Summary of Meteorological Field Experiments in the Taklimakan Desert,China∗

HE Qing,JIN Lili

(Taklimakan Desert Meteorology Field Experiment Station of CMA,Institute of Desert Meteorology,China Meteorological Administration,Urumqi Xinjiang 830002,China)

Abstract: In this paper, the latest progress of meteorological research in Taklimakan desert is summarized. On the basis of summarizing the progress of desert meteorology in recent years, the scientific problems of desert weather and climate are discussed,desert land-air interaction,desert atmospheric environment in Taklimakan Desert meteorology field scientific experiment base are explored,and the key scientific problems faced by Taklimakan Desert field scientific experiment results are summarized.Based on multi-scale land-air interaction over rugged terrain,the characteristics and parameterization of sand dust weather,the basic ideas of the key scientific problems of Taklimakan Desert meteorology are discussed,and some preliminary suggestions for further research on Taklimakan Desert meteorology are proposed.

Key words: Taklimakan Desert; Field Atmospheric Scientific Experiment; Land Surface Process; Physics of Blown Sand;Atmospheric Environment

0 Introduction

The Taklimakan Desert with an area of 33.76 ×104km2, is the largest desert in China, located in the vast areas of between north latitude 37°～41°and east longitude 77°～90°,in the Tarim Basin in the southern Xinjiang of China,it is 1 070 km long from east to west and 410 km wide from south to north[1]. The sand-dust weather in Taklimakan Desert lasts for a long time, spans the whole spring and summer seasons, and the westerly belt can bring the dust formed by the extremely strong dust storm to thousands of kilometers away,which has a direct impact on the climate and environment changes.

Since the 1980s, based on short-term observations, Chinese scientists have successively carried out research on the Taklimakan Desert,and achieved basic achievements in the field of desert climate. Since its establishment in 2002,Urumqi Institute of Desert Meteorology of China Meteorological Administration has carried out field observations on sandstorm weather with different intensities in the Taklimakan Desert, and obtained the grading test indicators of sandstorm. Since 2005, with the joint support of the China Meteorological Administration and the Ministry of Finance, the Urumqi Institute of Desert Meteorology of the China Meteorological Administration has successively established experimental stations for meteorological observation of desert boundary layer and wind sand observation experimental fields in the Taklimakan Desert,Gurbantunggut Desert and Badain Jaran Desert(in Inner Mongolia Autonomous Region,China),and it has built the Atmospheric Observation Network of the near surface boundary layer in the North-South longitudinal section,which provides the key technical support for the major disastrous weather forecast and early warning in the desert and its surrounding areas.With the establishment of the desert atmosphere comprehensive observation and research platform,the research and development of boundary layer detection technology and integrated method has become suitable for desert area,as well as suitable for the systematic monitoring data and theoretical research of desert boundary layer Many problems such as the inhomogeneity of desert undulating surface, the understanding of structure change from near surface layer to boundary layer, and the role of desert land surface process in the climate system have been gradually solved. At the same time,observation experiments on desert land-atmosphere interactions and sand transport processes were used to develop and parameterize a numerical model of desert land surface processes,so as to provide basic data and technical support for the refined observation of catastrophic desert weather and to improve mesoscale numerical predicting models.

Therefore, this paper attempts to summarize the research achievements and key scientific problems of Taklimakan Desert meteorology on the basis of in-depth thinking about the field scientific experiments and related scientific problems of Taklimakan Desert meteorology,and preliminarily proposed the basic ideas to solve the key scientific problems of desert meteorology on non-uniform underlying surface. It will provide a reference for further research on desert meteorology in the future.

1 The Scientific Experimental Base of Taklimakan Desert Meteorological Field

The Scientific Testing Base of Taklimakan Desert Meteorological Field of the China Meteorological Administration is located in the Tazhong Petroleum Operation Area in the hinterland of the Taklimakan desert, with a geographical location of 38°58′N，83°39′E,1 099.3 m above sea level. The experimental base was established in July 1996 and incorporated into the National Basic Meteorological Station of China Meteorological Administration on January 1, 1999. In 2002, the Tazhong Atmospheric Environment Observation Experimental Station of Urumqi Desert Meteorological Institute of China Meteorological Administration was established and incorporated into the National Benchmark Meteorological Station of China Meteorological Administration in 2008. At present,the station has been continuously observed for more than 20 years.The observation results of meteorological field experiment base in Taklimakan Desert of China Meteorological Administration may represent the basic characteristics of atmospheric physics and atmospheric environment in Taklimakan Desert. The experimental base has representative and reference value for the study of land air interaction on the underlying surface of mobile desert, desert boundary layer and turbulence, desert atmospheric composition concentration and its change. It is an ideal experimental site for the study of physical process of atmospheric boundary layer of mobile desert,land surface process of quicksand and dust atmospheric environment. Based on the whole boundary layer detection and a variety of experimental data analysis,the structure characteristics of desert weather system obtained from the test base have good regional representativeness in southern Xinjiang,China. Therefore,the long-term desert positioning observation will effectively promote the basic and applied research in the field of desert meteorology.

