The Analysis of Pre-Monsoon Dust Storm Over Delhi Using Ground Based Observations

Three major sequential widespread dust events were experienced in the northern parts of India in May 2018. A signicant impact of these pre-monsoon dust storms on the aerosol characteristics over the Indian National capital region (NCR) has been studied using remotely sensed ceilometer and ground-based measurements at Indira Gandhi International (IGI) airport, New Delhi, India. From the results, it is noticed that after each consecutive dust activity, the signicant inclusion of dust aerosols loaded in the Free Troposphere (FT). Consequently, the direct impact on the lower atmospheric parameters like increase in daily average temperature (by 4–5 K), stepped up (stepped down) diurnal cycles of longwave uxes (shortwave uxes) has been recorded within 15 days of dust span. Mainly, the adverse meteorological and radiation features noticed before rst dust storm (DS1), which pinpoints the sudden intrusion of dust over NCR, Delhi. However, this dust storm has extensively impacted in terms of the vertical dust loading, surface boundary layer mechanisms, and socioeconomic way. Therefore, the detailed analysis of vertical dust distribution and its interaction with middle tropospheric processes has been carried by using the vertical normalized attenuated backscatter coecients accompanying the radiosonde observation. The aloft oating dust layer up to 3–4 km has been noticed even after shallow rainfall and persisted at almost the same height for the next 34 hrs due to low-level clouds. Meanwhile, the sub dust layer below 1 km is formed due to local activity, which also sustains for a long time. Moreover, the cumulative losses in terms of the impediment in airline operations (delay and diversion), live causalities, and deaths were estimated at US$1.3 million over these dust period.

Moreover, the adverse impact of dust aerosols loading in the atmosphere has been seen globally, either directly or indirectly, on different aspects. For instance, dust aerosols directly in uence the atmospheric radiative budget by scattering and absorbing radiation uxes including both Shortwave (SW) and Longwave (LW) radiation, and indirectly affecting the cloud microphysics (Twomey et al., 1977;Albrecht et al., 1989;Miller et al., 2004). However, the long term effects of SDS have been noticed on monsoon, hydrological cycle, climate, human health, environment, and socio-economic activities ( The present study investigates the deadly dust storm events using the backscatter product of state of the art ceilometer and extensive surface meteorological observations at the Indira Gandhi International (IGI) airport, New Delhi, India, during May 2018. The capabilities of the Lufft CHM 15k Nimbus ceilometer using high-temporal vertical aerosol backscatter pro les have been analysed and used to describe the dust aerosol activities before, during, and after the dust storm event. The geographical and climatic conditions of the study area and observational datasets are presented in Sect. 2. And, the integral analysis of dust storms and their socioeconomic impacts are comprehensively described in Sect. 3.

Observational Site And Datasets
The Winter Fog Experiment (WiFEX) observational site is located at IGI airport, where this air eld is surrounded by industrial and urban areas and encompasses nearly 2,066 hectares (~ 5000 acres); in New In order to gather the information of prevailing weather parameters across such extreme events, a wide range of instruments, including a ceilometer, 20 m ux tower (equipped with multi-meteorological sensors), and PM sampler, has been operationalized during pre-monsoon dust activity occurred within a period of 15 days in May 2018. Here, entire observational datasets have been represented in the Indian Local Standard Time (LST: UTC + 5:30). The Lufft CHM 15k Nimbus ceilometer (1064 nm) installed on the roof (7 m Above Ground Level (AGL)) of the observatory instruments lab, has recorded Attenuated Backscatter Coe cient (ABC). Additionally, the diurnal variations of meteorological parameters like wind vector, air temperature, relative humidity, and precipitation have been recorded with the help of the multicomponent weather sensor WXT 520 (Vaisala Oyj). Also, the Net radiometer at 2 m height from the ground is collocated with other instruments used to record Radiation uxes (longwave and shortwave radiation). Throughout this study, 1 min high temporal resolution meteorological parameters along with these radiative uxes have been analyzed by taking an hourly average. Moreover, the particulate matters (PM 10 and PM 2.5 ) over the study site (nearly 0.8 km away from the ceilometer) have been monitored by the Central Pollution Control Board (CPCB) using the PM sampler. This 60 min averaged PM concentration followed the beta attenuation method. The planetary boundary layer height (PBLH) and vertical thermodynamic structure were derived at two distinct time intervals by employing radiosonde ascent data (http://weather.uwyo.edu/upperair/sounding.html) from the Ayanagar, New Delhi (nearly 10 km far from the observational site).

