TITLE: UH Aerosol Sizing (OPC) AUTHORS: Antony Clarke, Steven Howell, Mark Litchy University of Hawaii, Department of Oceanography 1000 Pope Rd. Honolulu, HI 96822 USA showell@soest.hawaii.edu tclarke@soest.hawaii.edu http://www.soest.hawaii.edu/HIGEAR/ mark@ctassociatesinc.com 1.0 DATA SET OVERVIEW: Introduction: This file describes aerosol particle size data obtained with an optical particle counter (OPC) aboard the NCAR C-130Q aircraft. The accompanying files provide size distributions in dN / d log D format from 0.15 to 7.8 micrometers at a resolution of 135 channels per decade. The OPC operated at reduced humidity compared with ambient conditions. While the distributions as given are adequate to show general features of the size distributions, people wishing to pursue quantitative work with these data should be aware that no corrections have been made for particle composition, relative humidity, or sampling efficiency. These issues are discussed below in more detail. In addition to normal (unheated) operation, sample air was periodically diverted through heated tubes to give some indication of the chemical composition of the particles. In the clean marine boundary layer, it has been demonstrated that particlulate volume lost after heating to 150 C corresponds well to measured sulfuric acid and heating to 300 C drives off ammonium sulfates [Clarke, 1991]. As the air sampled during INDOEX was rarely pristine, interpretation of the heated data is less straightforward. In addition to sulfuric acid, heating to 150 C will remove some nitrates and organic compounds, while heating to 300 C should remove most organic material as well as as ammonium sulfates. Refractory material that survives heating to 300 C is primarily dust, sea-salt, fly ash, and soot. Thermograms used to measure organic and elemental carbon indicated that the boundary between the two was probably nearer 380 C than 300 [Mayol-Bracero et al., 2002], so we raised the hottest tube to 380 C in RF09 and subsequent flights. Time period covered by the data: Flights RF07 (28 February 1999) to RF18 (25 March 1999) are included in this submission. The OPC was operational for all flights, but a malfunction compromised data from the first 6 flights. Contact us if you need data from those flights or the ferry flights. Physical location: Aboard the NCAR/NSF C-130Q aircraft. 2.0 INSTRUMENT DESCRIPTION: We used a modified LAS-X ASASP Optical Particle Counter. Essentially, the optics remain as manufactured, but the electronics have been replaced with a fast, wide dynamic range log amplifier and a 256 channel pulse height analyzer. The OPC received sample air through one of the 75 liter per minute inlets of the Community Aerosol Inlet [Blomquist et al. 2002]. A software-controlled valve determined whether the sample flow bypassed or went through a virtual impactor [Porter et al., 1992] that enriched large particle concentrations. Before entering the OPC, another set of software-controlled valves determined whether the sample flowed through unheated or heated inlet tubing [Clarke, 1991]. There were two heated channels. Before flight RF09 (7 March, 1999), they were set at 150 and 300 C. All later flights were at 150 and 380 C. Because OPC response is a function of composition and hence water content, changes in relative humidity affect sizing. To reduce this effect, sample air was diluted 50:50 with dessicated air. Specifications: Size range: 0.15 to 7.8 micrometers Sizing accuracy: Dependent on particle refractive index, shape, and diameter. Most likely about 5% up to 0.4 micrometers, 10% above 0.4. Large dust particles are often so aspherical that "diameter" is difficult to define, so sizes must be regarded as effective optical diameters. Sizing precision: <2% for particles of the same composition, shape, and orientation. Sensing angle: 35 to 140 degrees, but since it is an active cavity device, it is doubly sensitive from 60 to 120 degrees. Flow rate: 0.3 liters per minute +/- 5% Sampling period: Approximately 1 minute 3.0 DATA COLLECTION AND PROCESSING: The OPC itelf reported counts every 3 seconds. During flight, samples were averaged over approximately one minute periods, separated by 3 seconds to eliminate artifacts due to valve switching. While flying level, sampling rotated between unheated, 150 C and 380 (or 300) C channels. While ascending or descending, the heated channels were omitted to enhance time resolution. Description of derived parameters and processing techniques used: The OPC counts the number of particles in each of 256 voltage bins, which are proportional to the logarithm of the scattering intensity within the viewing angles of the optics. Calibrations with nearly monodisperse latex (0.1 to 3 micrometer) and glass (2 to 7 micrometer) spheres (Duke Scientific) were used to establish the relationship between voltage and diameter. A procedure similar to Clarke [1991] was used, but a cubic spline was used rather than piecewise polynomial fitting. This calibration procedure assumes that scattering intensity is monotonically related to particle diameter. Since the Mie scaattering intensity curves typically dip down between 0.4 and 0.6 micrometers and (more weakly) between 1.2 and 2 micrometers, sizing is ambiguous in these ranges. However, strongly absorbing aerosol such as we saw during INDOEX suppresses the downward wiggles in the curve, reducing this ambiguity. Because scattering intensity varies with refractive index as well as particle size, the calibration curve had to be altered to reflect the composition of the particles being sampled. Changes in real refractive index can be accomodated with a simple shift in the curve [Hand et al., 2000]. Absorption (the imaginary part of the refractive index) has a more complicated effect, increasing scattering of the smallest particles and reducing it for particles larger than about 0.8 microns [Clarke or Pinnick and Garvey?]