Fig 1 Location of meteorological field test base in the Taklimakan desert

Since 2005,with the support of the China Meteorological Administration and the Ministry of Finance,the 80 m gradient tower flux detection system,radiation detection system and eddy correlation detection system have been established in order to observe the surface structure,water and heat flux exchange and surface radiation energy budget of Taklimakan Desert. At the same time,the tethered airship detection system,wind profile radar detection system and UAV were introduced to detect the meteorological elements and vertical distribution of dust concentration in the boundary layer. In 2016,with the support of the Ministry of Finance,a 100 m gradient iron tower flux detection system(40°49′N,84°18′E,932 m above sea level),was established, and it built the boundary layer detection and contrast system in the hinterland of desert and the northern transitional zone.

2 Overview of Meteorological Achievements in Taklimakan Desert

Since the 1970s,a group of scientists led by Charney J G has studied the dynamic mechanisms of arid climate formation such as albedo and heat balance in the Sahara Desert and the Sahel, and they found that high albedo causes the desert heat sinks,which leads to the decrease of precipitation[2,3].Since the 1980s,Henderson-Sellers[4]and Cunnington[5]have carried out detailed analysis and research on the albedo of different underlying surfaces;Hwang[6]studied the effect of soil moisture on the surface energy balance of bare soil;Thomas T.Warner’s[7]study covered all aspects of desert climate,including the causes of large-scale and local-scale drought,desert precipitation characteristics,sandstorm,desert climate change,desertification,desert land surface physics, numerical simulation of desert atmosphere and the impact of desert climate on human beings,and it provided a summary of climate and surface attributes of desert regions in the world.

Since the 1930s, the research methods on sandstorm or sandstorm movement have made great progress. Bagnold’s works laid the foundation for the study of aeolian sand physics[8]. Before the 1970s,these experimental methods mentioned above were the mainstream tools for the study on sandstorm or its movement, and were used by Chepil[9], Yakubov[10],Hecun Malone[11],Zingg[12],Owen[13]etc. In the late 1980s,especially since the 1990s,due to the development of science and technology,high-frequency,automated,small-scale monitoring instruments have been used in the research of sand movement,such as high-speed photography monitoring system[14].

Some researches on the interaction between boundary layer and atmosphere in the early stage were carried out in Xinjiang where is the largest desert area of China, but it mainly limited with the area of desert fringe. Since the 1980s,the comprehensive scientific investigation and research on the Taklimakan Desert mainly studied the severe environmental problems in the desert, such as sandstorm, dust tornado, sandstorm, strong drought, high temperature, severe cold, strong light radiation and so on,and preliminarily the weather climate dynamics and thermal mechanism of the Taklimakan Desert were understood.

2.1 Desert climate and sand-dust environment

The climate of Taklimakan Desert in recent 50 years showed an obvious trend of“warming and wetting”. The annual average temperature increased abruptly in 1970, the precipitation showed an increasing trend, the number of precipitation days showed a decreasing trend, the precipitation increased abruptly in 1981, while the potential evapotranspiration and surface dryness decreased abruptly in 1981. Temperature, precipitation, potential evapotranspiration and surface dryness have periodic changes on quasi-3-year,8-year and 16-23-year interannual scales,respectively[15,16].

