The aerosol direct radiative effects (DRE) in the ultraviolet, shortwave, and longwave range due to an unusual dust storm during 21 March 2012 have been quantified from surface measurements of aerosol optical depth (AOD) and radiation fluxes at Delhi, a western Indo Gangetic Plain station. The intrusion of dust over Delhi caused an increase in daily average AOD at 500 nm from 0.6 to 0.8, with a corresponding decrease in Angstrom exponent from 0.4 to subzero value. The dust severely affected the incoming solar radiation flux in the UV, shortwave, and longwave regions. The DRE at surface in the UV and shortwave regions decreased from −4.6 to −5.9 Wm−2 and from −68 to −86 Wm−2, respectively, while the longwave DRE increased from 27 to 45 Wm−2.
 Atmospheric aerosols are now known to affect the energy balance of the Earth-atmosphere system through aerosol-radiation interaction [Twomey1977; Charlson et al., 1987; Haywood et al., 1997; Ackerman et al., 2000; Koren et al., 2004]. Desert mineral dust is the main natural aerosol, on the global scale, that decisively affects the energy balance of climate system [Satheesh and Moorthy, 2005]. There have been several studies on the aerosol radiative forcing highlighting the effects of mineral dust aerosols, but most of these are limited to the shortwave wavelength region only. This is in spite of the fact that aerosol forcings in the longwave region, as well as ultraviolet (UV) region, are also important, and neglecting these forcings would cause large errors in climate models [Satheesh and Lubin, 2003], particularly when there is a significant amount of mineral dust aerosols in the atmosphere. The mineral dust aerosols are also known to be good absorbers in the UV region [Patterson et al., 1977; Sinyuk et al., 2003; Torres et al., 2005]. Studies show that longwave aerosol radiative forcing is even comparable in magnitude to the forcing due to greenhouse gases [Vogelmann et al., 2003] or that with the shortwave forcing, especially during the mineral dust loading in the atmosphere [Tegen et al., 1996].
 The Indo-Gangetic Plain (IGP) is known to have frequent dust storms during the premonsoon period (April–May) that considerably increases the mineral dust aerosol over the region. The impact of these dust aerosols on the shortwave radiation flux has been studied in the past at a few places in the IGP region [Singh et al., 2005; Dey and Tripathi, 2008; Singh et al., 2010]. However, there is hardly any study on the effect of dust on the ultraviolet and longwave radiation flux in the region, in spite of the fact that the tropical aerosols are known to cause a large difference in the top of the atmosphere and surface forcings [Jayaraman et al., 1998; Satheesh and Ramanathan, 2000]. Recently, there was an unusual dust storm (which usually occur during May–June over the northwest Indian region) event that started in the western part of Middle East and engulfed the IGP region reaching Delhi on 21 March 2012. The cause of this intense dust storm seems to be the convergence of two different fronts, front that carried dust from Iraq and Kuwait and the dust laden front from the southeastern Iran. The MODIS (Moderate Resolution Imaging Spectroradiometer) on NASA's Terra satellite image of 19 March showing the spatial sweep of the dust event across Iran, Afghanistan, and Pakistan is shown in Figure 1a. The plumes from the storm were thick enough to completely obstruct the visibility of the land and water surfaces below, and they have traversed thousands of kilometers from the Red Sea to Afghanistan and from the Arabian Peninsula to India. Figure 1b shows the transport of dust plume on 20 March across the Arabian Sea to reach India.
 In order to reconfirm the dust event, the UV aerosol index (AI) map derived from Ozone Monitoring Instrument during 19–23 March along with the 5 day isentropic air mass back trajectories arriving at the measurement site, Delhi, were examined. It clearly evidenced the advection of the dust plume to the measurement location (figure not shown). Further, the cross-correlation analysis of the AI data between different locations, such as Iraq, Iran, Afghanistan, Pakistan, and Delhi, was carried out during the event at zero, 1 day, and 2 day lag. The maximum correlation for AI between the stations Iran/Iraq and Delhi was found in the 2 day lag, implying that the plume was reached from the source (Iran-Iraq region) to the measurement location (Delhi) within 2 days.