Methodology
The raw datasets i. images but it has deteriorated the air quality of urban and rural areas drastically. In order to recon rm these dust events, hourly averaged diurnal variation of particulate matters along with the in uential surface meteorological drivers have been examined, which is represented in Fig. 3. Frequently, the PBLH extend up to 2 − 3 km over the northern part of the Indian subcontinent during the summertime period as a result of intense solar heating (Patil et al., 2013). Nevertheless, Fig. 3(a) clearly demonstrate, the uplift and vertical extent of dust mainly in uenced by the surface meteorology within the atmospheric boundary layer which could stretch PBLH above 3km. Moreover, the diurnal variation of PBL is the response of radiation, winds and energy balance. In the early mornings (05:30 LST), the stable boundary layers at calm wind condition persisted below 500 m AGL. In contrast, convective boundary layer heights around late afternoons (17:30 LST) ranged between 1000 m AGL and 3800 m AGL, pointing toward colossal vertical mixing. Meanwhile, prior to each dust storms, a dry atmospheric agreement lower atmosphere has been noticed which is conclude from 20 − 30 % reduction in RH values during maximum temperature range (Fig. 3(b)). On the other hand, convective instability is built up during each dust events through an increase in low level moisture (by 30 − 40 %) corresponding to rapid fall down in temperature (about 10 K) in 6 hours of interval. In addition, a sudden shift in wind vector shown in Fig. 3(c), has conferred to vertical dust mixing/turbulent activities.
However, these suddenly impacted dust storms directly affected the radiative uxes which could be clearly analysed by comparing it's diurnal variations obtained a day prior to dust storm (Solid line), during the dust storm (Symbolic line), and the day after the dust storm (Dash line) reported in Fig. 4. Generally, in most of the dust storm cases worldwide, it has been observed that longwave ux (includes both Incoming and Outgoing) values one day before (one day after) dust activity are high (less) as noticed in DS2 and DS3 event depicted in Fig. 4(a) and Fig. 4(b). Interestingly, the opposite result seems in the case of DS1, where dust day shows trivial raised in longwave uxes compared to the former day. Concisely, the sudden subsidence of non-local dust over NCR, Delhi without a prior day warming, could be a reason for that. The gradually increased diurnal values of longwave ux before each dust episodes show additional radiative energy has caused a warming of the lower atmosphere. Afterwards, the Incoming longwave ux; signi cantly decrease from its peak within an hour during DS1 (from 468 to 439 W m − 2 ), DS2 (from 470 to 448 W m − 2 ), and DS3 (from 495 to 450 W m − 2 ). Additionally, such rapid fall in the percentage of Outgoing longwave ux (6%, 5%, and 10%) along with minor cut off in shortwave uxes (includes both Incoming and Re ected) due to these huge dust intrusion in a short time could yield the cool next hours/days relative to preceding warmer hours/days. The increased daily mean of surface temperature (by 4-5 K within 15 days dust span), stepped up (stepped down) diurnal cycles of longwave uxes (shortwave uxes) from DS1 to DS3 are utter pinpoints that after each consecutive dust activities, the inclusion of dust aerosols has loaded in FT.
However, in order to reveal these meteorological and radiative features associated with dust activities in pictorial form, the ceilometer backscatter pro les have been combined and analysed for the same dust span. Figure  , and RF4 (6.7 mm) at IGI airport has been recorded. Particularly, the rainfalls after DS1 and DS3 event are associated with WDs which bring ample amount of moisture and maintain low level clouds, especially over NCR which are depicted in Fig. 2(a), Fig. 2(c) and Fig. 5. Additionally, it has been noticed in ceilometer backscatter pro le that the RE4 after DS3 was relatively strong enough to washout oating dust and pollution in the atmosphere which was speci cally persisted in FT after DS1. By considering comprehensive facts of all dust episodes in terms of socioeconomic impact, prevailing meteorology, and direct and indirect repercussion of radiative uxes, the detailed analysis of DS1 has been carried out.