In the hinterland of Taklimakan Desert,the days of floating dust and blowing sand were on the rise,while the days of sandstorm were on the decline,from January 2004 to December 2009. In spring and summer,the dust weather is frequent,and the monthly average mass concentration of Total Suspended Particles(TSP)mainly concentrated in March to September,with the highest concentration in April and May(the average monthly mass concentration of PM10at the height of 4 m is 846.0 μg·m−3),and then decreased gradually Most of the sandstorms in Taklimakan Desert are systematic. During the sandstorm,the concentration of dust particles is the highest,and the greater the wind speed is,the higher the particle concentration is.The concentration of PM10at 80 m height is much higher than that of PM2.5and PM1.0,while the mass concentration of PM2.5and PM1.0at 80 m height is significantly lower than that at 4 m height,and the daily variation of dust concentration is completely consistent with the diurnal variation of wind speed[17−21]. When the wind speed reaches at 3.6 m · s−1, the concentration of black carbon aerosol begin to increase with the wind speed, and there are obvious differences in the concentration level of black carbon aerosol under different wind directions[22−24]

The dust aerosols which are transported into the air changes the absorption and energy distribution of solar radiation on the surface. This special weather condition inevitably makes the thermal effect of Taklimakan Desert, which bring its own particularity[25]. The aerosol in the hinterland of Taklimakan desert has the largest scattering effect at the solar radiation of635 nm,followed by 525 nm and the smallest at 450 nm. The daily change of the three-band scattering coefficient is consistent with the PM10mass concentration, showing a single peak variation: higher at night and lower during the day; the annual change of those three-band scattering coefficient is basically the same,which is close to that of PM10;the three-band scattering coefficient is the largest under sandstorm,the second is blowing sand,and the smallest is floating dust;there is a significant positive correlation between those three-band scattering coefficient and mass concentration of PM10. The correlation between PM10mass concentration and 525 nm scattering coefficient is the largest,followed by 450 nm and 635 nm[26−28]. In the two main areas(Pishan-Hetian-Minfeng and Xiaotang-Tazhong),sand-dust events occurs more than 80 days per year. The center of sandstorm is in the north part of the desert(Xiaotang, 46.9 days), but the center of sand-blowing(Pishan, 86.4 days)and the area with frequent floating dust events (Minfeng, 113.5 days) are located in the southwest and the south edge of the desert respectively. The occurrence of sandstorm generally decreased from 1961 to 2010,and the occurrence of blowing sand showed an upward trend from 1961 to 1979,and then decreased as a whole. The main reason is that the temporal changes of sand-dust events are mainly affected by strong winds and daily temperatures,the average correlation coefficients are 0.46 and 0.41 for these variables respectively[29].

2.2 Desert land-air interaction

2.2.1 Atmospheric boundary layer of Desert

In the field of atmospheric boundary layer of desert,scholars at home and abroad have carried out a number of researches and achieved some important research results. A deep convective boundary layer up to 5.5 km was observed in the Sahara Desert,with a clear structure of the residual layer[30].A convective boundary layer over 4 000 m thick was observed in the Hexi Corridor area of China[31]. The structure of the“cold island effect”was firstly discovered in the oasis of the boundary layer in the arid zone in the 1980s. Hu Yinqiao et al[33]discovered the phenomenon of inverse humidity in the desert atmosphere near an oasis in the“Heihe Experiment”in the 1990s,and the characteristics of the thermal inner boundary layer under the interaction between the oasis and the desert were summarized. In Dunhuang, China,it is found that there was a convective boundary layer with a thickness of more than 4 000 m in the clear sky in summer,and the height of the stable boundary layer at night might also exceed 1 000 m[34−37]. The convective mixing layer in the Badain Jaran Desert of China may reach 3 000m in summer[38].

In the hinterland of Taklimakan Desert,the development of clear skyturbulence is intense in summer,and the development of convective boundary layer is very profound, with the maximum height up to 4 km[39−41], and the monthly average thickness of the stable boundary layer is 570 m,the residual mixed layer(RML)is 2 700 m and the top cover of residual inversion layer is 350 m respectively[42]. There are strong“cold island effects”and“wet island effects”in artificial green spaces in the desert hinterland[43]. The local microclimate in the center and fringe of the artificial greenbelt in the hinterland of the Taklimakan Desert is mainly manifested in high wind speed and humidity in spring,summer and autumn,while temperature and humidity in winter are mainly affected by inverse temperature and inverse humidity[44].

2.2.2 Land surface processes of desert

In the hinterland of Taklimakan Desert,there is inversion phenomenon in the near surface layer at night in summer,and the change of daytime temperature is opposite;the surface radiation balance is mainly positive,and the surface heat exchange is dominated by turbulent influenza heat. The surface sensible heat and latent heat change with the fluctuation of the solar altitude angle. The maximum value of latent heat appears in the early morning,and the peak value of sensible heat appears at noon[45]. The distribution of sensible heat flux and latent heat flux on the surface of Taklimakan desert exhibits great seasonal and regional differences. In spring, the sensible heat flux tends to decrease in the north of the desert, and increase in the south. In spring,the latent heat flux tends to decrease in the north and northwest of the desert,and increase in the south and southeast of the desert, while the trend in winter, summer and autumn is not very obvious[46−47]; The surface heat transfer between artificial green land and natural sand surface in desert hinterland is mainly sensible heat transfer, but the sensible heat flux of natural sand surface is higher and the latent heat flux is lower[48].

