Parsivel correction

Updated correction (based on 2DVD+ MPS)

Recent work has demonstrated that the 2DVD does not properly sample the small drops and misses the “drizzle mode” (e.g. Thurai et al, 2018).

A new correction has been developed using a combination of DVD and MPS measurements in order to reconstruct (to the extent possible) the full DSD, including the drizzle mode. This is described in detail in Raupach et al., 2019.

Updated correction (based on 2DVD)

Raupach and Berne (2015) provides a method for correcting DSDs measured by Parsivel disdrometers to better match a collocated 2DVD. Since publication, several updates have been made to the correction factors. The latest correction factors are available for public use on this web site.

The factors are provided as .Rdata files that can be read by the R programming language. Each object is a list containing:

  1. filters: a list of correction factors for each class of rain intensity. Each list entry is a data.table containing the centre of the drop size class (dropSize [mm]), the class number (class) and the correction factor S (dropPerc).
  2. breaks: a list of the breaks in rain intensity classes [mm/h].
  3. maxSpreads: the maximum allowed spread for each class of rain intensity.

The latest factors can be downloaded here:

Updates made since publication of Raupach and Berne 2015 are the following:

  • Parsivel 2 data was supplied without zeros so it was not possible to know whether a missing time step was a zero or not available. The results in Raupach and Berne (2015) show Parsivel 2 factors trained with no zeros included in integrated DSDs. Zeros are now included (updated December 2016).
  • The calculation of the Parsivel sampling area was updated to fix a bug in which the width and height were reversed (updated January 2017).
  • The technique was retrained excluding hours for which the 2DVD misesd more than one minute of sampling time (updated April 2017).
  • The sampling area calculation was updated so that it uses $L(B-D)$ instead of $L(B-D/2)$ (see Raupach and Berne (2015) Eq. 5). This is because the Parsivel instruments we use automatically remove drops that touch the edge of the sampling area (updated September 2017).
  • The sampling area calculation for the 2DVD was updated to fix a bug found in the manufacturer’s code; the areas changed by less than 0.05% (updated September 2017).

We note the following points that were ambiguous in the original publication:

  • The results for DSD moment comparisons (Figures 12 and 13 and Tables 8, 12, A5, A6, and A7) show results comparing DSDs before and after the drop concentration correction was applied; thus the “before” results are DSDs to which the velocity correction and non-physical drop filter has already been applied.
  • The results for rain rate comparison show the difference made by the entire correction process, thus the before results are DSDs to which only the non-physical drop filter has been applied.

If you plan to use the correction factors in your work, or have any questions about the correction factors, please contact Alexis Berne or Tim Raupach.