The Inversion of Aureole Measurements to Derive Aerosol Size Distributions

J. T. Twitty Department of Meteorology, University of Wisconsin Madison 53706

Search for other papers by J. T. Twitty in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

An iterative method to invert size distributions from simulated scattered radiance measurements at small angles from the sun has been investigated. The inferred size distributions were represented by piecewise linear and cubic spline functions. Various relevant characteristics were investigated and it was found that:

  1. The inverted size distribution was insensitive to the number of knots in the piecewise linear spine.

  2. Within the range of sensitivity, the choice of initial guess had little effect on the inverted size distribution.

  3. Five percent random noise in the simulated radiances appreciably deteriorated the result but variations are still tolerable when compared with other methods for determining size distributions.

  4. Ale inverted distribution was insensitive to the index of refraction used in the kernel for particle radii r>1 µm.

  5. The choice of wavelength between 0.40µm and 0.70µm has negligible effect on the inverted distribution.

  6. A range of tropospheric aerosol size distributions give acceptable inverted results.

  7. The cubic spline representation can give reasonable inverted distributions, but may become unstable.

Abstract

An iterative method to invert size distributions from simulated scattered radiance measurements at small angles from the sun has been investigated. The inferred size distributions were represented by piecewise linear and cubic spline functions. Various relevant characteristics were investigated and it was found that:

  1. The inverted size distribution was insensitive to the number of knots in the piecewise linear spine.

  2. Within the range of sensitivity, the choice of initial guess had little effect on the inverted size distribution.

  3. Five percent random noise in the simulated radiances appreciably deteriorated the result but variations are still tolerable when compared with other methods for determining size distributions.

  4. Ale inverted distribution was insensitive to the index of refraction used in the kernel for particle radii r>1 µm.

  5. The choice of wavelength between 0.40µm and 0.70µm has negligible effect on the inverted distribution.

  6. A range of tropospheric aerosol size distributions give acceptable inverted results.

  7. The cubic spline representation can give reasonable inverted distributions, but may become unstable.

Save