What caused the unseasonal extreme dust storm in Uzbekistan during November 2021?

An unseasonal dust storm hit large parts of Central Asia on 4–5 November 2021, setting records for the column aerosol burden and fine particulate concentration in Tashkent, Uzbekistan. The dust event originated from an agropastoral region in southern Kazakhstan, where the soil erodibility was enhanced by a prolonged agricultural drought resulting from La Niña-related precipitation deficit and persistent high atmospheric evaporative demand. The dust outbreak was triggered by sustained postfrontal northerly winds during an extreme cold air outbreak. The cold air and dust outbreaks were preceded by a chain of processes consisting of recurrent synoptic-scale transient Rossby wave packets over the North Pacific and North Atlantic, upper-level wave breaking and blocking over Greenland, followed by high-latitude blocking over Northern Europe and West Siberia, and the equatorward shift of a tropopause polar vortex and cold pool into southern Kazakhstan. Our study suggests that the historic dust storm in Uzbekistan was a compound weather event driven by cold extreme, high winds, and drought precondition.


Introduction
On 4-5 November 2021, an extreme dust storm hit large parts of Uzbekistan, Tajikistan, and Turkmenistan, causing property damage, socioeconomic disruption, and a surge of respiratory illnesses.The event has been described in the media as the worst dust storm ever recorded in Uzbekistan.Tashkent, the most populous city of Central Asia, reportedly suffered extremely high concentrations of fine particulates (PM 2.5 ) resulting in an increase of acute respiratory problems and ambulance service calls [1].Poor visibility and hazardous weather also caused automobile accidents and power outages in surrounding regions [1].
Climatologically, dust outbreaks in Central Asia are most common between late spring and early summer due to frequent cold intrusion from high latitudes and sufficiently dry and exposed soils during the early growing season [2,3].The 2021 November dust event was a rare occurrence during the boreal cold season, when dust emission is usually suppressed by seasonally high precipitation and possible early onset of snow cover [4].The event was reportedly triggered by a cold air outbreak (CAO) linked to the Siberian High, which was found to extend abnormally westward to the Caspian Sea from its typical center of action during winter 2021 [5].The CAO reportedly triggered record snowfall and persistent cold extremes across China during 6-8 November 2021 [6,7].In addition, Central Asia suffered severe drought and record high temperatures in 2021, which may have enhanced the soil erodibility and susceptibility to wind erosion [8].
While past studies shed some light on the meteorological aspect of the unseasonal dust storm in Uzbekistan, several key questions remain unanswered, including: (1) how intense was the dust storm from a climatological perspective?(2) What atmospheric processes triggered the cold air and dust outbreaks?and (3) how did the regional hydroclimate contribute to the dust outbreak?This study presents observational evidence of the record-breaking aerosol burden and particulate pollution following the dust outbreak (section 3.1), and investigates the atmospheric dynamics (section 3.2) and hydroclimate preconditions (section 3.3) associated with such an unseasonal extreme dust event in Central Asia.

Data and methods
Two long-term observations are used to evaluate the dust event intensity in Tashkent: Hourly PM 2.5 concentration and Air Quality Index reported from the U.S. Embassy in Tashkent, and daily average coarse-mode aerosol optical depth (AOD 550 nm) derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) level 2 deep-blue and dark-target merged AOD and fine mode fraction [9].Surface synoptic observations and ERA5 reanalysis are used to investigate the precursory atmospheric processes leading up to the dust outbreak.Specifically, we identify two upper-level dynamical features related to persistent high-impact surface weather: atmospheric blocking and recurrent synoptic-scale transient Rossby wave packets.Blocks are defined as regions with persistent negative anomalies of 500-150 hPa vertically averaged potential vorticity (PV) exceeding −1.3 PVU (1 PVU = 10 −6 Km 2 s −1 kg −1 ) and a spatial overlap of at least 70% between successive 6 hourly time steps for at least 5 days [10,11].Recurrent Rossby wave packets can repeatedly pass and amplify at the same longitude in the same phase, resulting in recurring ridging or troughing patterns and persistent cold or hot extremes [12,13].The strength of the transient Rossby wave packets is described by a 'R' metric, which is a timeand wavenumber-filtered signal derived from the Hovmöller diagram of 250 hPa meridional winds averaged over 35 • N-65 • N [14].
In addition, we examine the role of tropopause polar vortices (TPVs) in the CAO development using the TPVTrack method [15].TPVs are subsynopticscale, coherent tropopause-based vortices with typical radii of 100 to 1000 km and lifetimes of days to months, characterized by a local minimum of dynamic tropopause potential temperature, a cyclonic PV anomaly in the Arctic lower stratosphere, and a lowered tropopause often to the 500 hPa level or below [16][17][18].The equatorward advection of TPVs and associated cold pool can trigger intense CAOs in the midlatitudes [19][20][21].Finally, we assess the drought severity and impact on the soil erodibility using three long-term datasets, including the Climate Research Unit Time Series monthly precipitation and potential evapotranspiration (PET) [22], European Space Agency Climate Change Initiative blended passive-active microwave soil moisture product [23,24], and MODIS Climate Modeling Grid global level 3 monthly Normalized Difference Vegetation Index (NDVI) product [25].The microwave soil moisture product represents the top 2 cm of the soil and is thus closely related to the inter-particle cohesion and wind erodibility of top soil [26].

