• Open Access

Micropulse lidar-derived aerosol optical depth climatology at ARM sites worldwide

Authors


Abstract

[1] This paper focuses on climatology of the vertical distribution of aerosol optical depth (AOD (z)) from micropulse lidar (MPL) observations for climatically different locations worldwide. For this, a large data set obtained by MPL systems operating at 532 nm during the 4 year period 2007–2010 was used to derive vertical profiles of AOD (z) by combining the corresponding AOD data as an input from an independent measurement using nearly colocated multifilter rotating shadowband radiometer (MFRSR) systems at five different U.S. Department of Energy Atmospheric Radiation Measurement (ARM) Program sites—three permanent sites (SGP in north-central Oklahoma, at 36.6°N, 97.5°W, 320 m; TWP-Darwin in the tropical western Pacific, at 12.4°S, 130.9°E, 30 m; and NSA at Barrow on the North Slope of Alaska, at 71.3°N, 156.6°W, 8 m) and two mobile facility sites (GRW at Graciosa Island in the Azores, at 39°N, 28°W, 15 m; and FKB in the Black Forest of Germany, at 48.5°N, 8.4°E, 511 m). Therefore, amount of data used in this study is constrained by the availability of the MFRSR data. The MPL raw data were averaged for 30 s in time and 30 m in altitude. The diurnally averaged AOD (z) profiles from 4 years were combined to obtain a multiyear vertical profile of AOD (z) climatology at various ARM sites, including diurnal, day-to-day, and seasonal variabilities. Most aerosols were found to be confined to 0–2 km (approximately the planetary boundary layer region) at all sites; however, all sites exhibited measurable aerosols well above the mixed layer, with different height maxima. The entire data set demonstrates large day-to-day variability at all sites. However, there is no significant diurnal variation in AOD (z) at all sites. Significant interannual variability was observed at the SGP site. Clear seasonal variations in AOD (z) profiles exist for all five sites, but seasonal behavior was distinct. Moreover, the different seasonal variability for the lower level (0 to ~2 km) versus the level above indicates a contribution of different types of air masses from different sources. The lower annual mean AOD (z) values (0.093 ± 0.033 for daytime and 0.093 ± 0.05 for nighttime) observed near the surface at GRW are not unexpected for maritime aerosols (mostly sea salt), and the corresponding higher values at SGP (0.118 ± 0.038 for daytime and 0.11 ± 0.05 for nighttime), FKB (0.124 ± 0.042 for daytime and 0.127 ± 0.047 for nighttime), and TWP (0.13 ± 0.078 for daytime and 0.14 ± 0.073 for nighttime) are usual for continental aerosols. The annual mean AOD (z) values observed near the surface during daytime and nighttime for NSA were 0.1 ± 0.042 and 0.09 ± 0.037, respectively. These results will aid the scientific community in understanding aerosol properties and boundary layer dynamics and in improving the incorporation of aerosol radiative effects into global climate models.

1 Introduction

[2] Atmospheric aerosols, solid and liquid particles suspended in the air from natural or human sources, are one of the most variable components of Earth's atmospheric environment [Box et al., 2002]. They influence the radiative budget of the Earth directly by scattering and absorbing solar radiation [Charlson et al., 1991; Haywood et al., 1997; Ramanathan et al., 2001] and indirectly by serving as cloud condensation nuclei [Coulier, 1875; Aitken, 1880], possibly modifying cloud albedo and lifetime [Twomey, 1977; Twomey et al., 1984; Albrecht, 1989; O'Dowd et al., 1999; Spurny, 2000; Lohmann and Feichter, 2005]. Aerosol optical depth is a quantitative measure of the extinction of solar radiation between the point of observation and the top of the atmosphere and is hence a proxy for the integrated columnar aerosol particle.

[3] Atmospheric aerosol studies, involving radiative forcing (RF) analysis, aerosol-cloud interactions, and global aerosol modeling, require accurate information on aerosol optical depth [e.g., Haywood et al., 1997; Kaufman et al., 1997; Seinfeld and Pandis, 1998; Chin et al., 2002; Satheesh, 2002; Kondratyev et al., 2006]. The uncertainties associated with aerosol RF are larger than those associated with any of the other major components of RF that affect climate change [Solomon et al., 2007]. Variable physical and optical properties, relatively short atmospheric lifetimes, and large spatial and temporal variability complicate efforts to account for RF impacts of aerosols in climate models [Rogers et al., 2009]. Therefore, understanding the temporal and spatial variability of aerosol optical properties from varying climate regimes is important for relating aerosols to their sources, quantifying the effects of transport and transformation on the aerosols, and better understanding their contribution to RF and climate change. The vertical distribution of aerosol particles in the atmosphere, one of the key components directly affecting the RF [e.g., Haywood et al., 1997; Franke et al., 2003], is by meteorological conditions such as relative humidity, wind speed and direction, and turbulence [Pilinis et al., 1995; Smirnov et al., 1995].

[4] Multifilter rotating shadowband radiometer (MFRSR), Sun photometer, and other passive measurements of solar transmission provide some information on the vertically integrated physical and optical properties of aerosols [e.g., Shaw, 1983; Harrison and Michalsky, 1994; Harrison et al., 1994; Dubovik and King, 2000; Smirnov et al., 2000; Michalsky et al., 2001; Alexandrov et al., 2004a, 2004b; Michalsky et al., 2010]. But they provide no vertical distribution information. Active-based lidars can effectively and continuously profile the vertical variability of aerosol particles in the atmosphere with high spatial and temporal resolution [Measures, 1984]. Thus, they have proven popular field instruments [Sassen, 1994; Eloranta et al., 2000; Hair et al., 2001; Turner et al., 2001, 2002; Voss et al., 2001; Chazette, 2003; Ferrare et al., 2006; Welton et al., 2006; Rogers et al., 2009; Andrews et al., 2011]. However, aerosol lidar studies covering climatically diverse regimes are still scarce, especially with respect to quantitative discussions of the vertical distribution of aerosol loading.

