Convective–stratiform separation using video disdrometer observations in central Oklahoma – the Bayesian approach

• Precipitation microphysics are characterized with 2-Dimensional Video Disdrometer (2DVD) data, collected in central Oklahoma.
• Precipitation events are separated as convective (CO) and stratiform (ST) rain by applying a multi-variable Bayesian classification algorithm to the 2DVD dataset.
• The corresponding shape-slope (μ-Λ) relations of the constrained gamma (CG) distribution are derived for these two rain classes to improve rain retrieval.

Application of 2-Dimensional Video Disdrometer (2DVD) data, collected in central Oklahoma, to the problem of convective-stratiform rain separation is presented. The partition into convective (CO) and stratiform (ST) periods is achieved by applying a multi-variable Bayesian classification algorithm to the 2DVD dataset. It turns out that the CO-ST separation methods developed for measurements with one type of disdrometer may not work optimally on measurements with a different type of disdrometer. Similarly, single/dual parameter, or simple threshold separation methods may not be able to adequately separate CO and ST rain types. The corresponding shape-slope (μ-Λ) relations of the constrained gamma distribution are derived for these two rain classes. These constrained gamma relations are then used for rain drop size distribution (DSD) retrievals, and the results are compared with those obtained from the exponential distribution and the unified μ-Λ constraint previously proposed. It is demonstrated that the results based on the convective-stratiform separation yield more accurate DSD retrievals with respect to the exponential distribution and moderate improvements in comparison to unified μ-Λ constraint.