Observed extreme dust burden in Tashkent
According to geostationary satellite observations, the initial dust outbreak began around 03:30Z (08:30 AM) 4 November 2021 from an agropastoral area in the Ordabasy District of southern Kazakhstan (see figure S1 for details).The source area consisted of a mix of semiarid steppe (used as natural pasture) and rainfed croplands near the lower Arys River.Three hours later, MODIS onboard Terra detected a coneshaped dust plume advancing south (figure 1(a)) and sweeping across several highly populated regions the next day (figure 1(b)), including the Fergana Valley, Gissar Valley, and the foothills of Tien Shan-Pamirs mountains.Located 150 km downwind, Tashkent experienced record-breaking particulate concentration and persistent unhealthy air quality following the dust outbreak (figure 1(c)).The U.S. Embassy at Tashkent observed a peak PM 2.5 concentration of 978 µg m −3 on 5 November 2021 (figure 1(c)).It was the highest PM 2.5 ever observed across all U.S. Embassy locations in Central Asia (figure S2).The Centre of Hydrometeorological Service of Uzbekistan (Uzhydromet) reported even higher PM 2.5 levels in excess of 2000 µg m −3 or 30 times the permissible level in Tashkent [27].The extreme dust burden was also observed by the MODIS coarse-mode AOD record.Particularly, the dust storm caused the highest AOD in Tashkent during the boreal cold season (November-April), as well as the second highest annual mean AOD since 2001 (figure 1(d)).