[5] One goal of the U.S. Department of Energy Atmospheric Radiation Measurement (ARM) Program is to collect extensive data sets from climatically different regimes worldwide that can be used to study, and hence improve, the treatment of radiative transfer in the atmosphere, particularly with respect to aerosols and clouds [Stokes and Schwartz, 1994]. ARM has established five permanent research sites (SGP in north-central Oklahoma at 36.6°N, 97.5°W, 320 m; three sites in the tropical western Pacific (TWP-Manus at 2°S, 147.4°E, 4 m; TWP-Nauru at 0.5°S, 166.9°E, 7 m; and TWP-Darwin at 12.4°S, 130.9°E, 30 m); and NSA at Barrow on the North Slope of Alaska, at 71.3°N, 156.6°W, 8 m), plus two mobile facilities (AMF1 and AMF2) on land and ocean locations throughout the world (Figure 1). These sites are equipped with diverse arrays of passive and active remote sensing, as well as in situ instrumentation [e.g., Mather et al., 1998]. Details on the instrumentation at each site are online at http://www.arm.gov/instruments/. The micropulse lidar (MPL) [Spinhirne, 1993; Spinhirne et al., 1995], operating at 532 nm, is an integral part of the ARM active remote sensing instrument suite at each site.

Figure 1.

Locations of ARM fixed sites and mobile facility deployments.

[6] In this paper, we present a range-resolved aerosol optical depth (AOD (z)) climatology for data collected from MPL systems at three ARM sites (SGP, NSA, and TWP-Darwin) and mobile facility sites including Graciosa Island in the Azores (GRW; 39°N, 28°W, 15 m) and the Black Forest of Germany (FKB; 48.5°N, 8.4°E, 511 m) during cloudless periods. The data used in this study span a 4 year period from 2007 to 2010. We examine vertical profiles of AOD (z) from hourly observations, as well as the multiyear vertical profile of AOD (z) climatology, including diurnal, day-to-day, and seasonal variabilities. The measurements and analysis method are presented in section 2, results and discussion in section 3, and summary and conclusions in section 4.

2 Measurements and Analysis Method

[7] To the limit of signal attenuation, the continuously running and autonomous MPL system collects vertical profiles of scattering from optically thin clouds and aerosols and molecules. Polarized MPL systems have been operating regularly since late 2006, giving rise to an extensive database of both daytime and nighttime backscatter profiles at the ARM sites described in section 1. The MPL system uses a coaxial “transceiver” design with a telescope shared by both transmission and receiving optics. The system achieves the American National Standards Institute eye-safe standard by using low output energies (≤10 μJ) and beam expansion to ~20 cm diameter, combined with high pulse rate (2500 Hz). At standard ARM temporal and spatial resolution (i.e., 30 s and 30 m) settings for data collection, the system is capable of measuring backscattered signals up to an altitude of 15–25 km with reasonably high signal to noise [e.g., Campbell et al., 2008]. High sensitivity is achieved through the use of a diode-pumped, frequency-doubled, pulsed solid-state laser (Nd:YAG at 532 nm) with narrow field of view (~100 µrad), narrow interference filters (~0.3 nm full width at half-maximum intensity), and avalanche photodiode photon counting detection. More extensive descriptions of this system are given by Campbell et al. [2002], Welton et al. [2002], and Flynn et al. [2007].

[8] In total, approximately 1460 days of continuous observations were collected during the 4 year period at SGP, TWP-Darwin, and NSA, whereas 580 days of observations were collected during a 2 year period at GRW, and 250 days were collected during a 1 year period at FKB. Of these days, 280 at SGP, 281 at TWP-Darwin, 48 at NSA, 10 at GRW, and 15 at FKB were found as cloudless periods and have available corresponding data from nearby MFRSR measurements. Hence, these are the observations used in this study (described below).

[9] The raw data used in this study, obtained from the ARM data archive center, are in the form of photoelectron count rate profiles. The observed raw photoelectron count rate represents return signals from molecules, aerosols, and clouds, background noise (detector noise, moonlight, starlight, airglow, scattered city lights during nighttime, and sunlight during daytime), and instrument noise, such as detector afterpulsing. The background signal, derived from data averaged between 45 and 55 km, is subtracted from the detector dead-time-corrected raw signal at each altitude, z. The remaining signal is normalized by multiplying by the square of the range from the laser and dividing by pulse output energy. The signal is also corrected for afterpulse (i.e., the result of cross talk between the laser pulse and the detector) and overlap effects by using a technique described in Campbell et al. [2002] to produce the normalized relative backscatter (NRB) signal, PNRB (z), in units of [counts km2(µj µs) −1], given by

display math(1)

where C is the MPL system constant, β (z) is the backscatter coefficient, σ (z) is the extinction coefficient, the R subscript denotes a Rayleigh (molecular) scattering quantity, and the P subscript denotes a particle (aerosols, clouds) quantity. The signal between the surface and 180 m is often saturated as the system recovers from the transmit pulse; thus, the atmosphere below 180 m is assumed to be well mixed, and σ and β are held constant through this layer. The PNRB (z) profiles are further averaged over hourly intervals.

[10] Aerosol backscatter coefficient profiles, β (z), are derived using the cloud-free PNRB (z) profiles and the method of Fernald [1984] as

display math(2)

where zc is the maximum altitude where measurable aerosols are present, S is aerosol backscatter-to-extinction ratio, a = SR/S, and SR is the molecular backscatter-to-extinction ratio (3/8π sr−1). The system constant C is determined by using the PNRB (z) signal, the corresponding column integrated aerosol optical depth (AOD) values derived from the nearly colocated MFRSR measurement at 500 nm, and the modeled molecular scattering coefficient, βR, in the cloud-free region above the aerosol layer. The clear air layer was defined as regions where the MPL data exhibit signal-to-noise ratio (SNR) greater than 6, and the NRB signal was found to decrease with altitude in the same manner as expected for a Rayleigh-only NRB signal. The SNR is estimated as [Spinhirne, 1993; Campbell et al., 2008]

display math(3)

[11] The procedure used to determine C, δPNRB (z) and associated error calculation in this study is similar to the one described in Welton et al. [2002], Campbell et al. [2008], and section 2.2.

[12] Because wavelengths of the MFRSR and MPL systems are different, MFRSR AOD values were converted for use directly with MPL data at 532 nm by using the Angström relation [Angström, 1964] as

display math(4)

where τ500 and τ415 are AOD values at wavelengths 500 and 415 nm, respectively. The MFRSR AOD values are available only during daytime; hence, corresponding C values can only be derived during daytime. Nighttime C values are assumed using a linear interpolation between the late afternoon and following early morning calculated C values. A large change in the magnitude of C overnight may produce large uncertainties in the nighttime results; thus, periods with large changes were excluded from this analysis. The acceptable interpolated C values on average were found to change less than 10% overnight which adds extra uncertainty in the nighttime AOD (z) profiles. The maximum nighttime period over which the derived C values could be interpolated to obtain an estimate of AOD during the MPL operations was 16 h. The βR and σR are derived by using the values of temperature and pressure derived from the NRL-MSISe00 model [Picone et al., 2002].

