Dust storms are a catastrophic weather phe-nomenon mainly occurring in arid and semi-arid re-gions,in which a lot of dust is advected by strongwinds and horizontal visibility is less than 1 km(Cen-tral Meteorological Bureau,1979). They are generallyassociated with small probability and great threat, and induced by specific large-scale circulation patterns and the development of certain weather systems. Researchon weather systems that cause dust storms mainlyconcentrates on three aspects: weather dynamic pro-cesses,spatiotemporal distributions of dust storms and associated cause-effects, and variations of related me-teorological elements. Joseph et al.(1980)reportedvisibility to decrease sharply,wind velocity to increaserapidly, and relative humidity to increase greatly dur-ing dust storms in India. Taha et al.(1991)pointedout that the average wind velocity decreased by 0-2 m s¡1 in an oasis compared with an open field.Goudie and Middleton(1992)noted the lack of aglobal model describing the occurrence and frequencyof dust storms. Natural processes are the main de-terminant in the frequency of dust storms. A winderosion index was proposed to forecast dust stormsin order to decrease wind erosion by modified windregimes and soil moisture conditions(McTainsh et al., 1998). Desouza et al.(2011)researched evolutioncharacteristics of a dust storm day in Oman: highwind velocity was closely correlated with a fast movingcold front driving suspended s and within a zero vege-tation zone. Muhammad et al.(2012)proposed thata hybrid approach consisting of satellite imagery and wireless sensor networks is most suitable for detect-ing and predicting s and and dust storms of all types.Dust storms in Middle East and Southwest Asia canbe grouped into two main categories: the Shamal duststorms and frontal dust storms(Hamidi et al., 2013).

Research on dust storm weather in China beganin the 1970s. Zhou et al.(2002) and Qian et al.(2003)concluded that the Hotan area in the south-ern margin of the Tarim basin is one of the high fre-quency centers for strong or exceptionally strong duststorms in northern China. Wang et al.(2003)pointedout that there are few dust storms in northern Xin-jiang and many dust storms in southern Xinjiang, and that dust storms occurred frequently from the 1960sto 1970s,before decreasing obviously after the 1990s;they also noted that dust storms mainly appear fromApril to August. Liu et al.(2004)researched the frac-tal characteristics of strong dust storm weather sys-tems. Bai et al.(2005)pointed out that there are twoconcentrated areas of dust storm events in the Tarimbasin: the eastern Tarim basin,centered in the Kash-gar and Artush area; and the southern Tarim basin,centered in Cele and Hotan. Meteorological factorschange sharply before and after the passing of duststorms: air pressure increases,air temperature de-creases, and wind velocity increases drastically(Wang et al., 2000). Lu and Liu(2006)found a close correla-tion between increasing vegetation cover and decreas-ing dust storm frequency. The vegetation coverageis considered an important factor for dust storm fre-quency changes in northern China. Fan et al.(2010),Zhang et al.(2010), and Hai(2011)found that the oc-currence of dust storms sharply increases the relativehumidity,air pressure, and wind velocity. However,Niu et al.(2007)concluded that air temperature isrelatively high near the ground before dust storms, and decreases sharply over a short period after duststorms. There is a process of decreasing air tempera-ture and increasing relative humidity in a dust stormevent(Li et al., 2009). The photosynthetically activeradiation(PAR) and total solar radiation during duststorm events are obviously less than on cloudy and sunny days(Yao et al., 2006; Feng et al., 2009).