Atmospheric dynamics of cold air and dust outbreaks
On 4 November 2021, surface synoptic observations near the dust source revealed a reversal and rapid rise of sea level pressure, a sudden drop of temperature, and a shift from calm condition to sustained northerly winds (figure 2(a)), which collectively indicated the passage of a cold front.Persistent wind gusts of The atmospheric blocking and CAO mechanisms are further examined using the dynamic tropopause potential temperature maps shown in figure 3. A high-amplitude ridge developed over northeastern Canada on 27 October 2021 (day −8), initiating RWB and blocking near Greenland (B1 in figure 3).A split flow pattern developed around the blocking ridge and the October 2021 nor'easter located equatorward off the Mid-Atlantic coast (W).The meridional PV gradient intensified downstream in association with the superposition of the polar and subtropical jets (circled area in figure 3(a); see figure S3 for cross section along 45 • W), driven by the simultaneous poleward transport of an anticyclonic PV anomaly in the warm sector of the nor'easter along the Gulf Stream and the equatorward incursion of a cyclonic PV anomaly from a TPV over the Lincoln Sea.A deep extratropical cyclone with a central pressure of 963 hPa (L) developed in the left exit region of the superposed jet, likely due to the strengthened ageostrophic transverse circulation in the jet exit and the polar cyclonic PV anomaly [29].The warm sector of the extratropical low featured a poleward ascending flow of low PV air which favored upperlevel ridging and block onset over Northern Europe (B2 in figure 3(b)) [30,31].As the amplifying ridge built into the Barents Sea, the Ural and West Siberia region experienced RWB, blocking onset, and formation of an elongated high PV trough (or PV streamer) downstream (figure 3(c)).The deepening trough was embedded with an equatorward shifted TPV which was situated near the Kara Sea two days earlier (see the TPV trajectory in figure 3(d)).The TPV experienced significant stretching and formed via splitting a new cyclonic coherent tropopause disturbance (CTD) over southern Kazakhstan (figure 3(d)).
The TPV and CTD were dynamically similar, with the difference being that the TPV formed and spent much of its lifetime in polar regions, which allows a tropospheric-deep cold pool to form within and underneath the TPV [18,32].Indeed, figure 3(e) shows that both the TPV and CTD featured a lowered tropopause to around 500 hPa and strong wind circulations surrounding them.The cold air triggering the CAO was associated with the TPV, as indicated by the isentropes bending upward into the TPV as well as the extremely low 1000-500 hPa thickness (figure 3(e)).The CTD formation in southern Kazakhstan, however, facilitated the deep penetration of the TPVassociated cold pool into the midlatitudes.At the ground level, the highest winds developed just behind the surface cold front (figures 3(e) and (f)), and ahead of rapid low-level anticyclogenesis (anticyclonic center denoted as H in figures 3(c) and (d)).Further analysis reveals that the anticyclonic center rose to 1050.6 hPa on 4 November at a rate of 8 hPa d −1 .The explosive anticyclogenesis and attendant sustained postfrontal northerly flows contributed to the intense dust outbreak in southern Kazakhstan and subsequent rapid dispersion to large parts of Central Asia (figure 2(b)).The tropopause-level diagnosis suggests that the cold air and dust outbreaks were associated with blocking development over the Ural and West Siberia regions via a planetary wave train propagating from the North Atlantic.The wave propagation was supported by a strong meridional PV gradient at the Atlantic jet entrance region, but was followed by wave breaking directly downstream and poleward of the jet exit near Iceland (figure 3(a)).The upper-level wave breaking and blocking onset are closely related to changes in the Atlantic jet position, storm track, and leading patterns of the North Atlantic winter climate variability [34][35][36][37].During the prior month of the dust storm (October 2021), frequent blocking accompanied by strong positive 500 hPa geopotential height anomalies occurred in the vicinity of Greenland, and to a lesser extent, over Northern Europe and West Siberia (figure 4(a)).Consistent with the high-latitude blocking, the eddy-driven polar jet shifted well southward into the subtropics over the central and eastern Atlantic (figure 4(a)).Based on the 1951-2020 climatology, extreme blocking conditions occurred near Greenland during October 2021, resulting in the second highest October Greenland Blocking Index (GBI) (2.3) since 1950 (figure 4(c)).The extreme positive GBI and associated equatorward shifted jet position corresponded to an exceptionally negative pattern of the North Atlantic Oscillation (NAO), as indicated by the second lowest October NAO score (−2.0) since 1950 (figure 4(c)).Extreme negative NAO and Greenland blocking favor enhanced meridional airmass transport contributing to more frequent CAOs over Eurasia and North America [38][39][40][41].

Hydroclimate precondition and enhanced soil erodibility
While the CAO and attendant high winds triggered the dust outbreak onset, land surface conditions play an important role in modulating the location and intensity of dust emission, as indicated by the highly localized source activation despite the powerful frontal system (figure 1(a)).Notably, hydroclimate precondition, including soil moisture and vegetation, affects the aeolian sediment availability by increasing the soil inter-particle cohesion, sheltering dry or exposed surfaces, and reducing the near surface wind momentum through drag partition [4,42].Drought is an important precondition for increasing wind erosion in semiarid areas, and has been linked to previous intense dust periods in Central Asia [42,43].
Central Asia suffered widespread precipitation declines in 2021, especially in the high mountains where most of the annual precipitation is received during the cold season and released as snowmelt runoff during the warm season (May-October) (figure 5(a)) [44,45].As the headwater area for the Arys river and surrounding agropastoral regions in southern Kazakhstan, western Tien Shan-Pamirs (boxed area in figure 5 ).In addition to drought, land use intensification from agricultural and pastoral productions may have further enhanced the sediment availability and localized dust emissions, as previously observed in southern Kazakhstan [43].
The recent prolonged drought of Central Asia occurred during a 'triple-dip' La Niña lasting for the past three winters of 2020-2023 (figure 5(e)).This unprecedented event continued the predominant multi-year La Niña and La Niña-like conditions since the turn of the century.A recent study suggested that La Niña events are associated with belowaverage precipitation during the cold season, belowaverage soil moisture and vegetation cover during the following warm season, and consequently aboveaverage dust burden over Central Asia [43].Previous studies also showed an increasing ENSO influence on the hydroclimate in Central Asia since the 1990 s, due to an increasing frequency of Central Pacific La Niña events characterized by anomalously cold central Pacific and warm western Pacific resulting in an enhanced zonal sea surface temperature (SST) gradient across the west Pacific (figure S4) [43,[46][47][48][49][50][51].The seasonally persistent tropical thermal anomalies and zonal SST gradients foster coherent Rossby wave responses over East Africa and Central and Southwest Asia, often as part of a global zonal band of upper level anticyclonic anomalies leading to widespread precipitation reductions across the Northern Hemisphere midlatitudes (figures S5 and S6).The growing oceanic forcing of precipitation modifications over Central Asia has been largely attributed to the rapid warming and expansion of the tropical Indo-Pacific Ocean [52,53].