Vertical dust distribution and its peculiarities -Before Dust Storm (DS1) Time-height cross section of ceilometer backscatter pro le and surface meteorological observations at IGI airport illustrated in Fig. 6 for thorough analysis of DS1. In addition, Table 1demonstrates the variations in half-hourly averaged ABC values within two vertical heights (0-0.5 km and 0.5-1 km) corresponding to prior, during, and after the dust storms and posterior to rain events, respectively. Similarly, for the same time variations in half-hourly averaged vertical attenuated backscatter pro les for the DS1 event represented in Fig. 7.
The residual aerosol layer between 2-3 km height and low aerosol concentrations in FT during a stable night period has represented in Fig. 6(a). However, the aerosol backscatter signals, wind speed, and the temperature gradually increased in morning hours (06:00 LST) of DS1 event. During this cloud-free daytime, the maximum hourly averaged temperature of 36.37℃ is recorded 1 hr before the DS1 (between 16:00 LST to 17:00 LST), indicates the local convection. Therefore, the depth of aerosols layers nearly 1-2 km in the daytime period, is mainly in uenced by local convective instability and turbulent venture. This turbulence in the mixed layer is convectively driven.
Moreover, in Fig. 6, the vertical ceilometer backscatter pro le shows the turbulent activity is more dominant within 1 km before the dust storm event. From Table 1 and Fig. 7, it has also been noticed that the half-hourly averaged ABC value from 16:10-16:40 LST (20 min before DS1) within 0.5 km of the boundary layer (0.73 × 10 − 6 m − 1 sr − 1 ) recorded slightly less than ABC value (0.80 × 10 − 6 m − 1 sr − 1 ) within 0.5-1 km. During the clear day, unstable conditions might have created strong winds and vertical convection, which created the vertical aerosol mixing. However, these unstable conditions and convective eddies during the daytime period cause non-local transport above the mixed layer. The gradually increasing scattered vertical backscatter pro le (black line) after 2 km also shows the non-local dust transport over the IGI airport.  represent intense dust activity within 0.5 km from 17:00 LST to 17:30 LST due to the convective activity.
However, the small ABC value (0.45 × 10 − 6 m − 1 sr − 1 ) between 0.5 − 1 km heights represents the shallow dust layer during the same period. From Fig. 7, the vertical backscatter pro les difference between prior (black line) and during (red line) DS1 demonstrates positive values, especially above 0.5 km. Also that, sharply decreasing vertical backscatter pro les (shown in Fig. 7 by a red line) comparing with increasing vertical relative humidity (shown in Fig. 8(a) by a blue line) up to 1 − 1.5 km during DS1 indicates frequently wet deposition of the upper-level dust in the lower level.
The half-hourly average ABC value within a height of 0.5 km increased by about 7.3 times over 1 min from 17:00 LST to 17:01 LST. In this case, the ABC values within 0.5 km height suddenly increased over a few min of intervals. Also, the sharp dropdown in ABC within 0.5-1 km height shows the sudden impact of developed dust storms at IGI airport, Delhi. However, this dust storm is developed somewhere before reaching to the observational site. A similar result was noticed by Kawai  Meanwhile, the oating dust layer reached the FT and mixed with clouds. From 10 km ceilometer backscatter pro le (not given here), it is observed that cloud base height decreased from 8 km at 17:40 LST to 2 km at 19:00 LST. Later, two scatter rainfall events viz. RE1 and RE2 were recorded after the dust storm DS1. The RE1 and RE2 occurred within intervals of 2 hrs and 34 hrs from the dust storm period. These two scatter rainfall results from the western disturbances, oating dust layer and its interaction with cloud, which is clearly depicted in Fig. 2  Vertical dust distribution and its peculiarities -After Rain In a period of 03:30 hrs (between red and blue pro les), the ABC values underneath of 0.5 km sharply reduced by 20 times to dust time ABC values (5.63 × 10 − 6 m − 1 sr − 1 to 0.28 × 10 − 6 m − 1 sr − 1 ).