2.1 Calculation of AOD (z)

[13] The aerosol extinction coefficient, σp (z), is calculated according to

display math(5)

[14] Aerosol optical depth, AOD (z) is calculated as the integral of the extinction coefficient from z to the maximum altitude zc as

display math(6)

where i represents bin numbers and is equal to 0, 1, 2, 3, …, N − 1, and z0 = 0. To solve for aerosol profiles from equations ((2)) and ((5)), an initial value of S equal to SR is used, and the equation is solved iteratively by adjusting the assumed value of S until AOD (z0) and the corresponding MFRSR AOD agree within ~ 0.01. This is done for each profile and for all data used (Figure 2). S values are further discussed in section 3.3 and are shown in Figures 11a–11e for all five sites as frequency plots. The assumption of constant extinction between the surface and 180 m in this calculation might offset MPL AOD (z) values compared to the corresponding values from MFRSR measurements. However, this effect is negligible.

Figure 2.

Total number of observations made by month at each site. The days were defined by MFRSR and MPL data availability during clear days.

[15] An example of PNRB (z), β (z), σp (z), AOD (z), C, and AOD at the SGP site is shown for 15 April 2008 in Figures 3a–3f with corresponding uncertainties, which are described in more detail in the following section. This example shows the evolution of the convective boundary layer (CBL) in the time-height display (Figure 3a) of PNRB (z). This evolution of CBL depth is mainly due to the variations of solar radiation and heat flux from the surface, stability characteristics of the free troposphere above, synoptic conditions, and local terrain [Coulter, 1979; Cohn and Angevine, 2000]. Aerosol particles are mostly confined to the CBL during both night and day, with a relatively diffuse aerosol layer extending to 5 km.

Figure 3.

Example profiles related to finding AOD (z) for 15 April 2008, at SGP. Included are (a) vertical time section (30 s) of PNRB (z), 24 h average of (b) PNRB (z), (c) β (z), (d) σp (z), (e) AOD (z), and (f) column-integrated AOD (z0). In Figure 3f, MFRSR AOD (lower red dots), corresponding interpolated C values (upper red dots) and calculated C values (upper black dots) are shown. In Figures 3c and 3d, the values from Rayleigh scattering are shown in green. The corresponding 1 sigma measurement errors are given in dashed red lines.

[16] Daily-averaged vertical aerosol distributions in Figures 3b–3f reflect the pattern of predominant aerosol loading confined to the PBL and a thin extended aerosol layer above, falling below detectable limits near 4–5 km. The effect of propagating the 1 sigma uncertainties is shown in dashed red lines for all corresponding quantities. The relative uncertainty for PNRB (z) profiles varies between 12% at the surface and 15% at 4 km. For the β (z) profile, the uncertainty at the surface is 22% and 28% at 4 km. The uncertainty for σp (z) profiles at the surface is 25% and at 4 km is 30%, whereas the uncertainty for AOD (z) profiles at the surface is 28% and 48% at 4 km for this individual day. Large uncertainty in AOD (z) at 4 km is due to the lack of sufficient vertical integration for AOD (z). The results of diurnally averaged AOD (z) profiles for the data (Figure 2) from 2007 to 2010 at different ARM sites are discussed in section 3.

2.2 Uncertainty Analysis

[17] The uncertainty in MPL photon counts, P (z), is calculated using Poisson statistics as

display math(7)

where N is the product of the number of intervals (75,000 for 30 s average) and bin time intervals (200 ns for 30 m resolution in this study) [Kovalev and Eichinger, 2004].

[18] The uncertainty in the background B is also given by the form of equation (7). The corresponding uncertainty in PNRB (z) is calculated according to

display math(8)

where δAraw(z,E) is the uncertainty in the measurement of afterpulse, δE is the uncertainty associated with measuring the outgoing energy, and δO (z) is the uncertainty in the determination of the MPL overlap function. The details of the data processing procedure, as applied in this study, are reviewed by Campbell et al. [2002] and Welton et al. [2002]. Further description of the influence on the different sources of uncertainty on the MPL signal processing can be found in the work of Welton and Campbell [2002] and Campbell et al. [2008].

[19] The uncertainty in aerosol backscattering coefficient profiles, δβ (z), and aerosol extinction profiles, δσ (z), are calculated [Campbell et al., 2008; Young et al., 2008] as

display math(9)

and

display math(10)

where β (z) is the backscattering coefficient, δβR (z) is the uncertainty in molecular backscattering coefficient, T2 and δT2 are the values of two-way transmittance and its uncertainty, respectively, and δS is the uncertainty in backscatter-to-extinction ratio. The random errors in AOD (z) profiles are estimated by following the procedure discussed by Young et al. [2008] at each altitude, z.

[20] Uncertainties in the AOD (z) profiles are due to both random errors and systematic errors in the lidar measurements [Liu et al., 2006; Young et al., 2013] and atmospheric variability. Young et al. [2013] gave a detailed analysis of the sources of error and how they contribute to the overall uncertainty. In this case, the principal systematic errors include (1) the uncertainty in the overlap correction (most important in the lowest 2 km of the atmosphere) caused mainly by temperature dependence of the receive optics, (2) afterpulse and dead-time correction, (3) and variability in the lidar ratio S. The random errors, on the other hand, arise from the uncertainty in photon counting and energy monitoring. During daytime, the background signal increases, with a resulting reduction in the signal-to-noise ratio, which causes uncertainty to increase accordingly. Atmospheric variability is mainly due to the geophysical variability, which is discussed more in section 2.3.

[21] The average energy monitor value ranged from 3.77 to 3.87 μJ and the relative uncertainty in energy is less than 1%. The MPLs at all sites are maintained in climate-controlled environments with temperature variation held within ±2°C; the resultant uncertainty in overlap function is estimated at 6–10%. Welton and Campbell [2002] indicated that relative uncertainty in the overlap correction is typically on the order of 5–10%. The MFRSR-derived AOD at 500 nm (converted to 532 nm to match the lidar AOD (z) values) has been used as a constraint on the determination of σp (z) through equations ((2)) and ((5)). Therefore, the uncertainty of AOD (z) is also associated with the uncertainty on MFRSR AOD. The relative uncertainty for the Rayleigh scattering terms in equation (9) is estimated at a combined 5% on βR. [Campbell et al., 2008; Chazette, 2003]. Previous studies, as reported by Welton et al. [2002, and references therein], showed that the uncertainty for S ratio for continental and marine aerosols is on the order of 20%. An uncertainty of 20% is assumed for MFRSR AODs [Johnson et al., 2008]. Furthermore, the uncertainty associated with interpolation between daytime and nighttime C values is assumed to be about 10%.