Nonetheless,the above-mentioned studies mainlyfocused on the spatiotemporal characteristics of duststorms and associated regional background circula-tions; few studies have examined variations of the me-teorological factors such as wind,air temperature,rel-ative humidity, and solar radiation before,during, and after dust storm events,especially across different l and surface types in an oasis-desert ecotone. This paperaims to investigate the spatiotemporal characteristicsof meteorological factors over different l and scape typesbefore,during, and after typical dust storms in theoasis-desert ecotone in Cele,Xinjiang,China, and at-tempts to provide a research basis for the preventionof dust storms in similar environments worldwide.2. Brief introduction to the experimental area

Cele County lies between the southern margin ofTaklimakan desert and the northern Kunlun Moun-tain(35°l705500-39°3000000N,80°0302400-82°1003400E)(Fig. 1). It is a typical continental warm temperatezone. The area is hot and has little precipitation insummer,with long periods of sunshine and enoughlight and heat resources. The temperature differencebetween day and night is large and the highest and lowest recorded temperatures are 41.9 and -23.9℃,respectively. The annual average precipitation is 35.1mm, and the annual potential evaporation is 2600 mm.

Fig. 1. Distribution of(a)weather stations and (b)topography along the transect of the study area.

Figure options

It is known that wind-blown s and disasters arefrequently influenced by two main prevailing wind di-rections,namely,northwest and northeast. On av-erage,s and -blowing weather occurs on 25.2 days peryear in Cele,with a maximum of 59 days. Galesexceeding level 8 occur from 3 to 9 times annually(Zhang,1995). Disasters due to blown s and limit so-cial and economic development in Cele County. Theannual prevailing wind directions are westerly and west-northwesterly,which together occur for 62.43%-76.25% of the time, and the subordinate wind direc-tion is northwesterly,which occurs 17.75% of the time(Yang,1990). In recent years,the annual averagenumber of days with dust storms has been 16.5 inCele. Dust storms mainly occur in spring and sum-mer(Liu et al., 2009). The oasis-desert ecotone ismostly distributed in the western and northern partsof Cele County,while to the west of Old Cele Riverthere is shifting s and y l and , and to the south is Tak-limakan desert. The annual average wind velocity at2.0-m height is 2.36,1.75,1.60, and 0.96 m s¡1 atthe sites of shifting s and y l and ,semi-fixed s and y l and ,fixed s and y l and , and oasis,respectively. The averagesoil water content at the soil surface(0-20 cm)in theoasis-desert ecotone is very low(generally < 1%).

Lin(1988) and Zu et al.(2005)identified Takli-makan desert as being located between 36° and 42°N; and from the viewpoint of the planetary wind system,the upper air circulation over the desert is mainly in-fluenced by midlatitude westerlies. The Tarim basin issurrounded by mountains on three sides,with the Ti-betan Plateau asserting a dominant impact on the cir-culation situation,which influences the developmentof the zonal circulation. During winter,easterly and westerly airflows are prevalent to the east and to thewest of Keriya River in Taklimakan desert,respec-tively(Fig. 1).3. Experimental design for field observation and research methodology

Four weather stations were installed on differ-ent l and scape types(i.e.,shifting-s and frontier,semi-fixed s and y l and ,fixed s and y l and , and inner oasis)along the prevailing wind direction. Weather sta-tions comprised of HOBO U30 meteorological survey-ing instruments(USA ONSET Company). The siteof weather station No. 1 was representative of a sur-face of shifting s and y l and ,on the surface of which thevegetation mainly comprised sparse and tall Tamarixramosissima shrubs,a small coverage of Hexinia poly-dichotoma, and Halogeton. The average height ofTamarix nebkhas was from 2.0 to 3.5 m. At the siteof observation station No. 2,the surface vegetationmainly comprised Tamarix shrubs,Alhagi sparsifolia, and Karelinia caspica communities, and the terrainwas undulating,which represented semi-fixed s and yl and . The site of observation station No. 3 repre-sented fixed-s and y l and ; it was located at the marginof an artificial oasis, and the surface vegetation mainlycomprised a comparatively flourishing A. sparsifoliacommunity with relatively flat terrain. The site ofweather observation station No. 4 was representativeof the oasis,in which cotton and pomegranate wereplanted with vegetation coverage(VC)of more than80%. Here,the l and was flat and s and y.