2.2.2.1 Solar radiation of desert

The Taklimakan desert is rarely affected by human activities, and the influencing factors of solar radiation are mainly related to clouds and dust aerosols[49]. The average annual amount of total radiation,scattered radiation and horizontal direct radiation in the hinterland of Taklimakan Desert are 6 619 MJ·m−2,3 507.8 MJ·m−2and 2 203.5 MJ·m−2,respectively. The daily peak values of total radiation in typical sunny days are 2.4 times of scattered radiation and 1.5 times of direct radiation,respectively. The increased value of scattered radiation in sandstorm days is basically consistent with the total radiation[50].The scattering radiation in the desert hinterland increases in autumn and winter, raises exponentially with the rise of solar altitude angle,and decreases rapidly with the increase of atmospheric quality[51].The average annual total ultraviolet radiation in the desert hinterland is 320.7 MJ·m−2,with the maximum value of 62.5 w·m−2(June)and the minimum value of 29.3 w·m−2(December);the annual total UV-B radiation is 8.59 MJ·m−2,and the annual peak is 2.51 w·m−2in June. The frequent occurrence of sand-dust weather in spring and summer leads the conspicuous fluctuation of diurnal variation of ultraviolet radiation. In floating dusty days,sandy sky days and sand-dusty days,the relative attenuation of ultraviolet radiation is 26%,38% and 45% respectively. The attenuation of ultraviolet radiation caused by sand-dust weather is 2～4 times than that of caused by cloud cover weather[52−53].

2.2.2.2 Thermal effect of desert and soil

The average soil heat capacity,thermal conductivity and thermal diffusivity in the hinterland of Taklimakan Desert are 1.559（± 0.140）× 106J·m−3K−1, 0.234（± 021) w · M−1· k−1and 1.504（± 0.110）× 107m−2s−1respectively; Soil heat capacity and heat conductivity present obvious seasonal changes,with stable low values in winter and unstable high values in summer[54]. The surface emissivity of Taklimakan desert is the highest near the oasis,reaching 0.93,the value of oasis desert transition zone is 0.91 ～0.92, the value of other desert areas is 0.90 ～0.91, and the value of desert hinterland area is less than 0.90[55]. The annual average value of surface albedo in the hinterland and northern edge of Taklimakan desert is 0.27[56],and the daily average value of albedo has obvious seasonal variation, which is higher in winter and lower in summer. The daily variation of surface albedo with snow cover is varied 0.18～0.97,daily average value is 0.60,more inclined to reverse“J”type,showing a pattern that morning is higher than evening,and the average difference between morning and evening is 0.13. There was a negative correlation between snow albedo and surface temperature,the correlation coefficient was-0.71,and a negative correlation between snow albedo and soil moisture at 5 cm depth, the correlation coefficient was -0.74[57].The surface albedo is varied in different weather in the northern edge of Taklimakan Desert, and it decreases in rainy days and increases in snowy days. When the clearness index is<0.3,the surface albedo fluctuates greatly. Soil moisture impacts on the daily variation of surface albedo in the northern edge of the desert,and the surface albedo is sensitive to the shallow soil moisture in the range of 0.097-0.13[56]. Surface albedo on the northern edge of desert(α)and specific emissivity(ε)are 0.27 and 0. 91 respectively,which are consistent with the values of Taklimakan Desert hinterland and North American Basin Desert,andαandεvalues can be obtained from remote sensing products[58].

The annual average value of soil heat flux at 1 cm in the hinterland of Taklimakan desert is 1.9 w · m−2, the annual maximum value is 334.1 w · m−2, and the annual minimum value is -184.2 w · m−2; The basic featureis summer>spring>autumn>winter, each soil layer of soil heat flux shows obvious diurnal variation characteristics, and there are certain differences in diurnal variation characteristics under different weather conditions,and it is most significantly affected by the weather at soil depth of 1 cm[59]. On the daily variation scale,the soil heat flux in the hinterland and the northern edge of the desert has obvious diurnal variation characteristics. The daily average variation of the soil heat flux in the hinterland of the desert is less than that in the northern edge in January，in April they are relatively close,and in July and October the variation range of the soil heat flux in the hinterland of the desert is significantly higher than that in the northern edge[60].