2.3 Variability in AOD (z) Profiles

[22] The AOD (z) variance, Sτ2 (z), was calculated using multiyear monthly averaged AOD (z) profiles and the individual diurnally averaged AOD (z) profiles for that month and for each site. Sτ2 (z) and the root-mean-square value, Sτ (z) for each season was calculated by averaging the daily Sτ2 (z), profiles for a particular season and taking the square root of that seasonal average for each site. Sτ (z) values were further divided by corresponding mean AOD (z) to calculate the relative variability of AOD (z), Srel (z). The results of Srel (z) for all seasons and for all sites are shown in the inset plots of Figures 4, 6-9, 12, and 13 described below.

Figure 4.

(a) Vertical profiles of multiyear seasonally averaged AOD (z) at SGP from the 2007–2010 observations during daytime periods, for winter (black), summer (red), fall (October–November; green), and spring (March–April). Inset plots are the profiles of corresponding relative standard deviation, Srel (z). The corresponding 1-sigma measurement errors are given in horizontal bars. (b) Vertical profiles of multiyear seasonally averaged AOD (z) at SGP from the 2007–2010 observations during nighttime periods, for winter (black), summer (red), fall (October–November; green), and spring (March–April; blue). Inset plots are the profiles of corresponding relative standard deviation, Srel (z). The corresponding 1-sigma measurement errors are given in horizontal bars.

[23] Srel (z) values increase with altitude for all sites. The Srel (z) values at SGP show clear seasonal variability with larger values in summer, winter, and spring during both daytime and nighttime periods varying between 0.25 and 0.99 at the near surface and at 4 km. The fall Srel (z) values at SGP are relatively smaller compared to the other three seasons especially at higher altitudes, varying between 0.25 and 0.60 at the near surface and at 4 km. At TWP, maximum Srel (z) is observed in fall (0.4 near surface) during both daytime and nighttime periods. In spring and winter, it is close to 0.98 at 2.15 km and minimum Srel (z) values observed in summer at 2.15 km. However, at near surface, the values are minimum in spring and winter varying between 0.25 and 0.3. At NSA, maximum Srel (z) values observed in summer (0.22 near surface and 0.85 at 3 km) and minimum Srel (z) observed in spring (0.20 near surface and 0.8 at 3 km) during nighttime. The corresponding values are slightly smaller during daytime period. Minimum Srel (z) values observed at GRW in spring vary between 0.22 near surface and maximum Srel (z) values of 1 observed in all seasons at 2.15 km. There is no significant difference in Srel (z) during daytime and nighttime periods. Similarly at FKB, Srel (z) values show a clear seasonal variation with Srel (z) values varying between 0.4 near surface in summer and 0.60 at 2.25 km in spring. Overall, variability in FKB site is relatively smaller as compared to the other sites, but the amount of data used at this site is relatively small. The daytime Srel (z) values at FKB site are slightly larger as compared to the corresponding nighttime values.

[24] As with the large variability in AOD (z) reported here, Matthias et al. [2004] presented vertical profiles of aerosol extinction coefficient derived from regular Raman lidar measurements over Europe within the framework of the European Aerosol Research Lidar Network and reported the relative standard deviation of the values on the order of 0.50–0.80 below 2 km layer and indicated that relative variability of more than 1 can be reached in the 2–5 km layer. Eck et al. [2009] performed a long-term monitoring of aerosol optical properties at boreal forest AERONET, Aerosol Robotic Network site in interior Alaska, and reported large relative variability AOD sometime exceeding 1.0.

3 Results and Discussion

3.1 Climatology of the Aerosol Optical Depth

[25] A multiyear average of AOD (z) from each month was developed during daytime and nighttime periods for the 4 year period for SGP, TWP, and NSA; a 2 year period for GRW; and a 1 year period for FKB. These AOD (z) profiles are examined to assess seasonal behavior, which are defined by groupings of the similar data exhibiting similar characteristics for a common period of the year. Average values obtained at elevated heights are dominated by very small values that are often too small to be realistic or reasonable and cause the average profile to be unrepresentative. Therefore, AOD (z) values for heights well above the mixed layer are summarized in the form of probability distributions presented in section 3.2.

3.1.1 SGP Site

[26] Figures 4a–4b illustrate the result of averaging the data at SGP seasonally below 4 km during daytime and nighttime periods. These averaged AOD (z) profiles show that most aerosol extinction is concentrated below 2 km, with diffuse structure extending up to 4 km. The AOD (z) averages show a clear seasonal variation that differs below and above an apparent transition region at 2.5 ± 0.5 km. Below that altitude, the maximum AOD (z) values (0.17 ± 0.08 for daytime and 0.16 ± 0.066 for nighttime) are observed in summer (May, June, July, August, and September) and the minimum values (0.082 ± 0.023 for daytime and 0.080 ± 0.03 for nighttime) in winter (December, January, and February) near the surface. However, above the transition altitude region, maximum values are observed during winter and minimum values during summer. The spring (March and April) and fall (October and November) values lie in between the summer and winter values during nighttime. However, during daytime, the minimum values were observed in spring and fall. The inset plots shown in Figures 4a–4b are seasonal average profiles of the relative standard deviation in AOD (z), as described in section 2.3. The corresponding measurement uncertainties with ±1 standard deviation are presented as horizontal bars in each figure and are also given in Tables 1 and 2 for daytime and nighttime periods, respectively. Very small difference in AOD (z) values was observed between daytime and nighttime periods. Below transition altitudes are slightly larger values observed during daytime, whereas at higher altitudes, the values are slightly larger during nighttime. However, the uncertainties at higher altitudes are large.

