File Mode

For batch mode processing, the NetCDF files are made available via anonymous FTP at ftp://lidar.ssec.wisc.edu/data in a directory under your username.   You will be asked for a user name and files will appear in a ftp sub directory with this name. Each file will have a unique name. When processing is complete, the files are also combined and provided in a compressed Tar-BZip2 file.

Data Quality Mask

The NetCDF will contain a data quality mask in an array called qc_mask. It has the same dimensions as the lidar and/or the radar data array.   qc_mask can be used to mask data that does not meet user specified data quality standards.   The bit planes of this array contain ones when corresponding variable meets the specified threshold and zeros when the vairable fails the test.   Generation of the mask has no effect on the data.

qc_mask bitplane identification:

  • bit0 = data_ok_mask =logical and of bits 1--->9.
  • bit1 = lidar_ok_mask =lidar data is present.
  • bit2 = l2_lock =laser is locked to iodine filter wavelength.
  • bit3 = laser_seeding =laser wavelength is locked to seed laser.
  • bit4 = molecular_SNR =mol_signal/photon_counting_err_in_mol_signal> specified threshold.
  • bit5 = backscatter_SNR =backscat_xsection/photon_counting_err_in_backscat_xsection> specified threshold.
  • bit6 = mol_signal_lost =number of molecular photon counts has not decreased below specified value.
  • bit7 = min_lidar_backscat=lidar backscatter cross section above specified value.
  • bit8 = min_radar_backscat=radar backscatter cross section above specified value.
  • bit9 = radar_ok_mask =radar data is present.
  • bit10= aeri_ok_mask =aeri data is present.
  • bit11= aeri_qc_mask =aeri data has passed a quality check.

    QC thresholds:

  • minimum lidar backscatter and minimum radar backscatter provide masking for all points where these two quantities fall bellow the specified value. minimum radar backscatter can be specified in either 1/(m sr) or dBz.
  • I2_lock The HSRL uses an iodine adsorption filter to separate molecular scattering from particulate scattering. A servo-loop adjusts the laser output wavelength to center of an iodine adsorption line. When the servo fails to maintain this adjustment the calibration of the particulate scattering cross section and the optical depth are degraded. Some applications, such as cloud detection and cloud altitude determination are not very sensitive to errors in tuning--In these cases the I2_lock mask is not needed. For applications which need accurate calibration, this field can be used to create a mask marking bad data. This mask parameter monitors the leakage of transmitted light through the I2 filter. The I2_lock field accepts values between 0 and 1. When set to 0 or left blank none of the data is masked. Setting this value to 0.99 provides the most stringent test for frequency locking.
  • Laser_seeding This threshold is not user adjustable. It is fixed by the software to insure that at least 1% of the laser pulses that were transmitted within a given data record were locked to the seed laser frequency. Notice that the hardware rejects data from unseeded laser shots so that it is not included in the lidar profiles.
  • minimum molecular count The particulate backscatter cross section is computed from a ratio that contains the molecular signal in the denominator. When the molecular signal is completely lost due to attenuation by a cloud, small fluctuations in the molecular background count lead to large fluctuations in the derived backscatter cross section. These can be masked by providing a value for the minimum molecular count. All data points at altitudes beyound the altitude where the number of raw molecular counts first falls bellow this value are masked. Increase this value until the mask eliminates spurrious echos above dense clouds. For short time averages(a few seconds) a value of 1 usually suffices, larger values are required for longer averages(eg. a 180 second, 45 meter average may require values ~100).
  • molecular_SNR =mol_signal/photon_counting_err_in_mol_signal > specified threshold.
  • backscatter_SNR =particulate_backscat_xsection/photon_counting_err_in_backscat_xsection > specified threshold.   

    Raw Data

    These variables are subset of collected values used in processing data.  Channel Counts are the raw photon counts from detectors, after the pileup and afterpulse corrections have been applied.

    Radar Quantities

    These processed quantities are retrieved from a co-located radar.

    MMCR Backscatter Cross Section

    The radar backscatter cross section, in units of 1/(m sr), is computed from the MMCR Reflectivity which is provided in units of dBz.

    HSRL/MMCR Cooperative Quantities

    These quantities are calculated by combining lidar data with data from the co-located NOAA 8.6 mm wavelength radar. When cooperative quanities are requested using Generate RTI Image and Data Plots all ice clouds are assumed to be composed of bullet rosette crystals and water clouds are assumed to be composed of spheres. Other ice crystal shapes can be invoked by using Process and Export Data as a NetCDF.

    Effective Diameter Prime

    eff_diameter_prime = ((9*Volume^2)/(pi*Area))^.25 = effective diameter prime is the fundamental quantity derived from the combination of lidar and radar backscatter cross sections, where Volume^2 and Area refer to the average volume-squared and average area of the cloud particles as defined by Donovan and Lammeren JGR, v106, D21, pp 27425. The only user supplied quantity required for this computation is the backscatter phase function for ice crystals. p180_ice=p(180)/(4*pi) = ice crystal backscatter phase function. p(180)/(4*pi) can be computed from hsrl measurements for each point within the cloud, however the measurement is sensitive to altitude averaging lengths and averaging times. Short averages are likely to have excessive noise. Long averages are likely to mix regions of low and high scattering. As a result, we have decided to allow the user to specify this value as a cloud average. It is recommended that an independent analysis of the data be used to determine an appropriate value for p180_ice. Errors in effective_diameter_prime are proportional to the 4th-root of the error in p(180)/(4*pi).