After the scatter rainfall RE1, the ABC values (blue line, less than 0.5 × 10 − 6 m − 1 sr − 1 ) became lower than dust event values (red line) below 1 km height. Meanwhile, the primary pollutant namely, PM 10 (PM 2.5 ) mass concentrations recorded 108 µg m − 3 (17 µg m − 3 ) between 20:00 LST and 21:00 LST shows a drastic improvement in air quality. Surprisingly, raised ABC values in blue pro le above 1 km represents that oating dust load was still remained above 1 km of lower atmosphere. However, the oating dust around the height of 3-4 km AGL was persisted almost for 34 hrs from the time of DS1 depicted in Fig. 5. The horizontal spike (higher ABC values) in blue pro le above 4 resulted from high response of ABC to low level moist clouds which cross veri ed from cloud base height of ceilometer product. Therefore, the presence moist low level clouds trapped oating dust underneath of 4 km height and avoided to extend it into the FT. The secondary shallow dust layer formed underneath of oating dust layer correspond to different dust source after 9 hrs intervals of RE1. This lower sub dust layer within 1 km is generated from the local dust pollutants, while the overhead oating dust layer within 4 km was residual part of the dust storm DS1.
From the radiosonde pro le at 05:30 LST on 03 May 2018 shown in Fig. 8(b), the number of interesting facts has been noticed. In particular, (1) the strong inversion (ground level values missing up to 200 m) of 6.8 ℃ under the potential temperature gradient of 19.67 K km − 1 (11.8 K in a height range of 0.6 km) in calm wind condition indicates diminutive convection (high atmospheric static stability) and feeble thermal turbulence within 0.8 km (close to 900 mb), (2) reducing values of relative humidity in vertical pro le (from 68 % to 24 %) under the inversion layer depth perfectly signifying the dry warmer air is held above cold surface, (3) Sharpe turn in temperature gradient and relative humidity pro le above the inversion layer, trapped dust aerosols within 1 km which exactly depicted in Fig. 5 as a sub dust layer, and (4) the inclusion of the moist air advection underneath of 4 km height have been noticed from the vertical pro les of RH (increased around 60 % above inversion layer) and potential temperature. However, very steep variation vertically in ABC (blue line in Fig. 7) from 4 km AGL to 5.5 km AGL could result of low cloud. Ultimately, it cloud be acknowledge that oating dust layers (sub dust layer) has been trapped under low-level clouds (inversion layer) and persisted at almost the same height for 34 hrs (23 hrs).

Conclusions
Severe dust storm analysis on 02 May 2018, has been carried to monitor, evaluate, and describe the aerosols activity before, during, and after a dust storm. This dust storm was a particular case because it observed that the storm had already developed before reaching IGI airport, shown in the ceilometer measurements and other ground-based observation data. From the perspective of disasters, a dust storm's sudden appearance is more dangerous than the gradually developed dust storm because of the minimal time for evacuation without forecast information. It is also noticed that after a dust storm, the upper dust layer remained at almost the 4 km height for 34 hrs. However, the sub dust layer was developed in the oating dust layer after 9 hrs of the dust storm, remained at the almost same height.
Overall, it suggested that the oating dust layer prevented from reaching the FT due to low-level clouds, and the lower layer was unable to extend above due to surface inversion.
During each dust storm, a signi cant increase in PM concentration observed. However, on 02 May 2018 dust storm, PM 10 concentration suddenly increased (4 times over 1 hr), while other days, dust storms showed gradually increment. The convective and unstable conditions followed by strong wind speed (> 8 m s − 1 ) represented a positive correlation with each dust storm period. After dust storms, the low-level clouds did a feeble exchange of oating dust between the boundary layer and the free troposphere. This oating dust interacts with clouds and precipitates under favourable conditions. However, the partial washout of oating dust following light rain. After the severe dust storm on 02 May 2018, the residual dust layer persisted for several days within 4 km of height above ground level.
This study is the rst report of this type of vertical distribution of dust in the dust affected area based on observation data. In a detailed analysis of aerosol-cloud interaction from the numerical simulations during the dust storm events are suggested in future studies.