From site Nos. 1-3,there was a natural vege-tation ecotone covering a distance of 4.3 km. Thethree weather stations were equipped with wind cupsat heights of 0.5,1.0,1.5,2.0, and 3.0 m(includinga wind vane)before July 2010. Subsequently,eachmeteorological station was equipped with wind cupsat heights of 0.5,1.0,2.0,4.0,8.0, and 10.0 m(in-cluding a wind vane). The observation parameters in-cluded wind velocity,wind direction,air temperature,relative humility(RH),PAR, and solar radiation. Airtemperature and RH probes were installed at heightsof 0.5 and 2.0 m, and the probes for PAR and solarradiation were installed at a height of 1.5 m. Windvelocity resolution was 0.19 m s¡1, and the precisionsof air temperature and RH were 0.001℃ and 0.1%,re-spectively. The wind direction resolution was 1.4°. Ameasurement baseline from site Nos. 1-3 in summer2010 was designed to observe VC(by using quadratsof size 100 m × 100 m) and topography. The VC in-creased gradually from site No. 1(3%)to site No. 3(67%)(the VC at site No. 2 was 27%).

The study area was located at the southern mar-gin of Taklimakan desert, and the instantaneous im-pact threshold wind velocity was 5 m s¡1 on shiftings and y l and (Chen et al., 1995; Yang et al., 2012). Thesurface roughness and threshold wind velocity were ex-pected to increase with increasing VC from the shiftings and y l and to the inner oasis. Temporal and spatialchanges of meteorological factors three days before,during, and after dust storms were analyzed. Thedata collection interval was 1 min for three observationperiods(13-15 May 2010,11-13 March 2011, and 31May-2 June 2012),except for a 5-min interval from 18to 20 March 2012,i.e.,four typical dust storms wereselected. S and blew before and after the dust stormevent on 14 May 2010,but the weather was sunny and breezy before the 12 March 2011 dust storm,withs and blowing and floating dust weather after the duststorm event. There was floating dust before the duststorm on 19 March 2011, and rain and cloud after-ward. The weather was overcast both before and afterthe dust storm on 1 June 2012. All observation timesare Beijing Time(BT).4. Spatiotemporal characteristics of meteoro-logical factors during typical dust storms4.1 Meteorological factors during the 13-15May 2010 dust storm event

From nighttime of 13 May to daytime of 14 May,there occurred intermittent blowing s and over theshifting s and y l and and there was some wind-blowns and on the surfaces of semi-fixed and fixed s and yl and . Average wind velocity decreased substantiallywith increasing VC. Compared with that on the shift-ing s and y l and of site No. 1,the average wind velocityat heights of 0.5 and 2.0 m was reduced by 17.63% and 29.35% at site No. 2, and 58.83% and 54.7% at site No.3,respectively. The wind-blown s and with the high-est intensity and frequency occurred at the shiftings and y l and site from 0930 BT 14 May,when the duststorm began to blow up(Figs. 2a and 2b). Duringthe dust storm,the average wind velocity at heightsof 0.5 and 2.0 m was reduced by 24.05% and 29.09% atsite No. 2, and 64.52% and 39.84% at site No. 3,re-spectively,compared with that for the shifting s and yl and of site No. 1. The s and blowing continued for217 min at site No. 1. Wind power began to waneafter 1930 BT 14 May until 0030 BT, and then it be-gan to strengthen again and blowing s and appeared.Compared with that of site No. 1,the average windvelocity at heights of 0.5 and 2.0 m was reduced by25.89% and 29.22% at site No. 2, and 66.92% and 42.87% at site No. 3 on 15 May,respectively.

Fig. 2. Changes in meteorological factors over the three l and scape types(Nos. 1-3)of the ecotone between 13 and 15May 2010:(a)wind velocity at 0.5 m,(b)wind velocity at 2.0 m,(c)air temperature at 2.0 m,(d)RH at 2.0 m,(e)PAR, and (f)solar radiation.