2.2.2.3 Concentration of carbon dioxide and ozone in desert

In the hinterland of the Taklimakan Desert, the land surface absorbs CO2in the daytime and emits CO2at night, and the surface absorption intensity is significantly greater than the surface emission,and the CO2flux is greatly affected by the atmospheric stability[61]. The soil respiration rate in the hinterland of desert is generally slow, but it shows obvious diurnal fluctuation and seasonal variation characteristics[62].However,the soil respiration rate of saline alkali land and quicksand land in the northern edge of desert is relatively slow[63]. There is interaction between the soil respiration and weak meteorological conditions,and they jointly regulate and control the variation of CO2concentration in the near surface layer[64]. The diurnal variation of soil respiration in winter showed a significant unimodal curve. There is extremely significant or significant positive linear relationship between soil respiration,air temperature in each layer and surface temperature at 0 cm,and it had obvious linear relationship with soil moisture at 5 cm,and surface temperature at 0 cm had the greatest contribution to it[65].

The concentration of surface ozone in the hinterland of Taklimakan Desert ranges from 33.8 g · m−3to 65.3 g · m−3,with an average concentration of 49.0 g·m−3±0.45 g·m−3;The change of ozone concentration shows weekend effect,which is smooth at night and dramatic during the day,reaching the lowest value around 09:00 and the highest value around 18:00.During the sandstorm,the ozone concentration decreases significantly[18]

2.2.2.4 Key parameters of desert land surface process

The range of dynamic roughness in the hinterland of the Taklimakan Desert is 2.7×10−5m～8.0×10−5m,and it reaches 21.04×10−5m～91.32×10−5m in winter[54].The peak value of dynamic roughness(z0m)is 5.858×10−3m,which is similar to Mojave Desert,Peruvian desert,Sonoran desert,Heihe desert and Badain Jaran Desert.In the north of the Taklimakan Desert,the peak value of thermodynamic roughness (z0h) is 1.965 × 10−4m, which is different from the hinterland of Taklimakan Desert. The annual average additional damping of heat transfer(kB−1)is 2.5,which is different from HEIFE Gobi and desert,but similar to Qinghai Tibet Plateau and HAPEX desert steppe. The diurnal variation of z0mand z0his not obvious, but the seasonal variation is obvious. The diurnal variation and seasonal variation of kB−1are not obvious. z0mis obviously affected by local wind direction. There are many undulating dunes in the prevailing and counter prevailing wind directions, which are consistent with the peak direction of z0m. The daily average values(24 hours)of turbulent dynamic transport coefficient(Cd) and turbulent thermal transport coefficient (Ch) are 6.34 × 103and 5.96 × 103, respectively , which is higher than the hinterland of Taklimakan Desert and gobi region,and similar to HEIFE desert. Under the prevailing wind direction(NNE–ESE),according to the similarity theory, the average Cdand Chhave the same magnitude. Under different wind directions,the relationship between Cdand Ch,wind speed(U)and stability parameter(Z/L)are different. When the wind speed is lower than 3 m·s−1and its minimum value is 1 m·s−1～2 m·s−1,Cdand Chdecrease rapidly. It should be pointed out that using sensible heat flux estimation is more important than using other estimation methods when obtaining the value of(ε)[58].

2.2.2.5 Turbulent flow of desert

Under weak instability or near neutral condition the scale of characteristic length of turbulent vortex in desert hinterland is the largest, and it decreases with the increase of instability, with the increase of stability it firstly decreases rapidly and then decreases slowly. The wind speed of near ground layer decreases firstly and then increases during the sandstorm transit.The downward transport of momentum is obvious at 10 m height,and the heat transfer shows slower rising tendency;Before the sandstorm passing through, the near ground is a weak and stable inversion layer, and the air is in a state of warm-dry condition. The vertical air flow at 10 m height shows a systematic downward movement. With the sandstorm outbreak,the turbulence exchange is significantly enhanced, and the air flow shows an upward movement trend, however the intensity is not large,and it still dominated by the horizontal turbulence energy[67].