 The impact of this “unusual” dust storm has been studied here, not only on the shortwave radiation flux but also on the ultraviolet and longwave fluxes as well. To our knowledge, this is the first ever study in the IGP region where impact of dust storm intrusion on ultraviolet and longwave flux is being reported. These parameters are important for climate studies in order to gauge the diurnal effects of dust aerosols. In this letter, we are also reporting the observed ultraviolet, shortwave, and longwave aerosol direct radiative effect at the surface from Delhi and the effect of dust storm intrusion on the same.
2 Instrumentation and Observations
 The solar flux measurements in the UV (280–400 nm), shortwave (285–2800 nm), and longwave (4500–42000 nm) spectral regions were done at Delhi (28.38°N, 77.10°E, 235 m above mean sea level) using Kipp & Zonen pyranometers model CUV-4, CMP-21, and CGR 4 Pyrgeometer during March 2012. The shortwave radiant flux measurements have an estimated experimental error of 3%, whereas global UV radiant flux measurements have experimental errors <10%. All these instruments are new and have been installed after proper calibration, certified from the company. The flux data were recorded in Wm−2 every 2 min, 24 h a day during the entire observation period particularly during 19–23 March 2012 centering the dust storm event day of 21 March 2012.
 For measuring aerosol optical depth (AOD), Microtops Sunphotometer (Solar Light Company, USA) has been used. The instrument gives column AOD at five wavelengths from ultraviolet to near infrared (340, 500, 675, 870, and 1020 nm) with full width at half maximum of ±2–10 nm. The instrument is periodically calibrated by the Solar Light Company, USA, every year to ensure the stability and reliability of AOD measurements. The Microtops observations were made only during the clear-sky periods. The data were taken at half hour intervals during the daytime (about 09:00–16:00 h) with a minimum of three consecutive observations at a time (within a short span <20 s). Out of these three successive observations, the one with the minimum AOD at 500 nm was used for estimating the daily average values. In addition, the contamination due to the subvisible cirrus has been avoided further by removing outliers lying beyond 2σ level (σ being the standard deviation) from individual day's observations. The maximum uncertainty in the optical depth estimations in each channel is ~ ±0.02 [Morys et al., 2001; Porter et al., 2001; Ichoku et al., 2002, Smirnov et al., 2009]. More details of the instrument and data analysis procedures may be found elsewhere [Srivastava et al., 2006; Lodhi et al., 2013].
 The meteorological parameters (relative humidity, wind speed, wind direction, temperature, and rainfall) have been recorded with the help of meteorological sensors, colocated with the other instruments. These sensors are made by Campbell Scientific Group, Canada. Wind sensors model is 05103 wind monitor, temperature and relative humidity (RH) probe, model HMP 45 C.
2.1 Prevailing Meteorology During Observation
 The temporal variation of the surface meteorological parameters such as temperature (T), relative humidity (RH), wind speed (WS), and wind direction are shown in Figure 2. A sudden shift in the direction of winds from southeasterlies to north/northwesterlies is observed on 20 March, and the pattern remains more or less the same for the rest of the period. In addition, wind speeds have also increased significantly toward 21 March. Further, the values of relative humidity and temperature are found to reduce drastically on 21 March. The reduction was ~ 25% compared to that on the control day (say 19 March). The pattern is more or less similar for 22 March, and thereafter the temperature gradually picks up. The surface meteorological parameters suggest the advection of dry air from the west Asian region. The spatial extent of the advection has been examined from the vertical profiles of RH and potential temperature (θ), derived from the radiosonde ascent data from the New Delhi Safdarjung station (http://weather.uwyo.edu/upperair/sounding.html). The profile of RH on the event day (21 March) shows a very steep variation vertically from ~ 40% at the surface to reach as low as 5% at ~ 3 km and remains constant thereafter, clearly indicating the prevalence of dust laden dry air mass in the entire lower atmosphere at the measurement site.