Table 1. Seasonal and Annual Mean AOD Values at the Five Sites During Daytime, Near the Surface, or at an Altitude of 2.15, 2.25, 3, or 4 kma
SeasonSite
SGPTWPNSAGRWFKB
  1. aThe numbers in parentheses are altitudes in kilometers; (0) represents the near surface.
Winter0.0820 (0)0.1173 (0)  0.1040 (0)
±0.023 (0)±0.063 (0)  ±0.04 (0)
0.0061 (4)0.008 (2.15)  0.0035 (2.25)
±0.0025 (4)±0.003 (2.15)  ±0.002(2.25)
Summer0.1712 (0)0.093 (0)0.091 (0)0.085 (0)0.1423(0)
±0.066 (0)±0.056 (0)±0.037 (0)±0.02 (0)±0.045 (0)
0.0041 (4)0.013 (2.15)0.009 (3)0.005 (2.15)0.01 (2.25)
±0.0017 (4)±0.008 (2.15)±0.0043 (3)±0.004(2.15)±0.008(2.25)
Spring0.116 (0)0.2300 (0)0.107 (0)0.122 (0)0.12 (0)
±0.042 (0)±0.112 (0)±0.044 (0)±0.051 (0)±0.08 (0)
0.0047 (4)0.027 (2.15)0.01 (3)0.004 (2.15)0.012 (2.25)
±0.0019 (4)±0.017 (2.15)±0.0035 (3)±0.001(2.15)±0.006 (2.25)
Fall0.088 (0)0.089 (0) 0.0720 (0) 
±0.031 (0)±0.068 (0) ±0.027 (0) 
0.006 (4)0.003 (2.15) 0.006 (2.15) 
±0.0024 (4)±0.0031(2.15) ±0.002(2.15) 
Annual mean0.118 (0)0.130 (0)0.1 (0)0.0930 (0)0.1240 (0)
±0.038 (0)±0.078 (0)±0.042 (0)±0.033 (0)±0.042 (0)
0.006 (4)0.008 (2.15)0.01 (3)0.005 (2.15)0.01(2.25)
±0.0025 (4)±0.006 (2.15)±0.0037 (3)±0.003(2.15)±0.004 (2.25)
Table 2. Seasonal and Annual Mean AOD Values at the Five Sites During Nighttime, Near the Surface, or at an Altitude of 2.15, 2.25, 3, or 4 kma
SeasonSite
SGPTWPNSAGRWFKB
  1. aThe numbers in parentheses are altitudes in kilometers; (0) represents the near surface.
Winter0.080 (0)0.116 (0)  0.10 (0)
 ±0.030 (0)±0.064 (0)  ±0.03 (0)
0.0089 (4)0.0073 (2.15)  0.003 (2.25)
±0.0047 (4)±0.003 (2.15)  ±0.0025(2.25)
Summer0.163 (0)0.1115 (0)0.080 (0)0.0880 (0)0.130 (0)
±0.0804 (0)±0.063 (0)±0.032 (0)±0.036 (0)±0.039 (0)
0.0075 (4)0.018 (2.15)0.0067 (3)0.005 (2.15)0.013 (2.25)
±0.006 (4)±0.008 (2.15)±0.003 (3)±0.004(2.15)±0.007(2.25)
Spring0.1100 (0)0.242 (0)0.100 (0)0.123 (0)0.150 (0)
±0.055 (0)±0.0983 (0)±0.04 (0)±0.075 (0)±0.06 (0)
0.0067 (4)0.0256 (2.15)0.009 (3)0.004 (2.15)0.006 (2.25)
±0.0046 (4)±0.011 (2.15)±0.004 (3)±0.002(2.15)±0.005 (2.25)
Fall0.084 (0)0.096 (0) 0.07 (0) 
±0.035 (0)±0.068 (0) ±0.046 (0) 
0.0065 (4)0.0083 (2.15) 0.006 (2.15) 
±0.0034 (4)±0.0063(2.15) ±0.002(2.15) 
Annual mean0.110 (0)0.140 (0)0.09 (0)0.093 (0)0.127 (0)
±0.05 (0)±0.073 (0)±0.037 (0)±0.056 (0)±0.047 (0)
0.0065 (4)0.015 (2.15)0.008 (3)0.006 (2.15)0.02(2.25)
±0.0045 (4)±0.005 (2.15)±0.0037 (3)±0.004(2.15)±0.009 (2.25)

[27] The seasonal trend of AOD (z) found below ~2.5 km in this study is consistent with that reported for the profiles of aerosol extinction coefficients by Turner et al. [2001] using Raman lidar at this site. However, the seasonal trend found above ~2.5 km from this study is different. The two studies are not coincident in time (approximately 1 decade apart) nor are they of the same duration (four winter periods for the present study, one for Turner et al. [2001]); thus, it is difficult to determine a single controlling factor for this difference. The present study assumes a constant value for backscatter-to-extinction ratio with altitude and the wavelengths are not the same (532 nm versus 355 nm). Given the relatively small values of extinction coefficient normally encountered above the mixed layer and the observed variance, these differences on a single year time frame are likely possible and deserve continued study.

[28] Two complementary analyses have been applied at different heights for daytime and nighttime values to further assess seasonal behavior (Figures 5a–5d). First, the complete 4 year set of daily daytime and nighttime estimations of AOD (z) are plotted for the surface and at 4 km versus time in days. A least squares fit with 12 cosine and sine functions, and a period ranging from 48 months to 4 months, is performed on the diurnal AOD (z) values, giving insight into the seasonal and interannual variability of AOD (z) (Figures 5a and 5b). The linear fits to the daytime and nighttime values are shown in light green and red curves, respectively. The multiyear monthly averaged AOD (z) values (Figures 5c and 5d) at both levels are also shown with ±1 standard deviation from the mean for each month and for daytime (light green curve) and nighttime (red curve). The annual cycle that is evident at SGP supports maximum AOD (z) values during summer and minimum AOD (z) during winter near the surface. The upper level consistently shows larger AOD (z) values during winter and lower values in summer, consistent with Figure 4. However, large day-to-day variability in AOD (z) daily mean values is reflected in Figures 4, 5c, and 5d. The standard deviation relative to the mean value is higher (~0.7) in the upper level than that in the lower level (~0.2–0.4). Figures 5a and 5b show some interannual variability; the summer maximum at SGP varies from 0.42 to 0.21 at the surface, whereas the winter maximum varies from 0.044 to 0.022 at 3 km.

Figure 5.

Daily estimates of AOD at SGP from January 2007 to May 2010, (a) at the surface and (b) at 4 km. In Figures 5a and 5b, the linear fits to the daytime and nighttime values are shown in light green and red curves, respectively. Also shown are multiyear monthly means, at the (c) surface and at (d) 4km. (e) S values. Light green and red curves in Figures 5c and 5d are for daytime and nighttime periods, respectively. The corresponding ±1 standard deviation from the mean in Figures 5c and 5d are shown by dashed curves.