Figure options

Before the occurrence of the dust storm on 13May,air temperature decreased and RH increased. Airtemperature decreased with intermittent rainfall from0000 BT to dawn on 13 May, and RH began to in-crease significantly until the morning of 13 May. Inthe daytime between 13 and 14 May,the order of RHvalues by site was No. 1 > No. 2 > No. 3, and showedan opposite trend at nighttime. The differences in airtemperature and RH were larger in the daytime thanat nighttime,regardless of the l and scape type(Figs.2c and 2d). During the dust storm between 1400 and 1900 BT 14 May,RH declined,while PAR,air temper-ature, and solar radiation first significantly increased, and then weakened(Figs. 2e and 2f). This is as-cribed to the decreased wind power after 1250 BT and the dramatic increase in dust concentration,which re-duced the solar radiation absorption and increased itsscattering,diffusion, and fluctuation. Later on,withthe increase of wind velocity,temperature and solarradiation increased gradually once again, and the RHdecreased gradually. Before and after the dust storm,air temperature during the daytime ordered by sitewas No. 3 > No. 2 > No. 1. This was due to the nat-ural vegetation,which increasingly acted to preventwind and block s and from the shifting s and y l and tothe oasis along the prevailing wind direction. Withthe decreased dust concentration toward the inner oa-sis,the air temperature,PAR, and solar radiation in-creased. The prevailing wind direction over the threel and scape types was all in a W and WNW direction;however,the wind direction was more variable closerto the oasis and more wind was in an E and S direc-tion(Fig. 3). This could be ascribed to the influenceof nebkhas.

Fig. 3. Spatial and temporal differences of wind direction for three landscape types (Nos. 1-3) between 13 and 15 May 2010.

Figure options

Air temperature first increased and then de-creased over the three l and scape types before and afterthe dust storm,while RH decreased throughout theevent. Temperature and RH at site Nos. 1-3 on 13May were 20.471,20.575, and 20.472℃, and 38.79%,38.15%, and 38.75%,respectively. Air temperature and RH at site Nos. 1-3 during the dust storm on 14May were 24.144,24.569, and 24.937℃, and 24.54%,23.93%, and 23.41%,respectively. Average air temper-ature gradually increased and RH gradually decreasedfrom the shifting s and y l and to the oasis; however,air temperature and RH differences between the threel and scape types were very small. Air temperature and RH at site Nos. 1-3 during blowing-s and weather on15 May were 21.221,21.154, and 21.082℃, and 17.9%,18.24%, and 18.67%,respectively. Average air tem-perature decreased and RH increased gradually fromthe shifting s and y l and to the fixed s and y l and , and the variation reduced. VC had a slight influence onair temperature and RH during the dust storm,whilethe distribution of dust concentration played a keyrole in changing air temperature,RH, and solar radi-ation. There was a consistent decreasing trend in thechanges of PAR and solar radiation from the shiftings and y l and to the oasis(Figs. 2e and 2f). The maxi-mum PAR and solar radiation occurred on the semi-fixed s and y l and before the dust storm on 14 May,followed by the shifting s and y l and and fixed s and yl and . During the blowing-s and day(15 May),PAR and solar radiation ordered by l and scape type was:shifting s and y l and > semi-fixed s and y l and > fixeds and y l and . This indicates that higher terrain and nebkhas on semi-fixed s and y l and is able to intercept,to an extent,blowing s and and floating dust.4.2 Meteorological analysis during the 11-13March 2011 dust storm event