In most cases, the distribution of turbulence velocity spectrum in the northern margin of desert meets the exponential rate of -2/3, and the degree of compliance of inertial sub region in high frequency section in vertical direction is higher,followed by horizontal direction; The concentration of CO2and H2O was lower; In general the temperature spectrum has a good correlation with dimensionless frequency. In most cases the fitting value of the co spectral slope of vertical wind speed and radial wind speed is closer to - 1, and more in line with - 4/5 oblique line in near neutral stratification condition. The peak value of the co spectrum is larger and about one magnitude under the stable layer junction than that of the unstable layer junction;The high frequency band co spectrum decreases in the near linear mode under the condition of unstable stratification.The peak wavelength of u spectrum decreases with the increase of stability, while the peak wavelength of V spectrum and t spectrum do not increase or decrease regularly with the increase of stability; The peak wavelengths of U,V,W and T are about 67 m～827 m,69 m～2 417 m,4 m～54 m and 12 m～661 m,respectively[68].

2.3 Study on Aeolian Sand Physics

2.3.1 Characteristics of surface environmental

The average grain size of surface soil in Taklimakan desert is 15.6 μm～250 μm. It consists of fine sand(125 μm～250μm),very fine sand(62.5 μm～125 μm),coarse silt(31 μm～62.5 μm)or its mixture[69]. The sand transported by aeolian sand flow in the hinterland of the desert and the adjacent strata(below 2 m height)of the northern margin is mainly fine sand(125 μm～250 μm),very fine sand (62.5 μm～125 μm),most of them are fine sand, accounting for 43.8 %～75.5% of the sediment transport. The content of coarse sand(>250 μm) in each height layer is very little[70]. The average grain size of sand transported by aeolian sand flow is 62.5 μm～125 μm,the average grain size of 5 cm near the ground is the largest,and that of 2 m is the smallest,which indicates that the smaller grain size can jump higher under the same conditions[71].

In the hinterland of Taklimakan Desert, sand with grain size larger than 0.3 mm has better transparency, and there are also a small number of sand particles with poor transparency, such as red and black sand particles. The surface abrasion of those sand particles is more obvious, the edges and corners are less, and the roundness is better. The distribution of roundness value is relatively concentrated,mostly between 0.7 and 1.0,and the proportion of sand particles with roundness value between 0.8 and 0.9 is the largest,especially at the height of 5 cm,10 cm and 200 cm;there are various shapes of sand particles with the size of 0.125 mm～0.3 mm. The roundness values of sand particles are concentrated in the range of 0.7 ～1.0,and 41.49%of them are less than 0.7; Among the sand particles with grain size between 0.074 mm and 0.125 mm,the number of strip sand particles increases, the shape of sand particles becomes more complex, the surface abrasion becomes lighter,and the edges and corners become sharper[72].

2.3.2 Study on wind blown sand flux

In the hinterland of Taklimakan Desert, the structure of 0 cm～100 cm aeolian sand flow completely conforms to the exponential distribution, but there is no such characteristic in the northern margin. The grain size of aeolian sand flow is mainly fine sand,very fine sand and silty sand,of which the very fine sand accounts for 43.8%～75.5%of the total amount.The amount of sand transport has a downward trend with the increase of altitude. The sand movement in the desert hinterland is mainly concentrated in the range of 20 cm～30 cm near the surface. The amount of sand transported by 0 cm can be used as the amount of creeping sand,and the amount of creeping sand accounts for the total amount of sand transport the proportion is about 11.6%. There is a big difference between the direction distribution of creeping sand transport and the wind direction distribution. The wind speed profile of the ground layer in the wind-sand flow is affected by the interaction of wind-sand,and no longer conforms to the logarithmic distribution. It is more in line with the power function distribution(u=azb), and the fitting coefficients are all greater than 0.93[70,71].

