Drop size distribution retrieval with video disdrometers and a UHF wind profiler

1
Drop size distribution retrieval with video
disdrometers and a UHF wind profiler

• Laura Kanofsky
• Advisor Phil Chilson
• 16 Feb 2006

2
Overview

• Background information
• Goal
• DSDs
• Equipment
• Retrieval method
• Current results
• Retrieved parameters
• Comparison between instruments
• Future work

3
Goal

• Retrieve the drop size distribution from a
  vertically pointed wind profiling radar

4
Goal

• Retrieve the drop size distribution from a
  vertically pointed wind profiling radar
• Why?
• DSDs are essential for calculating rainfall
  parameters from radar
• Assumed forms (Marshall-Palmer)
• Actual distribution

5
DSDs and rainfall parameters

• Reflectivity factor
• Rainfall rate
• Liquid water content

D is diameter, N(D) is the number of drops of
diameter D, rho is particle density, v(D) is
terminal fall velocity, w is vertical component
of air speed (positive upwards)
6
Characterizing DSDs with Dm and D0

• Median volume diameter, D0
• Mass-weighted mean diameter, Dm

7
Equipment Profiler

• UHF wind profiling radar
• Points vertically (range gates are heights above
  ground)
• 915 MHz useful for enhanced detection of the
  Rayleigh scattering component

A UHF profiler in its natural habitat
8
Equipment 2DVD

• 2-dimensional video disdrometer (2DVD)
• Drops fall through parallel light sheets
• Counts are binned by diameter range (e.g., all
  drops within .1-.2 mm)

9
Retrieval method
profiler desired quantities disdrometer
spectra
spectra
Z
noise reduction
density correction
assume a fallspeed relation
assume a fallspeed relation
Z
R
Dm
drop size distribution
drop size distribution
D0
10
Other available information
spectra
spectra
Z
soundings
noise reduction
2 disdrometers for some rainfall events
density correction
assume a fallspeed relation
assume a fallspeed relation
Z
R
Dm
drop size distribution
drop size distribution
D0
Locust image archive (archived surface maps and
satellite imagery)

Mesonet (rainfall)
KTLX radar images (Z)
11
Things to explore

• Effects of different choices in the intermediate
  calculations
• Various fallspeed relations
• Density corrections (standard atmosphere vs.
  sounding)
• Zero-ing out contribution from very small drops
  that don't exist (artifact of dividing by numbers
  near zero)
• Z from spectrum vs. Z from DSD
• Comparison between instruments
• Z, R, Dm, D0
• Lowest profiler gates
• How well does R match with the Mesonet?
• Variations with height

12
Current results

• Comparison of Z over time (profiler, disdrometer)
• Comparison of R over time (profiler, disdrometer,
  Mesonet)
• Stacked Z (profiler)
• Stacked R (profiler)
• Retrieved DSD
• Comparison of exponential model fit parameters to
  retrieved DSD
• Dm, D0 (profiler, disdrometer)
• Stacked Dm, D0 (profiler)
• Reflectivity gradients (profiler)

13
Range-time-intensity plot
14
Range-time-intensity plot
15
Comparison of Z
16
Comparison of R
17
Total rain for the 8 hr period
tipping bucket rain gauge, so the total may not
include rain that fell in the last 5 minute
period but wasn't enough to tip the gauge
18
Range-time-intensity plot
19
Stacked Z with 2DVD anchor point
20
Stacked R with 2dvd anchor point
21
Retrived DSDs from the profiler
22
Parameters from an exponential fit
23
Parameters from an exponential fit
24
Parameters from an exponential fit (profiler
gates 2(blue) and 3(red))
pay no attention to that units error behind the
curtain
25
Dm comparison
26
Dm stacked plot
27
Reflectivity gradients

• "Reflectivity dependence of reflectivity
  gradients observed by radar profilers" (W. Clark,
  C. Williams, P. Johnston, K. Gage, A. Tokay),
  P6R.6 from October's radar conference
• Calibration of vertically pointed radars by using
  a disdrometer
• How precise is the calibration?
• Limited by changes in the DSD in the last few
  hundred meters to the surface
• Degree of disagreement between profiler and 2DVD
  depends on reflectivity
• Vertical gradient of reflectivity

28
Near-future work

• Continue looking at more data for other rain
  events (convective vs. stratiform)
• Continue looking into reflectivity gradients
• Fit with modified gamma instead of exponential
• Filters
• Implement a lowpass filter to isolate the clear
  air signal
• Use clear air signal to estimate vertical
  velocity
• Modeling in velocity space
• Given an assumed model for the DSD, how would it
  be manifested in the velocity data?
• Possible to fit a model and get the parameters
  directly from velocity data?
• Any advantages to tucking the velocity-to-diameter
  conversion into the math?

29
Questions?
30
Mesonet rain gauge
31
Issues with D0
clipping-like effects possibly due to extremely
large numbers of small drops