The duration of blowing s and was 400 min onshifting s and y l and for the dust storm on 12 March2011. The maximum wind velocity was 11.69 m s¡1at 10.0-m height. Average wind velocity and s and blowing time decreased significantly from the shift-ing s and y l and to the semi-fixed s and y l and ,fixeds and y l and , and oasis; there was little wind-blowns and observed at the oasis(Figs. 4a and 4b). Theaverage wind velocity over the semi-fixed s and y l and ,fixed s and y l and , and oasis was decreased by 29.92%,29.01%, and 44.46%,respectively,compared with thatof 6.432 m s¡1 over the shifting s and y l and . Air tem-perature and RH differences between the four l and -scape types were significant for the first two nightsbefore the dust storm event. During the daytime forthe dust storm,or blowing s and ,or floating dust event,air temperature and RH over the four l and scape typeswere changing synchronously,with no obvious differ-ences between them. The RH difference was largerthan the air temperature difference in the horizon-tal direction(Figs. 4c-f). Air temperature increasedgradually and RH decreased in order of observationsite number through the oasis-desert ecotone,then in-creased at the oasis from site Nos. 1-4. However,thedifference was generally not obvious. Average air tem-perature at 2.0-m height on the semi-fixed s and y l and ,fixed s and y l and , and oasis increased by 3.96%,4.9%, and 9.56%,respectively; while average RH decreasedby 2.84% and 4.18%, and increased by 8.21%,respec-tively,as compared with the shifting s and y l and whereair temperature and RH were 9.396℃ and 19.37%.

Fig. 4. Changes in meteorological factors over the four l and scape types(Nos. 1-4)between 11 and 13 March 2011:(a)wind velocity at 0.5 m,(b)wind velocity at 2.0 m,(c)air temperature at 0.5 m,(d)air temperature at 2.0 m,(e)RHat 0.5 m,(f)RH at 2.0 m,(g)PAR, and (h)solar radiation.

Figure options

Both PAR and solar radiation for the fourl and scape types on every day over the three daysshowed increasing trends. Values started at 8.74×105,1.076×106,1.255×106, and 1.365×106¹mol m¡2 s¡1 and 5.228×105,6.017×105,6.748×105,7.232×105 Wm¡2 for the shifting s and y l and ,semi-fixed s and yl and ,fixed s and y l and , and oasis,respectively. Forthe semi-fixed s and y l and ,fixed s and y l and , and oa-sis,these values increased,respectively,by 23.19%,43.62%, and 56.23% for PAR, and by 15.09%,29.06%, and 38.33% for solar radiation,compared with the val-ues for shifting s and y l and . Feng et al.(2006) and Yaoet al.(2006)reported daily values of PAR during duststorm,overcast and rainy, and sunny days to be 4.09,12.4, and 21.4 MJ m¡2,respectively. PAR research atChaganmaodao feed base in Ejin Banner showed thatPAR on a sunny day was greatest,whilst PAR on anovercast and rainy day was secondary, and PAR ona dust storm day was least. However,in the present study,the value of PAR and solar radiation duringblowing-s and and floating-dust weather on 13 Marchwas least among the three days,which accounted forthe fact that intermittent blowing-s and and floating-dust weather after the dust storm had a further weak-ening effect on PAR and solar radiation,which was fargreater than during the dust storm and blowing-s and weather. The s and -blown wind direction was mainlyconcentrated in an SE and NE direction, and there wasno obvious difference between them. When wind wasblown from the oasis to desert with decreasing VC,s and and dust would be blown up easily and the dustconcentration would be higher in the air when it wascloser to bare shifting s and y l and ,leading to a greaterreduction in PAR and solar radiation. Average tem-perature,RH, and solar radiation gradually increasedfrom the shifting s and y l and to the oasis during and after a dust storm event; these trends were the reverseof those observed on a sunny day.

Fig. 5. Wind direction distributions for the four l and scape types(Nos. 1-4)during 11-13 March 2011.