[29] Michalsky et al. [2010] also found the highest AOD values in the summer and lowest in winter using 16 years of radiometric measurements from 1992 to 2008 at SGP. They further reported that the summertime maxima vary considerably year to year, which is consistent with the present near-surface results. Similar seasonal and interannual variations were also reported by Michalsky et al. [2001] using 8 years of radiometric measurements from 1993 to 2000 at SGP.

3.1.2 TWP-Darwin Site

[30] The results of an analysis of the TWP-Darwin MPL data, similar to that described above for the SGP, indicate the highest aerosol concentrations below about 1.5 km. Diffuse particles are present up to 2.15 ± 0.5 km. In contrast to the SGP, maximum AOD (z) values (0.23 ± 0.112 for daytime and 0.242 ± 0.0983 for nighttime) near surface are observed in spring (September, October, and November) and minimum values (0.089 ± 0.068 for daytime and 0.096 ± 0.068 for nighttime) in fall (April and May). Winter (June, July, and August) and summer (December, January, February, and March) values mostly lie in between spring and fall values at all altitudes (Figures 6a and 6b) during both daytime and nighttime. Tables 1 and 2 summarize AOD (z) values at all sites. AOD (z) values during nighttime are slightly higher than those during daytime periods in almost all seasons at TWP. As seen in the inset plot of Figures 6a and 6b, large relative standard deviations in AOD (z) profiles are observed at all altitudes and for all seasons, reaching about 1 at 2.15 km. The corresponding 1 sigma measurement errors are shown in horizontal bars.

Figure 6.

(a) Vertical profiles of multiyear seasonally averaged AOD (z) at TWP-Darwin from the 2007–2010 observations during daytime periods. Corresponding values shown are for monsoonal summer (red), dry winter (blue), fall (green), and spring (black) seasons. Inset plots are the profiles of corresponding relative standard deviation, Srel (z). The corresponding 1-sigma measurement errors are given in horizontal bars. (b) Vertical profiles of multiyear seasonally averaged AOD (z) at TWP-Darwin from the 2007–2010 observations during nighttime periods. Corresponding values shown are for monsoonal summer (red), dry winter (blue), fall (green), and spring (black) seasons. Inset plots are the profiles of corresponding relative standard deviation, Srel (z). The corresponding 1 sigma measurement errors are given in horizontal bars.

[31] Maximum AOD (z) values during the spring season likely reflect increased dust activity and biomass burning incursions [Bouya et al., 2010; Radhi et al., 2010]. The variability decreases as the wet summer season progresses and becomes relatively small in fall and early winter [Radhi et al., 2010]. Other studies have mentioned that biomass burning activity at the TWP-Darwin site and in the Northern Territory of Australia begins in June and becomes extreme in October [Bouya et al., 2010; Radhi et al., 2006, 2010]. Some researchers [Radhi et al., 2006, 2010], however, note that the winter season in Australia (Northern Territory) is dominated by lower wind speed and clean sky, resulting in relatively less dust emission during this season. This interpretation is consistent with relatively lower AOD (z) values observed in the present study during the winter.

3.1.3 NSA, GRW, and FKB Sites

[32] The results of an analysis of the NSA, GRW, and FKB MPL data, similar to that described above for the SGP and TWP-Darwin sites, are summarized in Figures 7-9. Multiyear seasonally averaged vertical profiles of AOD (z) show that most aerosol particles are concentrated below 1.5 ± 0.5 km, near the PBL region, with a thin elevated layer of diffuse aerosol extending up to 3 km at NSA, 2.15 km at GRW, and 2.25 km at FKB. The AOD (z) seasonal averages (Figures 7a and 7b) at NSA show a small but clear seasonal variability with maximum values (0.107 ± 0.044 for daytime and 0.1 ± 0.04 for nighttime) in spring (March to May) and minimum values (0.091 ± 0.037 for daytime and 0.08 ± 0.032 for nighttime) in summer (June to September) at near surface (see Tables 1 and 2). Winter values could not be calculated for NSA, because the MFRSR is inoperative during nighttime.

Figure 7.

(a) Vertical profiles of multiyear seasonally averaged AOD at NSA during daytime. Corresponding values are shown for summer (red), and spring (black) seasons. Inset plots are the profiles of corresponding relative standard deviation, Srel (z). The corresponding 1-sigma measurement errors are given in horizontal bars. (b) Vertical profiles of multiyear seasonally averaged AOD at NSA during nighttime. Corresponding values are shown for summer (red), and spring (black) seasons. Inset plots are the profiles of corresponding relative standard deviation, Srel (z). The corresponding 1 sigma measurement errors are given in horizontal bars.

Figure 8.

(a) Vertical profiles of multiyear seasonally averaged AOD at GRW during daytime. Corresponding values are shown for summer (red), fall (green), and spring (black) seasons. Inset plots are the profiles of corresponding relative standard deviation, Srel (z). The corresponding 1-sigma measurement errors are given in horizontal bars. (b) Vertical profiles of multiyear seasonally averaged AOD at GRW during nighttime. Corresponding values are shown for summer (red), fall (green), and spring (black) seasons. Inset plots are the profiles of corresponding relative standard deviation, Srel (z). The corresponding 1 sigma measurement errors are given in horizontal bars.

Figure 9.

(a) Vertical profiles of multiyear seasonally averaged AOD at FKB during daytime. Corresponding values are shown for summer (red), winter (green), and spring (black) seasons. Inset plots are the profiles of corresponding relative standard deviation, Srel (z). The corresponding 1-sigma measurement errors are given in horizontal bars. (b) Vertical profiles of multiyear seasonally averaged AOD at FKB during nighttime. Corresponding values are shown for summer (red), winter (green), and spring (black) seasons. Inset plots are the profiles of corresponding relative standard deviation, Srel (z). The corresponding 1-sigma measurement errors are given in horizontal bars.

[33] Some studies have indicated that spring in Barrow is the season that experiences the highest concentration of dust aerosol from southeastern Asia [Raatz and Shaw, 1984; Rahn, 1981a, 1981b, 1985]; this is consistent with maximum AOD (z) values observed in spring in the present study. Higher mean AOD (z) values during spring season are also consistent with the peak arctic haze occurrences during March, April, and May as reported by Shaw [1995] and Herber et al. [2002]. The higher values of AOD (z) at NSA in the spring are also due in part to the transport of smoke from Siberian forest fires [Brock et al., 2011].

