A dust event in the Mongolian Gobi Desert in spring 2016: Observations and vertical dust flux estimation

. A meteorological and dust monitoring tower with 20 m height set up at a Nomgon site in the Mongolian Gobi Deserts in 2010. The Nomgon monitoring tower equipped with wind speed sensors at 2, 4, 10 and 20 m height above the ground level (AGL), a wind direction sensor at 10 m height, a sonic anemometer to measure turbulent momentum flux at 8 m height and a soil moisture sensor at 5 cm depth. A dust storm was expected in the Mongolian Gobi from 30 April to 1 May 2016 and two aerosol monitors (Dust-Trak) were set at 0.9 and 2.95 m heights in the tower for measuring PM 10 . Above mentioned data from the tower during an intensive observation period (IOP) of the dust event are used in this study. In association with a surface cyclone, its frontal system and a trough aloft, dust concentrations of PM 10 increased during the dust storm period due to raised wind speed in the dry conditions of air and soil. The present study aims to estimate friction velocity ( ) and vertical dust flux ( F ) around the Nomgon site during the dust storm period. The observations and estimation results are presented.


Introduction
Large amount of dust aerosol lifts up in the air due to windblown dust emission and severe dust storm events [1,2,3]. Severe dust storm events make threats to human lives and cause considerable economic damages [4,5,6]. High concentrations of particles with diameters of less than 10 micrometers (PM 10 ) can result in negative impacts on human respiratory health [7]. Fine dust particles with sizes smaller than 20 μm suspended in the atmosphere can be transported faraway from source region and be deposited in downwind areas [8,9,10].
Windblown dust occurs where soil particles move by wind force. This process involves the dust cycle in the atmosphere. The dust cycle consists of three major physical mechanisms such as dust emission, transport and deposition [6,11]. Dust emission takes place only for wind (U) and friction velocities higher than their threshold values [12]. As * u is one of the most important parameters in dust emission studies [6]. * u is not the speed of the flow but it is a descriptor of turbulence intensity (or friction) in the surface layer [6]. Friction velocity is governed by the apparent roughness of the surface and atmospheric convection or stability [13,14]. Turbulent motion, wind speed friction velocities play a main role in parameterizations of dust emission in models [15,16,17,18,19]. Gradient measurements of wind speeds and turbulent momentum flux at the meteorological and dust monitoring tower with 20-m height at the Nomgon site in the desert steppe (Gobi) area of Mongolia are a main data of this study. In addition, we aim to measure PM 10 dust concentrations at two levels during an intensive observation period (IOP) of a dust storm. Based on these measurements, friction velocity and vertical dust emission flux around the Nomgon site are attempted to be estimated during the dust storm period.

Meteorological data
A meteorological and dust monitoring tower with a height of 20 m was established at the Nomgon site (42º50ʹ27.26ʺN, 105º7ʹ52.53ʺE) in 2010 (Fig. 1). The tower gradient measurement data including air temperature and relative humidity at 2 m level, wind speeds at 2, 4 and 10 m levels and wind components ( ) by a sonic anemometer at 8 m level, soil moisture 5 cm depth were used in this study (Table 1).

Soil and vegetation condition around the site
The Nomgon site area is located in the Desert steppe (Gobi) regions at the end of the Altai Mountain. The environmental land surface condition around the site is characterized by flat valley between hills, homogenous terrain with scarce vegetation.
Observations on soil and vegetation conditions are made around the Nomgon site in 30 April 2016 (Fig. 2).
The site is open to north (N) and north-west (NW) directions. We measured vegetation density in a valley from the site to N and NW. The measurement field is located in 3 km distance from the site. Vegetation density was scarce around the field and in the valley. Gravel covers the flat surface and loose fine dust is under gravel. Anabasis brevifolia was sparsely growing in the valley and average distance between roots of Anabasis brevifolia was 1-5 m.

Field measurement data
Before dust events, we made zero calibrations for the two instruments. For this purpose, we put two instruments at 0.9 and 2.95 m levels and measured PM 10 concentrations, simultaneously. Statistical methods are used for data analysis.
Measurements of PM 10 concentrations at 0.9 and 2.95 m levels were conducted over a two-day period at the Nomgon site during the dust storm period from 30 April to 1 May 2016 using two aerosol monitoring instruments (Dust Trak-8533). These PM 10 data are used in this study.

Friction velocity
Friction velocity ( * u ) is expressed as: ( 1) where, * u is friction velocity ( ) and it is constant in the surface layer or in the constant flux layer, -turbulent wind speed components along x, y, z axis (at 8 m level), respectively, -covariance between , -covariance between .