Figure options

The dust storm event lasted 23.5 h from 2310 BT19 to 2240 BT 20 March 2012(according to the timeof s and -blowing wind on the shifting s and y l and ), and was accompanied by gales and intermittent rainfall inlater periods of the event. The wind velocity at heightsof 0.5 m and 2.0 m weakened from the shifting s and yl and to the oasis over the four l and scape types, and thedegree of weakening at 0.5 m was significantly greaterthan that at 2.0 m(Figs. 6a and 6b). During thedust storm,average wind velocity at 2.0 m over thesemi-fixed s and y l and ,fixed s and y l and , and oasis was,respectively,24.65%,28.76%, and 48.73% less thanthat of the 5.56 m s¡1 over the shifting s and y l and .Average air temperature and RH at 2.0-m height from the shifting s and y l and to the oasis was 7.606,7.578,7.61, and 7.47℃, and 57.98%,58.48%,57.73%, and 60.51%,respectively,all of which were very close(Figs.6c-f). Air temperature and RH differences at night-time between the four l and scape types between 18 and 19 March were relatively significant. Values of PAR and solar radiation over the semi-fixed s and y l and weregreatest,followed by the shifting s and y l and ,fixeds and y l and , and oasis,in that order(Figs. 6g and 6h). During the dust storm,tall rough terrain and high density nebkhas on the semi-fixed s and y l and in-tercepted more blowing s and and floating dust fromwind-drift s and ,which increased absorption of PAR and solar radiation. The weakening effect on PAR and solar radiation during gale and rain conditions on19 March was also relatively significant,but less th and uring the dust storm and blowing-dust weather.

Fig. 6. As in Fig. 4,but for 18-20 March 2012.

Figure options

The main prevailing wind direction was WNW and E for the four l and scape types during the duststorm and there was little change of wind directionspatially,though the direction was more southerly atthe oasis site.

Fig. 7. As in Fig. 5,but for 18-20 March 2012.

Figure options

A dust storm occurred on 1 June 2012,duringwhich the maximum wind velocity was 15.77 m s¡1at a height of 10 m over the shifting s and y l and fron-tier, and the dust storm duration was 19 h from 0103to 2006 BT 1 June 2012. From the shifting s and yl and frontier to the oasis,the wind velocity at differ-ent heights was reduced greatly,especially in the oasiswhere there was almost no s and -driving wind(Figs. 8a and 8b). Average wind velocity at 2.0-m height overthe semi-fixed s and y l and ,fixed s and y l and , and oasiswas reduced by 21.08%,35.72%, and 82.21%,respec-tively,as compared with the 7.644 m s¡1 over theshifting s and y l and site during the dust storm event.Average wind velocity at a height of 0.5 m for the shift-ing s and y l and to the oasis margin sites was 6.52,1.45, and 1.63 m s¡1,respectively,which demonstrates theweakening effect on the wind of the natural vegetationin the oasis-desert ecotone, and of the shelter forestsin the oasis in summer.

Fig. 8. As in Fig. 4,but for 31 May to 2 June 2012.

Figure options

There was a pattern of rainfall and sharp declineof air temperature over the four l and scape types af-ter the dust storm event. The air temperature beganto slowly recover on 2 June and the biggest drop was13℃. Air temperature differences between the threel and scape types in the oasis-desert ecotone betweenthe night and morning were more than those in theperiod from afternoon to sunset, and there were no sig-nificant differences between them. However,especiallyduring night and morning periods,temperature differ-ences between the oasis-desert ecotone and the oasiswere more obvious. Air temperature in the oasis wasconstantly lower than that in the oasis-desert ecotoneafter the end of the dust storm, and this was shownduring the daytime of 2 June. RH differences betweenthe four l and scape types during and after the duststorm event were less than those before the dust storm(Figs. 8c-f). Average daily air temperature and RH at2.0-m height from the shifting s and y l and to oasis siteson 31 May were 27.624,27.551,27.31, and 24.799℃, and 20.77%,21.53%,23.12%, and 41.04%,respec-tively; on 1 June,they were 22.849,22.754,22.586, and 21.232℃, and 37.96%,38.76%,39.03%, and 48.39%,respectively; and on 2 June,they were 19.985,20.143,19.767, and 18.747℃, and 53.19%,52.13%,54.35%, and 65.74%,respectively. Air temperature graduallydecreased and RH gradually increased from the shift-ing s and y l and to the oasis sites before and during the dust storm. Air temperature was highest and RHwas least over the semi-fixed s and y l and during cloudyweather on 2 June.