[34] The GRW site (Figures 8a and 8b) experiences similar seasonal variation during daytime and nighttime period with maximum AOD (z) values during spring (April and May) and minimum values during fall (September and October) (see Tables 1 and 2). Corresponding values during summer (June, July, and August) lie in between spring and fall. Above ~1.5 km, the AOD (z) maximum is observed in fall and minimum during spring for both daytime and nighttime. No clear days are available during winter time and are not included in this analysis. The GRW site, in the Atlantic Ocean, is mostly dominated by maritime aerosols [Smirnov et al., 1995; Sayer et al., 2012]. However, Saharan dust storms are occasionally present over the GRW site [Carlson and Prospero, 1972], especially during late spring through early fall, with maximum aerosol concentrations at 1.5–3.7 km [Prospero and Carlson, 1972; Carlson and Prospero, 1972].

[35] During both nighttime and daytime periods, distinct seasonal patterns, albeit over only a single year, are evident at the FKB site (Figures 9a and 9b), with minimum AOD (z) values in winter (November and December) and maximum values during both spring (April and May) and summer (June, July, and August) (see Table 2) during daytime. However, during nighttime, the seasonal variation is different with maximum values in summer and minimum values in winter. Spring values lie in between summer and winter (see Table 1). No clear days are available during fall season at FKB. Matthias et al. [2004] performed seasonal studies on AOD (z) at three different sites in Germany and observed a distinct annual cycle with two maxima, one in spring and one in summer, and minimum values in winter at lower altitudes (below ~1.5–2 km), consistent with the pattern observed here in lower altitudes particularly in daytime period. As for SGP and TWP, the corresponding seasonal mean values of relative standard deviation at these three sites are shown in inset plots (Figures 7-9) and discussed in section 2.3. The corresponding 1 sigma measurement uncertainties are given as horizontal bars for each site and for each season (Figures 7-9).

3.2 AOD (z) Well Above the Boundary Layer

[36] Figure 10 summarizes results from elevated heights at each site. The height, zm, at each site above which the distribution of daily AOD (z) values becomes dominated by near-zero values of AOD (zm), was used as the height for these AOD (z) estimates (4 km for SGP, 3 km for NSA, 2.15 km for TWP and GRW, and 2.25 km for FKB). The most common AOD (zm) values are between 0.0075 and 0.01. GRW and SGP exhibit no values above 0.04, but NSA, FKB, and TWP indicate occasional significant AOD (z) values near 0.05. Almost all occurrences at NSA occurred during March–May 2008 and may well be the result of forest fires. Eck et al. [2009] performed a long-term monitoring of aerosol optical properties at boreal forest AERONET, Aerosol Robotic Network site in interior Alaska, and reported that elevated AOD (z) values were likely associated with boreal forest fires as well as arctic haze occurrences during March, April, and May, similar to results by Shaw [1995] and Herber et al. [2002]. Other studies report that southern Russian and Siberian open biomass and forest fires have contributed to the elevated aerosols at NSA [Brock et al., 2011], which was unexpected.

Figure 10.

Probability distributions of annual mean AOD (z) at different heights for all sites. SGP (black), NSA (red), TWP (green), FKB (blue), and GRW (yellow).

3.3 Extinction-to-Backscatter Ratio (S)

[37] The daily values of lidar extinction-to-backscatter ratio (S) were estimated for each site and for all data used (Figure 2). The results are shown in Figures 5e and 11a–11e. At SGP, the S values range from 15 sr to 75 sr with most common values between 28 sr and 35 sr (Figure 11a). The larger values are generally observed during spring and late summer (Figure 5e). The S values are more scattered at TWP site ranging from 13 sr to 64 sr with most prevalent values between 20 sr and 27 sr (Figure 11b). S values at NSA varied between 12 sr and 62 sr with most dominant values between 28 sr and 35 sr (Figure 12c). The ranges of S values at FKB and GRW sites are from 25 sr to 40 sr with dominant S values between 28 sr and 35 sr at FKB and 20 sr and 27 sr at GRW, respectively (Figures 12d and 12e). However, FKB and GRW have significantly fewer clear days available as compared to the other three sites.

Figure 11.

Histogram of daily estimates of S values at (a) SGP, (b) TWP, (c) NSA, (d) FKB, and (e) GRW sites.

Figure 12.

Vertical profiles of multiyear annually averaged AOD (z) during daytime and nighttime periods at (a) SGP and (b) NSA. Inset plots are the profiles of corresponding relative standard deviation, Srel (z). The corresponding 1 sigma measurement errors are given in horizontal bars.

[38] Omar et al. [2009] performed a study to determine different aerosol types and appropriate S values by using Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) aerosol model and estimated that the mean S values at 532 nm for continental aerosol and marine aerosols are 35 sr and 20 sr, respectively, which are close to the corresponding dominant values found in the present study. Young et al. [2013] studied lidar ratios for various aerosol and cloud types using version 3 CALIPSO data analysis and reported characteristic lidar ratios varying between 19 ± 9.5 sr and 70 ± 28 sr for water cloud and smoke, respectively. They also reported 25 ± 10 sr for clean marine, 35 ± 15.8 sr for continental, 40 ± 20 sr for dust, and 55 ± 22 sr for polluted dust aerosols.

[39] The first Caribbean campaign, which made frequent measurements of maritime aerosol using High Spectral Resolution Lidar at 532 nm, exhibited the values (25 sr) of lidar ratio smaller than the corresponding values (40 sr) observed over Mexico during the MILAGRO campaign that made the measurements of continental aerosols [Burton et al., 2012]. The same study further reported that aerosols observed over eastern and southeastern U.S. ranged from 25 sr to 50 sr, which are in good agreement with the values found in the present study.

