent tourism infrastructure might beinadequate to meet higher demand or

addit ional visi tat ion might increaseenvironmental stress.

The White Mountain Apache Tribe Wildlife andOutdoor Recreation Division has oversight oftwo alternative (non-winter) income generatingactivities: outdoor recreation permits and theTrophy Hunting Program. This latter includesthe world famous elk hunting but also thelesser-known but important pronghornantelope, bighorn sheep, bear, mountain lionand turkey hunting seasons [33]. In 1999,these combined activities generated $600,000profit. Such activities could help the tribe takeadvantage of climate change whilst reducingtheir economic vulnerability to climate changeimpacts on winter recreation.

To cope with climate variability and change wehave seen the importance of snowmaking asan adaptation strategy. Another strategy thatcould be pursued is to extend runs into higheraltitude. In Switzerland the ski industry hasused climate change as an argument toextend existing runs and open new ski runs athigh Alpine regions above 3,000 m againstenvironmental concerns [18]. This seems anunlikely option at either Sunrise or Snowbowlbecause there are few alternatives that wouldbe at higher elevation. However, if climatechange does increase temperatures to athreshold at which it is uneconomic to makesnow, we could anticipate that mountainmanagers might close the bottom sections oftheir runs and limit skiing to the higherelevation, more snow reliable terrain.

Another adaptation is to diversify risk. Largecorporate ski companies such as Vail Resortsand the American Skiing Company may beless vulnerable to climate change than singleresort operators because they are diversifiedcompanies, with real estate and warm-weathertourism businesses, and they are alsogeographically diversified, thereby reducingexposure to poor snowfall in one area [29].Both Sunrise and Snowbowl have no realestate ventures, their location on a reservationand USFS land preclude such options.However, Sunrise does have the option tofurther develop on-site accommodation, forexample, a new Ski in – Ski out lodge at theresort. The only onsite accommodation,

Sunrise Lodge, is located 11 km from theslopes. Large companies are also bettercapitalized and therefore able to makeinvestments in snowmaking. Financingsnowmaking and other expansions is anobstacle for both Sunrise and Snowbowl.However, both have the support of theirsurrounding communities to further developskiing in the region. Smaller operators maywish to investigate winter tourism weatherderivatives market to even out good and badseasons. A weather derivative19 is a contractbetween two parties that stipulates whatpayment will occur as a result of themeteorological conditions that occur in thecontract period. For example, the contractscould be structured to reduce weather-relatedrisk for peak periods over Christmas and NewYear.

Whilst the exact impacts of climate change arestill unknown there are other actions thatSunrise and Snowbowl management couldtake to ensure that every good snow day atthe resorts is fully utilized by skiers and profitsare maximized. Management could planactivities, marketing, and hiring based onENSO forecasts [34]: resorts could bettercapitalize on strong El Niño conditions, whilst,budgeting for increased snowmaking in LaNiña years. Sunrise’s webpage is in need ofan update; a new ‘public face’ could bettercapitalize on its position as the largest skiresort in the state, its excellent spring skiingconditions, and affordable learn to ski/boardpackages. Both resorts could review their skilift prices to ensure that they are not a limiting

19 A derivative is an instrument used by companies

to hedge against the risk of weather-related losses.

The investor who sells a weather derivative accepts

the risk by charging the buyer a premium. If

nothing happens, then the investor makes a profit.However, if the weather turns bad, then the

company claims the money. This is not the same as

insurance, which is for low-probability events like

hurricanes and tornados. In contrast, derivatives

cover high-probability events like a dryer-than-

e x p e c t e d s u m m e r .

http://www.investopedia.com/terms/w/weatherderi

vative.asp

factor to the growth of skiing. Themanagement may wish to introduce lower halfday skiing passes to encourage opportunisticskiers and off-peak prices for mid-week, preand late season skiing when the resorts areunderutilized. Family packages could also beintroduced to make the sport more affordable,this might be particularly important ifsnowmaking investments are realized at theresorts, because with more consistent skiseasons, families might substitute their usualvacation activities for a skiing vacation.

References

[1] Breiling, M. and P. Charamza (1999). Theimpact of global warming on winter tourismand skiing: a regionalised model for Austriansnow conditions, Regional EnvironmentalChange, 1 (1): 4-14.

[2] König, U. (1998). Tourism in a warmerworld: implications of climate change due toenhanced greenhouse effect for the skiindustry in the Austra l ian Alps.Wirtschaftsgeographie und Raumplanung,Volume 28, University of Zurich, Zurich,Switzerland.

[3] Scott, D., G. McBoyle and B. Mills (2003).Climate change and the skiing industry insouthern Ontario (Canada): exploring theimportance of snowmaking as a technicaladaptation. Climate Research, 23: 171-181.

[4] Northern Watertek Corp, Newletter –SkiFax #1,http://www.watertek.org/Skifax1.htm.Accessed 1.5.05.

[5] http://www.ratnik.com/snowmaking.html,accessed 1.5.05.

[6] US National Assessment Team (2000).CLIMATE CHANGE IMPACTS ON THEUNITED STATES: THE POTENTIALCONSEQUENCES OF CLIMATE CHANGEVARIABILITY AND CHANGE. US GlobalChange Research Program. CambridgeUniversity Press, New York, USA.

[7] Galloway, R. W. (1988). The potentialimpact of climate changes on Australian ski

fields. In Pearman G. I. (Ed) GREENHOUSE:PLANNING FOR CLIMATE CHANGE. CSIRO,Melbourne, Australia, 428-437.