PAR and solar radiation for the four l and scapetypes on 1 June were at their lowest among the threedays,due to the weakening effect on PAR and so-lar radiation during the dust storm and blowing-s and weather compared to cloudy weather. Over the threedays,PAR and solar radiation were greatest over thesemi-fixed s and y l and , and decreased sequentially oversite Nos. 3,1, and 4(Figs. 8g and 8h). The averagedaily values of PAR and solar radiation over the semi-fixed s and y l and on 1 June were 288.341 ¹mol m¡2 s¡1 and 163.128 W m¡2 respectively; these were reducedby 4.99%,6.95%, and 39.08%, and by 4.03%,10%, and 33.2% over the shifting s and y l and ,fixed s and y l and , and oasis,respectively.

The prevailing wind direction of wind-driven s and wind for the four l and scape types was W and WNW.There were no obvious differences on the whole in winddirection distribution,but wind variability was slightlyincreased at the oasis site(Fig. 9).

Fig. 9. As in Fig. 5,but for 31 May to 2 June 2012.

Figure options

During the last three dust storms,the temper-ature and RH difference in the vertical direction atnight for all four l and scape types was obviously greaterthan during daytime. At all sites at night,air tem-perature at 2.0-m height was higher than that at 0.5m,indicating a temperature inversion layer. Air tem-perature at 2.0 m was lower than that at 0.5 m inthe daytime and the RH changes in the vertical di-rection showed an opposite trend(Fig. 10). Thisis because the ground surface continuously receivesheat in the daytime and the air temperature increasesquickly near the surface; the surface loses heat quicklyat nighttime and part of the heat that is lost is storedin the atmosphere,including at the 2.0-m height. The air temperature difference in the vertical direction atnight in spring over the fixed s and y l and near the oasiswas significantly greater than that at the other threeobservation stations,while in the daytime at the oa-sis it was greater than that in the oasis-desert ecotone.The air temperature difference in the vertical directionin the oasis before and during the dust storm events insummer showed the greatest value; however,air tem-perature and RH differences in the vertical directionbetween the four l and scape types during and after thedust storms both in the daytime and at night were veryclose(Fig. 10). Air temperature and RH differencesin the vertical direction were largest at site No. 3.The RH difference in the vertical direction at the oa-sis was greatest before the dust storm and was greaterthan that at other sites. This indicates that heat and vapor near the surface before the dust storm were con-tinuously exchanging and migrating; furthermore,themoisture-heat exchange on the fixed s and y l and be-side the oasis was largest in spring. At the borderbetween the oasis and the oasis-desert ecotone,therewas an influence from the uptake of adhered conden-sational water by ungerminated vegetation and plantabsorption of solar radiation. Large vertical changeshelped to form natural channels for water and heattransfer in the horizontal direction. Before the duststorms,atmospheric stratification was very unstable and water and heat were exchanged frequently; how-ever,atmospheric stratification near the surface be-came relatively stable during or after the dust stormevents.

Fig. 10. Air temperature and RH changes in the vertical direction for the four l and scape types(Nos. 1-4)during thethree dust storms.

Figure options

Across the four dust storm events,average wind velocity reached a maximum on 1 June 2012. Whenaverage wind velocity were comparatively smaller,allwind profiles primarily fitted a logarithmic correlationon 14 May 2010 and 12 March 2011, and most windprofiles did not fit a logarithmic correlation when theaverage wind velocity were stronger in the dust stormevents on 19 March and 1 June 2012. Wind veloc-ity near the ground surface was reduced when windapproached or entered the oasis. The slopes of windprofiles under 4.0 m became reduced from the shiftings and y l and frontier to the oasis; wind velocity differ-ences near the ground surface in the vertical directionwere greater than those at 4.0-m height; and wind ve-locity vertical profiles over 4.0 m were close to a linear distribution(Fig. 11). Wind velocity at a height of2.0 m was greater than that at a height of 4.0 m atsite No. 3 on 19 March 2012, and this may be dueto strong winds near the oasis edge as a result of thebuffering effect of shelter-forests,causing strengthenedturbulence.