    If p180_ice is set to zero, the effective_diameter_prime is computed without using an assumed value for p180_ice. This must be used with caution due to noise sensitivity inherent in computing the scattering cross section from the derivative of the optical depth.

    The following assumptions are made:

  • The cloud at a given data point is either all water or all ice.
  • When the linear depolarization is < h20_depol_threshold the cloud is water.
  • The backscatter phase function for water droplets, p(180)/(4*pi) = 0.05
  • All particles are large compared to the lidar wavelength such that the optical scattering cross section is twice the projected area of the particle.
  • All particles are small compared to the radar wavelength.
  • The radar attenuation is negligible.
    • An effective_diameter_prime is computed for each data point in the image. It is derived from the ratio of the radar and lidar backscatter cross sections. Conversion of effective _diameter prime to the standard definition of effective_diameter requires assumptions about the particle size distribution and particle morphology in the case of ice crystals. Matlab function used to compute effective diameter prime

    Particle Measurements

    The combination of lidar and radar measurements provide robust measurements of the effective_diameter_prime. However, eff_diameter=(3*Volume)/(2*Area), where volume and area refer to averages over all particle, is the quantity which best relates particle mass to radiative influence. The total crossectional area of particles can be derived from the lidar scattering cross section. Using this with the effective_diameter allows calculation of number density and liquid water content. Conversion from eff_diameter_prime to eff_diameter requires assumptions about the particle size distribution and, in the case of ice crystals, particle morphology. We assume particle sizes given by a modified gamma distribution as presented by Deirmendjian, 'Electromagnetic Scattering on Spherical Polydispersions', Elsevier, NY, 1969:

    n(D) = a * D^alpha * exp(-b*D^g)   equation 1

    Where:
    D = Maximum dimension of particle
    a = Number_of_particles*g*b^((alpha+1)/g)/gamma((alpha+1)/g)
    n = number of particles per unit volume
    b = size distribution parameter
    alpha , g = user-provided size distribution parameters     Assumptions used in derivation of particle properties:
  • The size distribution is given by equation 1.
  • The cloud at a given data point is either all water or all ice.
  • When the linear depolarization is less than h20_depol_threshold the cloud is water.
  • The backscatter phase function for water droplets, p(180)/4pi = 0.05.
  • All particles are large compared to the lidar wavelength such that the optical scattering cross section is twice the projected area of the particle.
  • All particles are small compared to the radar wavelength and the particle polarizability is not a function of particle shape.
  • Radar attenuation is negligible.
  • The volume of ice in a particle is related to its max dimension, D, by: Volume = sigma_v * pi/6 * D_ref^(3-delta_v) * D^delta_v for water, sigma_v =1 and delta_v=3 so that Volume= pi/6 * D^3.
  • the projected area of an ice particle is related to D by: Area = sigma_a *pi/4 * D_ref^(2-delta_a) * D^delta_a for water, sigma_a =1 and delta_a=2 so that Area=(pi/4) * D^2.
    For more information on volume and area relationships in ice clouds see: Mitchell and Arnott, J. Atmos. Sci., Vol 51,no 6 and Heymsfield et al, J. Atmos. Sci.,Vol 61, May, 2004. Notice tha in this treatment the power law relationships presented in the above papers have been rewritten to make the proportionality constants sigma_a and sigma_v non-dimensional.
  • sigma_a is the fraction of the projected area of a sphere with diameter Dr which is filled by a particle with a maximum dimension of Dr. For spherical particles sigma_a=1.
  • sigma_v is the fraction of the volume of a sphere with diameter Dr which is filled by a particle with a maximum dimension of Dr. For spherical particles sigma_v=1.
  • The user may specify different values for delta_a and delta_v when the particle size D is less than Dr.
    • Matlab function used to compute effective diameter, number density and liquid water content from effective_diameter_prime This function calls: another Matlab function."

    Crystal Presets

    Power laws describing the area and volume of ice crystals as a function have cystal size(max dimension) been extracted from the literature and are provided in these presets When two power laws are provided for two size ranges the transistion occurs at size = Dr. If more than two power laws are provided we have selected two of the power laws to cover the entire range of sizes.

  • Solid spheres Ice crystals assumed to be solid spheres of ice.
  • Fire II (Arnott 1994) Arnott, W. P. , Y. Dong, and J. Hallett, Role of small ice crystals in radiative properties of cirrus: A case study, FIRE II, November 22, 1991.(1994), J. Geophys. Res.,99, pp 1377.
  • Hex columns(Mitchell 1996) Mitchell, D. L., Use of Mass- and Area-Dimensional Power Laws for Determining Precipitation Particle Terminal Velocities.(1994),J. Of Atmos. Sci., 53, pp 1716. A third power law for particles larger than 300 microns has been eliminated and the 100 to 300 mciron law used to describe particles larger than 300 microns.
  • Bullet Rosettes(Mitchell 1996) Ref same as above. 'Bullet Rosettes, 5-branches at -42 C. We have extended the lower limit of application to 60 microns from the 200 micron limit specified by Mitchell. Below 60 microns we assume that the crystal shape does not change with size.
  • Stellar Crystals(Mitchell 1996) Ref same as above. 'Stellar Crystals with broad arms. Small adjustments were made to proportionality constants in order to make power laws match at 90 microns. Powers were left unchanged. The power laws for particles less than 90 microns exceed area and volume fractions of 1 for very small crystals. This should not have a large effect on results.