[8] Breiling, M., P. Charamza, and O. Skage(1997). Klimasensibilitat OsterreichischerBezirke mit besonderer Berucksichtigung desWintertourismus. Report 97:1. Institute forLandscape Planning, Alnarp, Austria.

[9] McBoyle, G. R., and G. Wall (1992). Greatlakes skiing and climate change. In: Gill, A.and R. Hartmann (eds) MOUNTAIN RESORTDEELOPMENT. Center for Tourism Policy andResearch, Simon Fraser University, Burnaby,Canada, 77-81.

[10] Lamonthe and Periard (1998).Implications of climate change for downhillskiing in Quebec. Climate Change Digest 88-03, Environment Canada, Ottawa, Canada.

[11] Harrison, S. J., S. J. Winterbottom, and C.Sheppard (1999). The potential effects ofclimate change on the Scottish tourist industry.Tourism Management 20, 203-211.

[12] König, U. and B. Abegg (1997). Impactsof climate change on tourism in the SwissAlps. Journal of Sustainable Tourism, 5(1): 46-58.

[13] Elasser, H., and R. Bürki (2002) Climatechange as a threat to tourism in the Alps.Climate Research, 20, 253-257.

[14] Bloomfield, J. and S.P. Hamburg. 1997.Seasons of Change: Global Warming andN e w England’s White Mountains.Environmental Defense Fund, New York, NY.33 pp.

[15] USDA, 2005 Final Environmental ImpactStatement for Arizona Snowbowl FacilitiesImprovements Volume 1, Coconino NationalForest, Coconino County, Arizona.

[16] Mote, P. W., A. F. Hamlet, M. P. Clarkand D. P. Lettenmaier. 2005. Decliningmountain snowpack in western north America.Bulletin of the American MeteorologicalSociety, Vol. 86 (1).

[17] Gibson, L. J. and B. Evans (2002)Regional dependence on tourism: thesignificance of seasonality, Yearbook of theAssociation of Pacific Coast Geographers, 64:112-127.

[18] Bürki, R., H. Elsaser and B. Abegg(2003). Climate change – impacts on thetourism industry in mountain areas. 1st

International Conference on Climate Changeand Tourism, Djerba, Tunisia, 9-11 April,2003.

[19] Scott, D., G. McBoyle, B. Mills and A.Minogue (2004). Climate change and the skiindustry in North America: A Reassessment inMatzarakis, A., de Freitas, C.R., Scott, D.(eds.) 2004: Advances in TourismClimatology. Ber. Meteor. Inst. Univ. FreiburgNr. 12.

[20] Andrew Slough, Arizona’s Native Deep.http://www.skimag.com/skimag/features/article/0,12795,325810,00.html

[21] Mantua, N.J. (2000). PDO Index.http://jisao.washington.edu/pdo/. Accessed 10March 2005.

[22] Gershunov, A. and T.P. Barnett (1998).Interdecadal modulation of ENSOteleconnections. Bulletin of the AmericanMeteorological Society, 79(12): 2715-2725.

[23] Hidalgo, H. G. (2004), Climate precursorsof multidecadal drought variability in thewestern United States, Water Resources Res.,in press.

[24] McBoyle, G. R., and G. Wall (1992). Greatlakes skiing and climate change. In: Gill, A.and R. Hartmann (eds) MOUNTAIN RESORTDEVELOPMENT. Center for Tourism Policyand Research, Simon Fraser University,Burnaby, Canada, 77-81.

[25] The art of making snow, by CatherineLutz. Accessed Jan 5, 2005.http://www.snowmassvillagesun.com/home.php?content=article&article=1453

[26] Merideth, R., 2001. A Primer on ClimateVariability and Change in the Southwest, UdallCenter for Studies in Public Policy and theInstitute for the Study of Planet Earth, TheUniversity of Arizona, March 2001.

[27] Doherty, R. and L. O. Mearns (1999) Acomparison of simulation s of current climatefrom two coupled atmosphere-ocean GCMsagainst observations and evaluation of theirfuture climates. Boulder, CO: National Centerfro Atmospheric Research website,www.esig.ucar.edu/esig/doherty/index.html

[28] Thompson, R. S., S. W. Hostetler, P. J.Bartlein, and K. H. Anderson, 1998. A strategyfor assessing potential future change inclimate, hydrology, and vegetation in theWestern United States. USGS Circular 1153.

[29] Scott, D. (2003). Climate change andtourism in the mountain regions of NorthAmerica. 1st International Conference onClimate Change and Tourism, Djerba, Tunisia,9-11 April, 2003.

[30] US EPA (2003) Research andDevelopment, Nat ional Center forE n v i r o n m e n t a l R e s e a r c h .http://cfpub.epa.gov/ncer_abstracts/index.cfm.

[31] Gibson, L. J. and B. Evans (2001) WhiteMountains Visitor Tax Study, The University ofArizona.

[32] Scott, D., and G. McBoyle (2001) Using a‘tourism climate index’ to examine theimplications of climate change for climate as anatural resource for tourism. Proceedings ofInt’l Society of Biometeorology, Commission 5.October 5-10 2001. Halkidi, Greece. 69-98.

[33] KSG (2000) Wildlife and OutdoorRecreation Program: White Mountain ApacheTribe (Whiteriver, AZ). Harvard University.

[34] Weiher, R. (2000) Editor IMPROVING ELNIÑO FORECASTING: THE POTENTIALECONOMIC BENEFITS, NOAA.