3.4 Annual Mean AOD (z) for All Sites

[40] Multiyear annual mean AOD (z) values during daytime and nighttime for all five sites (Figures 12 and 13) demonstrate that aerosol loading differs significantly from site to site at all altitudes. The AOD (z) values are not significantly different between daytime and nighttime periods for almost all sites. Minimum AOD (z) values (Tables 1 and 2) are found at GRW at almost all altitudes, and the maximum values are found at TWP during both daytime and nighttime periods. Mean AOD (z) values close to the ground at FKB, TWP-Darwin, and SGP are very similar. Moreover, up to about 1 km from the ground, the smallest AOD (z) (at GRW and NSA) and the largest AOD (z) (at SGP, TWP-Darwin, and FKB) are clearly distinct; though, at higher altitudes the values vary. Corresponding mean AOD (z) values at TWP-Darwin and FKB are similar at almost all heights. Minimum mean AOD (z) values at GRW and the maximum values at SGP, TWP-Darwin, and FKB are not unexpected for maritime aerosols (especially sea-salt aerosols) and continental aerosols (dust, biomass burning aerosols, etc.), respectively. One possible explanation for larger AOD values at higher altitudes at SGP is the more frequent occurrence of dust and smoke advected downward of their source within the free troposphere [Kaufman et al., 2002; Rahn, 1981a, 1981b, 1985; Raatz and Shaw,1984].

Figure 13.

(a) Vertical profiles of multiyear annually averaged AOD (z) at TWP, GRW and FKB during daytime periods. Inset plots are the profiles of corresponding relative standard deviation, Srel (z). The corresponding 1-sigma measurement errors are given in horizontal bars. (b) Vertical profiles of multiyear annually averaged AOD (z) at TWP, GRW and FKB during nighttime periods. Inset plots are the profiles of corresponding relative standard deviation, Srel (z). The corresponding 1 sigma measurement errors are given in horizontal bars.

[41] The average maximum detectable aerosol altitude from our methods at SGP is 4 km, at NSA is 3 km, at TWP and GRW is 2.15 km, and at FKB is 2.25 km. There are contributions to AOD from above these altitudes, and this contributes to the uncertainty in absolute MPL-derived AOD (z) values. Livingston et al. [2005], for example, reported that typical AOD values measured from the NASA Ames Airborne Tracking Sun photometer (AATS-14) centered between 453 and 864 nm wavelengths at altitudes of ~ 8–12 km ranged from 0.003 to 0.009. However, the MFRSR AOD is used as a calibration factor for the MPL determination at the surface, to which contributions above 4 km are considered negligible and affect the MPL “calibration” very little. Values less than 0.01 at large heights are more appropriately represented as relative contributions to overall AOD because of the limited range of the integral above the mixed layer and lack of sensitivity. Although this detracts from upper level AOD (z) being an absolute measure of AOD (z) above the mixed layer, the relative values reported here can be compared from site to site because the MPLs in use are identical.

4 Summary and Conclusions

[42] Diurnally averaged vertical profiles of aerosol extinction coefficients derived from micropulse lidar (MPL) data, using corresponding aerosol optical depth (AOD) data as input from an independent measurement collected with colocated multifilter rotating shadowband radiometer (MFRSR) systems for five different Atmospheric Radiation Measurement (ARM) Program sites (SGP in north-central Oklahoma, TWP-Darwin in the tropical western Pacific, NSA at Barrow on the North Slope of Alaska, GRW at Graciosa Island in the Azores, and FKB in the Black Forest of Germany) were used to determine vertical profiles of AOD (z). The multiyear diurnally, monthly and seasonally averaged AOD (z) climatologies presented in this paper for the five sites were derived for data collected from 2007 through 2010. Significant results are as follows:

  1. [43] Most aerosols are confined to the convective boundary layer (CBL) region at all sites; however, the average aerosol maximum heights vary from 2.15 km (TWP-Darwin, and GRW), 2.25 km (FKB), 3 km (NSA), to 4 km (SGP). Lower layer aerosols are presumably generated locally, while diffuse higher layer aerosols are most likely from convective lifting and transport.

  2. [44] There are no significant differences in AOD (z) values during daytime and nighttime periods. However, large day-to-day variability of AOD (z) exists at all sites and all altitude levels.

  3. [45] Significant interannual variability is observed at SGP.

  4. [46] A well-defined seasonal variation in AOD (z) profiles found at all five sites exhibits different seasonal behavior at different sites. Moreover, the variability is often different for the lower level (roughly within the CBL) than for levels above (Tables 1 and 2). The patterns at the different sites are as follows:

    1. At SGP, lower level AOD (z) values are high during summer and low during winter, while the reverse is the true for upper levels.
    2. At TWP-Darwin, AOD (z) values are larger during spring than summer at all altitudes.
    3. At NSA, AOD (z) values are larger during spring than summer at all heights.
    4. At GRW, maximum AOD (z) values occur during spring in the lower level and during fall in the upper level, whereas the minimum values occur in fall at lower level and in spring at upper level, whereas summer values lie in between spring and fall for both heights.
    5. At FKB, maxima in AOD (z) mean values were found during both spring and summer during daytime whereas the maximum values were found in summer during nighttime; the minimum values were found in winter for all heights and during both nighttime and daytime periods.
  5. [47] A lower annual mean values (0.093 ± 0.03 for daytime and 0.092 ± 0.045 for nighttime) at GRW are not unexpected for maritime (mostly sea-salt aerosols) aerosols, and the higher values at TWP (0.13 ± 0.075 during daytime and 0.14 ± 0.073 during nighttime), SGP (0.116 ± 0.05 nighttime and 0.128 ± 0.038 daytime), and FKB (0.13 ± 0.042 daytime and 0.12 ± 0.047 nighttime) are typical of continental aerosols. Annual mean values found at NSA are 0.1 ± 0.04 and 0.091 ± 0.036 during daytime and nighttime periods, respectively.

  6. [48] Most prevalent values of lidar extinction-to-backscatter ratio (S) differ from site to site. At SGP, the values are between 28 sr and 35 sr, TWP 20 sr and 27 sr, NSA 28 sr and 35 sr, FKB 28 sr and 35 sr, and GRW 20 sr and 27 sr.

[49] These results demonstrate that different types and sources of air masses from different meteorological and climatic conditions likely contribute to atmospheric aerosol mass loadings in different altitude regions.

[50] The MPL has proven to be a very useful tool for the quantitative determination of the aerosol loading from the ground to nearly 10 km at multiple locations worldwide. These results from regular, extensive observations in diverse climate regimes are relevant to improved understanding of aerosol properties and boundary layer dynamics, as well as to improving global climate models by incorporating aerosol radiative effects. These AOD (z) profiles will also serve as a constraint to calculate the value added product for solar heating rates at various altitudes in the atmosphere at various ARM sites. Therefore, ARM scientific community will be benefited because of both the nighttime and daytime availability of these results.

Acknowledgments

[51] This work was supported by the U.S. Department of Energy (Office of Science, Office of Biological and Environmental Research) as part of the ARM Program, through contract DE-AC02-06CH11357.

Ancillary