In the desert hinterland, the wind direction of sand onset wind and sand transport potential are mainly distributed in ENE,NE and E directions[73]. In the process of sand-dust weather(sandstorm and sand blowing),with the increase of wind speed, the sand transport at each height also increases. In the two weather processes, the average grain size of sand firstly decreases and then increases in the vertical height. Using the data of H-sensit wind erosion sensor, anemometer and micro gradient sediment collection instrument, it is found that in the range of 85 mm height, with the increase of wind speed, the ratio of jump amount/creep amount decreases in a negative power function. When the wind speed is about 8.5 m · s−1, the ratio of jump amount/creep amount is about 8～10;When the wind speed reaches 6.9 m·s−1,the maximum jump is 1 952.01g;When the wind speed reaches 7.6 m·s−1,the maximum creep is 211.79 g. The ratio of total jump to total creep decreases exponentially with the increase of wind speed. The ratio of total jump to total creep at different wind speeds in a typical weather process is about 8～24. With the increase of wind speed, the sediment transport is more and more concentrated in the range of 0 mm～35 mm. There is a good linear relationship between the total amount of sand collection and the number of saltation particles recorded by the Sensit wind erosion sensor the averageR2value is 0.605 3,the average sand collection efficiency of the automatic sand collector at the height of 5 cm is 94.3%. During the observation period,there are significant differences in the amount of sand-dust transported by different sand dust weather processes. The maximum horizontal flux of sand-dust in a cross section of 2 cm (wide) × 5 cm (high) is about 190.335 kg, and the minimum is 1.2 kg. In dusty weather,the maximum sand transport rate appears at a height of 5 mm～15 mm,and the minimum appears at a height of 35 mm～85 mm. In the sand blowing weather,when the wind speed is greater than 9.2 m·s−1,the maximum sediment transport rate is 0 mm～5 mm. In sandstorm weather, the inflection point wind speed is 7.5 m · s−1.When it is less than 7.5 m · s−1,the increase of sand transport rate is not significant,when it is greater than 7.5 m·s−1,the increase of sand transport rate is significant[74−78]. Using BSNE sand collector,it was observed that the horizontal flux of sand dust on the top of sand dune and flat sand decreased significantly with height. Under these two underlying surface conditions,the change of horizontal flux of sand dust with height was in good accordance with the power function relationship[79].

The vertical sand flux increases with the rise of wind speed, and the maximum flux is concentrated in the afternoon.When it is 2 m and the wind speed is 2 m · s−1, the vertical sand dust flux in the hinterland and the northern edge of the desert is close to 3 kg·m−2. When it is 2 m and the wind speed is 6 m·s−1,the vertical sand dust flux in the hinterland and the northern edge of the desert is 10 kg · m−2. When the wind speed is 2 m · s−1, the horizontal dust flux in the hinterland and north edge is about 20 kg·m−2,and the value in the hinterland is slightly larger. When the wind speed is 4 m·s−1,the horizontal dust flux in the hinterland and north edge is about 40 kg · m−2, and the value in the hinterland is larger. When the wind speed is 6 m·s−1,the horizontal dust flux in the hinterland is about 70 kg·m−2,and the horizontal dust flux in the northern margin is about 65 kg·m−2[80].

The average particle size of Sandstorm in the hinterland of desert is 70 μm～85 μm. Because of the existence of sand dunes and valleys,the horizontal dust flux increases with the increase of height in the lower altitude,but remains unchanged when it exceeds above 32 m. The vertical distribution of boundary layer is controlled by wind speed, and the average flux varies from 8 kg·m−2to 14 kg·m−2. The size of dust particles PM100and below accounts for 60%～80%of the collected samples, in which PM0−2.5is 0.9 %～2.5%, PM0−10is 3.5%～7.0%, PM0−20is 5.0 %～14.0%, and PM0−50is 20 %～40%. The average vertical flux potential of dust is about 0.29 kg·m2,and particles smaller than PM20are transported from 80 m to the upper boundary layer and free atmosphere of the planet[81].

2.3.3 Transport mechanism of near surface sediment

In the hinterland of Taklimakan Desert, the climate factor index of annual average wind erosion (the main index to evaluate the potential wind erosion capacity of a certain area)is 28.3,13.9 in summer,0.7 in winter,and the average surface roughness is 6.32×10−5m,the surface roughness is small,which aggravates the degree of wind erosion of soil in this area[82].The critical friction velocity in the hinterland of the Taklimakan desert is 0.24 m·s−1. Yang Xinghua et al[83]found that the calculated results of the Lettau formula were the closest to the measured values. The critical friction velocity was 0.26 m·s−1in spring and summer,and the critical wind velocity was about 4.1 m·s−1at 2 m height;The change of dust flux was consistent with the change of wind speed and friction velocity. Zhou Chenglong et al[73]comprehensively considered the surface soil particle size,soil moisture,air density and other factors,and obtained the critical friction velocity(0.24 m·s−1～0.36 m·s−1)and critical sand blowing velocity(3.9 m·s−1～5.9 m·s−1,with an average of 5.1 m·s−1)at 2 m height in the hinterland of desert;The sand threshold in the desert hinterland was the highest in summer,the second highest in winter and the lowest in spring.