Fig. 11. Average wind velocity profiles for varying l and scape types(Nos. 1-4)during four dust storm events:(a)14May 2010,(b)12 March 2011,(c)19 March 2012, and (d)1 June 2012.

Figure options

The present results show that temperature and RH differences among four l and scape types were rela-tively small during dust storm events. However,theywere relatively large before dust storm occurrence, and this was related to invasion by cold,high-pressure air-flow. Significant differences in temperature and RHbetween the desert area and oasis provide an impor-tant basis for dust storm prevention and mitigation. Infuture,to accurately predict and prevent dust stormsfor oasis farml and maintenance,ground temperature,latent heat flux, and air pressure over different l and -scape types should be observed to further reveal thespatial variability of meteorological factors during duststorm events.

The present research also shows that VC,windvelocity,wind direction, and topography were mainfactors influencing the spatial differences during thedust storm events between the oasis-desert ecotone and the oasis, and that the vegetation distributionwas the most important factor(Qian et al., 2004).Chen and Zhou(2010)pointed out that local VC and soil factors are also key factors besides topography,wind velocity,wind direction, and atmospheric sta-bility in influencing the occurrence and migration ofdust storms. Through quantitative analysis between normalized difference vegetation index(NDVI) and aerosol index,a significant negative relationship be-tween them was demonstrated,indicating that theexistence of vegetation can restrain floating-dust oc-currence to a certain extent.

In spring,vegetation has not yet fully emerged and soil moisture is extremely low in the Cele area.Dust storms can cause blown-s and hazards, and influ-ence the safety of transportation and electric powerindustries,thereby damaging human health. Mean-while,they also affect plant photosynthesis,whichcan influence growth of natural vegetation and cropyields. Therefore,only through improving soil mois-ture conditions,increasing VC,banning unauthorizedlogging,overgrazing, and abuse of reclamation,canthe damage caused by s and storm events to oases beeffectively prevented and reduced. In summary,wedraw the following conclusions from the present work.

(1)Average wind velocity gradually decreasedwith increasing VC from the shifting s and y l and fron-tier to the oasis,with the maximum decrease in theinner oasis. During the dust storms,natural veg-etation in the oasis-desert ecotone demonstrated abuffering effect on the wind and an interception ofblown s and . It was more effective at these processesin summer than in spring.

(2)During dust storm events,differences in airtemperature and RH were small between the oasis-desert ecotone and the inner oasis. Air temperatureincreased and RH decreased gradually, and PAR and solar radiation increased significantly from the shiftings and y l and frontier to the oasis. When the prevail-ing wind direction was E and SES,PAR and solarradiation were significantly reduced during the duststorm events and the subsequent s and -blowing and floating-dust conditions, and this effect was muchstronger than for similar weather conditions withoutdust storm events such as s and -blowing and overcast and /or rainy days.

(3)Air temperature,PAR, and solar radiationshowed their greatest values over the semi-fixed s and yl and when the prevailing wind direction was W and WNW during dust storm events. Values at the othersites decreased in the following order: shifting s and yl and ,semi-fixed s and y l and , and oasis. This could beattributed to the high sun angle and more intercep-tion of blown s and and dust by the high terrain of thesemi-fixed s and y l and .

(4)RH differences among the four l and scapetypes during the dust storms in summer were morethan those in spring,which was caused by higher VC and the irrigation effect of the oasis. Air temperature and RH differences among the four l and scape typeswere very small both in the horizontal and verticaldirections; however,the differences increased abruptlybefore dust storm occurrence.

(5)The distribution of wind direction for the fourl and scape types changed very little during the duststorm events in summer; however,the prevailing winddirection was more varied when wind entered the oa-sis.