Vertical dust flux
Vertical dust flux for PM 10 concentration can be determined using following equation [8,9,10,15,20,21,22]: (2) where, Fvertical dust flux (mg m -2 s -1 ), difference between PM 10 concentrations measured at various heights ( ), difference between heights (m), friction velocity ( ), the vertical eddy diffusivity at z height it is expressed as: where, kvon Karman constant, it characterizes the dimensionless wind shear for statically neutral conditions.zero plane displacement (m), is the non-dimensional vertical profile function for heat defined as for the unstable stratification, for the stable stratification,is the Monin-Obukhov length scale that is formulated as: (4) where, is the mean air temperature ( ), is the gravitational acceleration and the kinematic heat flux.
Vertical dust flux is proportional to a power function of friction velocity [16,23] and it is as: (5) where, Fvertical dust flux ( ), a is coefficient, * u is friction velocity ( ), n is the power of function, it is between 3 and 7 [23].

Data analysis
For calibration instruments, we set up two Dust Trak at same levels (0.9 and 2.95 m) in the tower, and measured concentration of PM 1 PM 2.5 PM 4 and PM 10 during one hour for each two levels with 1-minute interval. The Dust Trak1 (DT1) was at 0.9 m level, while the Dust Trak2 (DT2) was at 2.95 m level. A linear regression equation (Eq. 6) between PM 10 concentrations at 2.95 m level by two DTs was found as: where, Y -PM 10 concentration values by the DT2, х -PM 10 concentration values by the DT1. We corrected as PM 10 concentration by the DT2 at 2.95 m level using Eq. 6. The corrected PM 10 concentration is used for next calculation.

Meteorological characteristics and dust concentration of PM 10
The dust event occurred in 30 April to 1 May in association with low-pressure system and its atmospheric cold front passed across the Gobi Deserts including the site. There was no precipitation over the Gobi Desert region in April 2016 [24]. Moreover, measured soil moisture was 7.7-8.8% and relative humidity varied from 13.4 to 40.0% at the site during the IOP. Thus, soil and air were dry around the site before and during the dust event. The wind speed increased reaching from 11 to 17 . Visibility was around 20 km in the morning of 1 May.

Estimation of friction velocity and the vertical eddy diffusivity
Friction velocity is computed using Eq. 1 that is based on turbulent wind speed components ( ) at 8 m height and their 10-min averaged covariance ( and ). Friction velocity varied from 0.0114 to 0.7175 from 12:00 LST on 30 April to 08:00 LST on 1 May 2016 and rose from 0.2043 to 1.1322 in the morning of 1 May (Fig. 4). In the Desert and desert steppe regions (the Gobi) of the country in spring season, Normalized Difference Vegetation Index (NDVI) is less than 0.24 [3,25]. As a result of less and scarce vegetation condition around the site (Fig. 2), zero plane displacement in Eq. 3 was neglected and the vertical eddy diffusivity was calculated at 10 m height using Eq. 3.

Vertical dust flux estimation
In horizontally-homogeneous boundary layers, flow properties vary only in the vertical direction [6]. We suppose the dust-concentration field is horizontally homogeneous. Vertical dust flux ( ) for PM 10 dust concentration is estimated with the gradient method using Eq. 2 using data from the gradient measurements of PM 10 at 0.9 and 2.95 m levels.
Vertical dust flux ( ) for PM 10 (Fig. 5). According to values of the parameters from the best-fitting, when in Eq. 5, coefficient of determination ( ) was 0.567563 and regression coefficient, a, was as 0.0084395. The relationship equation between friction velocity ( ) and vertical dust flux ( ) was defined as: .

Conclusions
The main results of this study are summarized as follows: The dust event occurred from 30 April to 1 May 2016 in association with passage of a low-pressure system and its atmospheric frontal activities over Mongolia. Data analysis of this study showed that air and soil were dry during the dust storm period. Relative humidity of air and soil moisture at 5 cm depth was 13.4-21.0% and 7.7-8.8%, respectively. Moreover, wind speed and PM 10 dust concentration reached to [15][16][17] and 88-338 , respectively. The vertical dust emission flux is estimated by using the gradient method with dust concentrations measured at two levels and vertical eddy diffusivity evaluated by the turbulent fluxes at 8-m height during the IOP. Moreover, the estimation was calculated for the unstable stratification. The vertical dust emission flux is closely related to the friction velocity . In the morning of 1 May 2016, estimated vertical dust flux ( ) for PM 10 dust concentration rose from 7.2062 to 47.0105 and evaluated friction velocity was high as 0.88- 1.06 in that time. Our estimated vertical dust flux values were comparable with other study results by [21,28,29].