Discussions and conclusions
The dust storm of 4-5 November 2021 in Uzbekistan, which originated from an agropastoral area of natural pasture and rainfed croplands in southern Kazakhstan, was an unseasonal extreme event with significant impact on regional air quality and socioeconomic activity.Our analysis suggests that this dust storm was a 'preconditioned' compound event caused by an extreme CAO and attendant postfrontal northerly winds (the driver) and a prolonged drought (the precondition) associated with a multi-year La Niña event (the driver of the precondition) [54].The cold air and dust outbreaks were preceded by a succession of planetary-and synoptic-scale processes during the prior week, including recurrent transient synoptic-scale Rossby wave packets over the North Pacific, upper-level wave breaking and blocking near Greenland, southward shifted polar jet and extratropical cyclogenesis over the North Atlantic, and high-latitude blocking over Northern Europe and West Siberia.This type of Atlantic-origin wave train has been previously identified as a primary mechanism of CAOs in Asia [55][56][57].Apart from the planetary waviness, the extreme CAO event was associated with the equatorward advection of an Arctic TPV and cold pools.While this case study is not sufficient to establish a causal link between TPV and CAO, equatorward shifted TPVs have been frequently associated with intense CAO events in mid-and low-latitude regions.For example, 40% of the most intense CAOs in the Fram Strait were associated with Arctic TPVs [19], while TPVs were involved in 85% CAOs in eastern North America [21].
A prolonged agricultural drought-characterized by persistent below-average precipitation and aboveaverage PET-desiccated the drylands of Central Asia, thereby creating a favorable precondition for dust uplifting from the dry, exposed soils.The severe drought was linked to an unprecedented 'triple-dip' La Niña event through teleconnection effect on the wintertime circulation and precipitation in Central Asia.Recent studies suggested that multi-year consecutive La Niña events, such as the recent occurrences of 2010-12, 2016-18, and 2020-23, may become more frequent under greenhouse warming [58].The heightened risk of prolonged drought, combined with rising temperatures and atmospheric evaporative demand, may continue to aggravate the surface water availability, soil erodibility, and dust outbreaks in Central Asia.

X Xi et alFigure 1 .
Figure 1.(a), (b) MODIS/Terra true color images on 4 and 5 November 2021.Black triangles denote the weather station at Arys, Kazakhstan.Open squares denote the geographic domain for computing the regional mean AOD over Tashkent.(c) Hourly PM2.5 and Air Quality Index reported at the U.S. Embassy in Tashkent.Horizontal red lines are the unhealthy for sensitive groups, unhealthy, very unhealthy, and hazardous levels.(d) Daily average coarse-mode AOD and PM2.5 in Tashkent.Observed values on 5 November 2021 are marked (AOD in black square; PM2.5 in red square).

Figure 3 .
Figure 3. (a)-(d) Potential temperature (shading) and winds on the 2 PVU surface (or dynamic tropopause) at 06 UTC of preceding days of the dust storm.Major features are Greenland blocking (B1), Ural blocking (B2), October 2021 nor'easter which eventually became Tropical Storm Wanda (W), jet superposition (black circle), extratropical cyclogenesis (L), rapid anticyclogenesis (H), tropopause polar vortex (TPV), and coherent tropopause disturbance (CTD).The white dotted line in (d) denotes the TPV trajectory at 06 UTC between 30 October and 4 November 2021.(e), (f) Vertical cross sections of wind speed (shading) and potential temperature (gray contours every 5 K) along the A-A' and B-B' black lines in (d).The thick black contour denotes the dynamic tropopause.The thick blue contour denotes the 1000-500 hPa thickness.The approximate locations of surface cold front and TPV-associated cold pool are shown.The weather station at Arys, Kazakhstan is denoted as a black triangle in all panels.
(a) received the lowest precipitation since 1990 (figure 5(e)).Meanwhile, the PET anomaly revealed persistent above-average atmospheric water demand (figure 5(b)), which exacerbated the already limited surface water.As a result, both the soil moisture and NDVI anomalies indicate widespread agricultural drought during the 2021 warm season (figures 5(c) and (d)).The severe drought reportedly caused massive crop failure and livestock deaths in Central Asia [8].A closer look at the dust source region (boxed area in figures 5(c) and (d)) reveals prolonged drought conditions since 2019, which could result in a cumulative risk of desertification and wind erosion (figure 5(e)