. Unfortunately, we do not know particle composition well enough at any given time to calculate refractive index and hence the correct modification to the calibration curve. We chose to use one calibration curve for the entire project. The submicron part of the curve was based a refractive index of 1.541 - 0.055i, which was calculated from the aerosol composition measured at the Kaashidhoo station [Cite Cass' group]. As the refractive indices of latex, glass, NaCl, and many silicates are near 1.55, the supermicron part of the calibration curve was not altered. At times, composition varied enough that further corrections might be desirable. Notable possibilities include clean Southern Hemisphere and free troposheric aerosol and pollution plumes elevated well above the boundary layer. In addition, the heated inlets alter composition and thus apparent size. No correction for this has been applied. When it was used, the virtual impactor enhanced large particle concentrations in a predictable way [Porter et al,. 1992]. The data have been corrected for this, assuming that large particle density was 2.1 g/cm^3. Due to the irregularities in Mie scattering, non-ideal characteristics of the log amplifier, and counting statistics, the calculated distributions are rather noisy. Therefore, the data presented here have been smoothed with a 29 point Gaussian window. Description of quality control procedures: Samples were edited to remove time periods when the laser power fell too low, the thermal channels were not at the correct temperature, and when obvious shattering of cloud particles occurred. Flights RF01 through RF06 were removed as a design flaw prevented the valves from closing properly, so a mixture of heated and unheated sample air was fed to the OPC. 4.0 DATA FORMAT: Data files are tab-separated ASCII text. A section describing the data format preceeds the data matrix. After the description is a 2-line header with column names on the first line and units on the second. Names for the particle size columns are the diameters in micrometers. (Strictly speaking, diameters are the geometric mean diameter of the size range included in the bin, but since the bins are narrower than the sizing errors, a precise definition is irrelevant.) Columns are as follows: 1) Start time of the sampling interval, YYYYMMDDhhmmss 2) End time of the sampling interval, YYYYMMDDhhmmss 3) Longitude from RAF netCDF file, degrees 4) Latitude from RAF netCDF file, degrees 5) Pressure altitude from RAF netCDF file, meters 6) Inlet temperature flag, 1=unheated, 2=150 C, 3=300 C for RF07 and RF08, 380 for all later flights. 7) Relative humidity of air entering the OPC, % 8) Ambient relative humidity from RAF netCDF file, % 9-223) dN / d log D in each size range, # cm^{-3} at standard conditions (25 C, 1013 hPa). Note: common log (base 10), not natural log (base e). Data files are name RFXX_UH_OPC_vY.txt where XX is replaced by the flight number and Y is replaced by the version number. This is version 1. 5.0 DATA REMARKS: The dN / d log D convention used here allows different instruments to be compared directly and allows easy "visual integration" when semilog plots are used (log diameter vs. concentration). To obtain concentrations over a size range sum the appropriate channels and multiply by 1/135 = 0.007407. As relative humidity changes, particles gain and lose water to a degree determined by composition. Since sample air was heated while accelerating to aircraft speed, heated or cooled by the cabin environment, and mixed with dessicated air before entering the OPC, sample RH was rarely the same as that outside the plane. Thus, the diameters seen by the OPC and reported here are not the ambient diameters. Sample air was drawn through the CAI, which has since been shown to be pretty lousy at transmitting particles larger than a micrometer [Blomquist et al., 2002]. As the haze was usually dominated by accumulation mode (mass median diameter about 0.3 micrometers), the CAI inefficiency can be ignored for many purposes. However, we did fly through some dust plumes, and many low altitude legs had significant sea salt. Anyone wishing to examine data from those periods should be aware that the supermicron data presented in these files dramatically under-reports the true concentrations. This has little effect on the total number of particles, but can seriously affect calculations of scattering and mass from the distributions. Losses in the CAI have not been sufficiently well characterized that we are willing to present corrected size distributions, but we are in a position to provide some estimates of the losses from measurements made during the Preliminary Evaluation of the Low-Turbulence Inlet project [ http://saga.pmel.noaa.gov/ aceasia/platforms/lt_inlet/final_report/ Pelti_ReportFinal13Sept00.pdf ]. At altitudes over 2 km (low humidity dust) the CAI efficiency for 1 um particles was roughly 90%, dropping to 60% at 2 um and 30% at 7 um. At low altitude (higher humidity, some sea salt aerosol, but still primarily dust), efficiencies at 2 and 7 um dropped to about 50% and 10%, respectively. 6.0 REFERENCES: B. W. Blomquist, B. J. Huebert, S. G. Howell, M. R. Litchy, C. H. Twohy, A. Schanot, D. Baumgardner, B. Lafleur, R. Seebaugh, and M. L. Laucks, An Evaluation of the Community Aerosol Inlet for the NCAR C-130 Research Aircraft, J. Atmos. Ocean. Tech. 18:1387-1397, 2001. A. D. Clarke, A thermo-optic technique for in situ analysis of size-resolved aerosol physicochemistry, Atmos. Env. 25A:635-644, 1991. J. L. Hand, R. B. Ames, S. M. Kreidenweis, D. D. Day, and W. C. Malm, Estimates of particle hygroscopicity during the Southeastern Aerosol and Visibility Study, J. Air & Waste Manage. Assoc. 50:677-685, 2000 O. L. Mayol-Bracero, R. Gabriel, M. O. Andreae, T. W. Kirchstetter, T. Novakov, J. Ogren, P. Sheridan, and D. G. Streets, Carbonaceous aerosols over the Indian Ocean during INDOEX: Chemical characterization, optical properties and probable sources, J. Geophys. Res. in press (October, 2002 is projected). Porter, Clarke, Ferry and Pueschel, Aircraft Studies of Size- Dependent Aerosol Sampling Through Inlets, J. Geophys. Res., 97:3815-3824, 1992.