There are some differences in the critical wind speed obtained by different time steps. With the decrease of time step,the critical wind speed is more and more refined. There are different parameterization schemes indifferent values of critical wind speed. Based on Marticorena and Shao schemes,the mean values of critical wind speed are 4.88 m·s−1and 6.24 m·s−1,respectively[84−85]. The minimum wind speed and critical starting friction velocity obtained by H11LIN wind erosion sensor are 6.0 m·s−1and 0.25 m·s−1respectively[86].

The saltation activity of Taklimakan Desert often occurs in the daytime of each season, and the most active time of saltation is about 11:30 local time and lasts about 16:30. From September 1,2008 to August 31,2010,the time of jumping activity accounted for more than 3%of the total time of the whole year,and inclined to reach the peak in the windy months of spring and summer. However, it tends to be the smallest in the months when the wind speed is weak in winter. In the extremely arid Tazhong area,precipitation has no significant effect on reducing sand saltation[87−88].

Gauss time fraction method developed by Stout(interval is 1 day), the obtained friction velocity threshold is 3.03 m·s−1～5.62 m·s−1(field test observation method),the value obtained by Kurosak method is 3.71 m·s−1～5.74 m·s−1(statistical calculation method),and the value given by Marticorena and Shao model is 4.87 m·s−1～4.90 m·s−1and 5.82 m·s−1～6.78 m·s−1(model parameterization method)[85]. The total horizontal dust fluxes estimated by Stout, Kurosak, Marticorena and Shao methods are 1 311.9 kg·m−1,1 166.4 kg·m−1,1 279.9 kg·m−1and 661.6 kg·m−1,respectively,while the observed values are 732.9 kg·m−1.The correlation coefficients between estimated values and observed values based on Stout,Kurosak,Marticorena and Shao methods are 0.75, 0.79, 0.77 and 0.83, respectively. According to the friction velocity threshold,the duration of sand jumping values estimated by Stout,Kurosak,Marticorena and Shao methods are 8 211 min,6 575 min,7 567 min and 3 463 min respectively,while the correct time is 6 208 min,5 646 min,5 986 min and 3 346 min respectively.

The threshold of wind velocity(TWV) mainly occurrs in the daytime (08:00～20:00), and in spring (6.4%), summer(5.6%), autumn (1.7%) and winter (0.8%). The DUP increased to 50% on June 17 and 90% on September 1. The DUP was observed from April 1 to July 31 for two consecutive years from September 1, 2008 to August 31, 2010. The result showed that 75.4% DUP was produced. And The results also show that different time resolution of wind speed will affect the calculation of DUP.When the wind speeds are measured for 1 minute,5 min,10 min and 15 min the DUP are 9.93×106m3·s−3,8.96×106m3·s−3,8.51×106m3·s−3,8.32×106m3·s−3,×106m3s−3,respectivly；The seasonal variation of DUP was as follows: winter>summer>autumn>spring. By analyzing the diurnal DUP and wind speed in different observation periods, it is found that the main deviation occurs in the morning and evening. If the average wind speed is measured at intervals of 10 minutes or longer,there will be some errors in assessing the damage caused by wind dust activities[89].

3 Conclusion

In the hinterland of the Taklimakan Desert, China Meteorological Administration has established the Tazhong atmospheric environment observation and experimenting station of Urumqi Institute of Desert Meteorology of China Meteorological Administration. It has conducted long-term research on the land-air interaction,the dynamic mechanism of weather and climate formation,the atmospheric environment and the mechanism of surface dust emission and its transportation in the Taklimakan Desert and its surrounding areas, for more than ten years. Systematic achievements have been made in desert climate and environment,desert land-air interaction,aeolian sand physics,etc. However,there are still some problems,such as the imperfection of the system,the lack of revealing and understanding of the problems,the lack of in-depth analysis of all kinds of influencing factors measured in the field,and the insufficient utilization.

In order to meet the increasing needs of national regional development strategy,disaster prevention and mitigation,meteorological services and scientific research, the desert meteorological research requires strengthening the observation and experimenting of land-air interaction and surface sand transportation process, enhancing the application ability of comprehensive observation data,and improving the research of desert climate change characteristics,sand transport law and desert micro meteorological characteristics.At the same time,it is necessary to obtain the key land surface process parameters which is essential to improve the regional numerical prediction model and improve the prediction ability of the model combined with remote sensing technology;It is also necessary to use modern air-space-ground observation methods,combined with big data processing techniques, to conduct in-depth research on the scientific issues such as: the radiative forcing of sand-dust aerosols in desert areas the influencing mechanism of positive and negative feedback effects of the complex coupling system between the desert and the atmosphere.