 The collocated measurements of the global surface flux and aerosol optical depth are used first to obtain the aerosol direct radiative forcing efficiency at the observation site in the three wavelength bands. For the forcing estimations, we have employed the differential method, which assumes that changes in the global irradiance in a spectral band are assumed as due to the changes in aerosol loading. In order to do this, a reference day among the clear-sky days for the whole month (of March 2012) was chosen based on the lowest mean aerosol optical depth at 500 nm. The 24 h average surface fluxes in different bands for the reference day (lowest AOD day) were subtracted correspondingly from the average fluxes for the rest of the days. The aerosol forcing efficiency was then calculated from the slope of the best fit line between daily mean surface flux and the corresponding mean aerosol optical depths. The forcing efficiency when multiplied with the average aerosol optical depth gives rise to the observed direct radiative effects (DRE) at the surface. This method is rather robust over the conventional method using radiative transfer models as this is neither sensitive to calibration uncertainties nor model assumptions [Conant, 2000]. The maximum change in direct flux due to the changes in the eccentricity parameter (Sun-Earth distance) and declination angle is found to be less than 2%. The total uncertainty in DRE estimates in this approach is also small and is reported to be ~ 7% [Srivastava, et al., 2011].
4 Results and Discussion
 The time series of spectral AOD at 500 nm is shown in Figure 3 for the period from 19 to 23 March 2012. The AOD value is significantly higher on 21 March in comparison with other days. The AOD at 500 nm increased from 0.58 to 0.74, whereas the AOD at 1020 nm increased from 0.41 to 0.69 due to the dust intrusion. The particular matter (PM10 and PM2.5) mass concentrations measured during the event day showed a sudden increase of about five to six times (from normal) at several locations in Delhi (A. K. Srivastava, 2013, personal communication). Estimation of the Angstrom coefficients α and β, using the Angstrom relation τ = βλ- α [Ángström, 1964], also clearly demonstrates the prevalence of dust dominance in the atmosphere as evident from the negative value of α and higher value of AOD. The reversal in the spectral dependences leading to negative values of α occurs usually during the periods of severe dust storms in the vicinity of the measurement location [Singh et al., 2004; Singh et al., 2005; Pandithurai et al., 2008; Gautam et al., 2010; 2011; Lodhi et al., 2013]. In the present case, α values decreased from 0.43 (on 19 March) to −0.02 (on 21 March) due to the dust intrusion.
 The time series of the global irradiance at surface observed at three spectral bands viz. ultraviolet, shortwave, and longwave/IR are shown in Figure 4. The downward UV and shortwave radiation was significantly less on the event day in comparison with other days, and an opposite pattern with an enhancement in the flux is observed for the longwave radiation. The peak UV flux reduced from 36 to 25 Wm−2, and the peak shortwave flux reduced from 844 to 714 Wm−2. The 24 h daily average UV and shortwave fluxes reduced from 9 to 7 Wm−2 and 221 to 199 Wm−2, respectively. The longwave radiation decreased after the event subsided during 22 March onward. The average downwelling longwave radiation (DLR) flux reaching the surface was maximum during the event at 385 Wm−2 and it reduced to 318 Wm−2 after the event. The reduction in UV and shortwave fluxes due to dust intrusion has also been reported earlier [Lubin et al., 2002; Antón et al., 2012]. The significant peak observed in the DLR on 21 March, despite the fact that both the temperature and humidity showed a decrease, is possible due to the emissions from dust in longwave region [Slingo et al., 2006]. The sudden fall in DLR soon after the event might be attributed to the fall in atmospheric temperatures and relative humidity (Figure 2) [Prata1996; Cho et al., 2008].
4.1 Surface Aerosol Direct Radiative Effect in UV, Shortwave and Longwave Regions
 The 1 month data obtained during March 2012 had 18 clear-sky days and the measurements of spectral AODs and global irradiance during these days have been used in this study. The irradiance data have been screened for the passing clouds, shown by the sudden dip in the diurnal variation of the (Gaussian shape) irradiance. Among the 1 month clear-sky data, the maximum irradiance at UV and shortwave corresponds to the minimum AOD at 500 nm accompanied with the minimum irradiance at the longwave region corresponding to the lowest AOD at 1020 nm. The diurnal irradiance at that particular day has been taken as the reference diurnal cycle (S0) at all the three bands. A change in the irradiance due to the aerosols (Ar), Ar = <Sr> − <S0> has been computed for each day by subtracting the diurnal average of S0 from the daily mean values of the irradiance (Sr) for each spectral band. The method assumes that the change in flux is forced by the changes in aerosol optical depths. In the present study, the 29 March has been taken as the reference day as the day corresponds to the lowest spectral AODs and maximum shortwave irradiances during the month. The slope of the linear least squares fit between the Ar and AODs yields the direct aerosol forcing efficiencies. These forcing efficiencies were then multiplied with the corresponding AODs to give the observed aerosol direct radiative effect.
 The estimated aerosol DRE at UV, visible, and IR regions during the study period are shown in Figure 5. The vertical lines through the bars represent the estimated standard errors in the forcing estimations, computed from the errors in the estimation of forcing efficiencies (from the scatterplots of average forcing versus AOD) and the uncertainties in the mean AODs. On the event day, the observed shortwave DRE (DRESW) is as low as −86 Wm−2, which is ~ 18 units less than that of the control day, 19 March (−68 Wm−2). The observed surface cooling in shortwave is much higher than those observed earlier during month of March [Pandithurai et al., 2008]. We have also observed a large increase in the longwave DRELW (as high as ~ 45 Wm−2) on 21 March in comparison with control day (~27 Wm−2). Similar values of longwave forcing of the order of 42 Wm−2 have been observed recently during a Saharan dust event at Lampedusa [di Sarra et al. 2011]. The DREUV also shows a change from −4.6 to −5.9 Wm−2 from control day to the dusty (event) day. Thus, we can see a change of 28% in DREUV, 26% in DRESW, and 40% in DRELW due to the intrusion of dust over Delhi during the dust event.
 A large change in DRELW, despite the weak increase in the DLR, signifies the large absorption of IR radiation by the dust aerosols. Examination of the outgoing longwave radiation during the period of study also showed a significant reduction during the event day. This may be due to the large absorption efficiency of the dust aerosols in the longwave region [Deepshikha et al., 2005, 2006; Moorthy et al., 2007]. Further, the longwave DRE due to dust aerosols at Delhi is found to offset the shortwave DRE by about 50%. A somewhat similar offset of shortwave forcing due to longwave was also observed in the Arabian Sea region [Nair et al., 2008]. However, an offset of only 25% was observed at Pune [Panicker et al., 2008], a site devoid of desert dust aerosols.
 The study of an unusual dust storm on 21 March 2012 in the western IGP region at Delhi has revealed marked changes in the AOD, the surface flux, and the direct radiative effects of aerosol at surface in the ultraviolet, short, and long wavelength regions. The results can be summarized as follows:
 The dust intrusion caused an increase in AOD at 500 nm from 0.58 to 0.74 and a subsequent decrease in Angstrom exponent α from 0.43 to −0.02.
 The diurnal averaged downwelling ultraviolet and shortwave flux decreased from 9 to 7 Wm−2 (peak flux decreased from 36 to 25 Wm−2) and 221 to 199 Wm−2 (peak flux decreased from 844 to 714 Wm−2), respectively, whereas the downwelling longwave flux increased from 376 to 385 Wm−2 due to the dust storm.
 It was found that the DREUV decreased by 28% (from −4.6 to −5.9 Wm−2), and DRESW decreased by 26% (from −68 to −86 Wm−2). About a 40% increase in DRELW (from 27 to 45 Wm−2) was also observed at Delhi during the dust event.
 We acknowledge the financial and other help from Council of Scientific and Industrial Research (CSIR) and Indian Space Research Organization—Geosphere Biosphere Programme (ISRO_GBP) for this work. We also thank the anonymous referees for their valuable suggestions and comments.