Transcript Document

Spatial and temporal variability of
drop size distribution from
vertically pointing micro rain
radar (MRR)
Clemens Simmer1 and Malte Diederich1
Presented by Alessandro Battaglia1
1University of Bonn
Focus on continental
BL clouds
Baltex Bridge
campaign
BBC-2
Outlines
Back to short time scale
Whatraincan
MRR
rain radar-based
retrieval
target?
Cabauw, May 2003
What’s
next?
Achievements in
the campaign
MRR concept
•Investigation
of drop
distributions
and
•Instrument
between
•Study ofintercomparisons
small
scale size
variability
of vertically
prec.
field.
consequences
for the
relation between
Z and R
pointing
radars, WR,
disdrometer,
rain gauges
AQUARadar
SOP
Micro Rain Radars: toward a 4D-remote
sensing of the rdsd at sub-WR pixel
Main advantage of the instrument: it avoids sampling errors thanks to its
sampling volume (static=150-105 m3 depending on range) so bridging
between gauge and RR resolution
Original goal: to capture the DSD variability of hydrometeors in a
volume similar to a weather radar pixel (hence better 4D-understanding
of in-homogeneity in clouds and precipitation),
Possible applications:
•Better understanding in the whole process
of R-retrieval from Z measurements for
WR (development of adaptive/dynamic ZR conversion other than fixed power laws);
•enhanced
validation
method
for
Weather Radar
polarimetric weather radar;
Resolution Cell
Micro Rain Radar
Resolution Cell
•comparison and validation with spectral
microphysical models.
Micro Rain Radar MRR-2 concept
•24.1 GHz Low power FMCW (Frequency Modulated
Continuous Wave) Doppler radar;
•Beamwidth 2 deg;
•Range res 10-200 m (70 m);
•Time res 10 s – 1 h (30 s);
•Cost around 10.000 euros
http://www.meteo.uni-bonn.de/
Outputs: from MRR Doppler spectrum (after noise subtraction), rdsd
grouped in 43 classes with drop diameters from 0.249 to 4.6 mm are
estimated. Attenuation correction are applied after computing Mie
extinction from retrieved dsd. Radar reflectivity factor Z, rain rate R
and mean fall velocity (first doppler moment) W are then derived.
Output layout
We get the vertical profile of DSD below the cloud base
Instrument Layout during BBC-2
Twin net
Disdrometer
Better to use multiple instruments of the
same type: if carefully calibrated, this
should eliminate all instrumental biases
Intercomparisons of DSD measurements
MRR 1
MRR 2
MRR 3
2D-Video Dis.
Accumulations of drop densities in 0.2mm
drop diameter intervals in 5 rainy days
Comparison with 3 GHz-TARA
Example of strongly attenuated rain event at 1800 m height
Attenuation correction
Measured Ze
Z (DSD)
noise level
Other comparisons …
Variability of Z/R ratios and power laws
from MRR and Disdrometer DSDs
• Observation of the evolution of Z and rain rate to
form Z-R relations and „power laws“ at different
altitudes
• Special attention is paid to track dsd height
variation (possible causes & consequences for
weather radar estimates)
Simple characterization of precipitation by Z/R
ratios:
• High Z/R: most reflectivity contributed from large drops
• Low Z/R: most reflectivity contributed from small drops
Combined disdrometer -MRR analysis of a BB event
Line: Z=250R1.4
+, +, +: MRR-measurements at 200, 800 and 1500 m
+ disdrometer at ground level
Towards identifying different ‘‘physically
homogeneous’’ parts in a raining event …..
Analysis of shallow convection event
MRRs are immune
to horizontal wind
+, +, +: MRR-measurements at 200, 800 and 1500 m
+ disdrometer at ground level
MRR-2 vs De-Bilt RR sampled volume
MRR2 sampled volume must be reconstructed by 3-D
distribution of rain drops + wind advection
Overlapping the radar grid
Up to 10% of the RR volume (0.3 x 0.3 x 1 km3) covered by each MRR.
Despite synchronization problems with time stamp of De Bilt scan (not
better than 20 s) found good correlation (0.94) with WR Z.
Can spatial variability be resolved with MRR?
Correlation of drop-numbers in single Doppler-bins for 30-second measurements
C(mrr1,mrr2)
C(mrr1,mrr2)
C(mrr1,mrr3)
C(mrr1,mrr3)
C(mrr2,mrr3)
C(mrr2,mrr3)
It seems we can!! Correlation increases where there is wind advection
and spatial homogeneity (as seen by the RR)
Assessing the errors in radar R estimates within a single event
caused by spatial in-homogeneity at MRR scales of DSD
By averaging consecutive and spatially distributed MRR samples we can
mimic a larger volume. Therefore we can compare R accumulated during
each event computed at different spatial scales:
• directly from DSD  AP(dsd)
• from Z (DSD-derived) by different Z-R  AP(Z).
This
variability
In this
other event
is an indicator of
different spatial
how different
spatial sampling
samplings
are
equivalent
affect
the Z-based Restimate
Experience gained in BBC-2
• Relatively new instruments (not a deus ex machina!), a lot gained in
BBC-2: evaluated instrument precision/error sources in reflectivity,
DSD, rain rate, noise levels (should be below 0 dB), attenuation
problems in heavy rain, stability of calibration, better time
synchronizations.
• MRR-2 can be used to study vertical evolution of DSD (thus to
address dsd variations by coalescences, evaporation, break up, …)
• MRR-2 revealed as a useful tool for studying spatial in-homogeneity
at short time scale inside RR volume. Errors introduced by using a ZR relationship derived by a ‘‘point measurement’’ to a RR volume can
be studied.
• To get a deeper insight we need better spatial coverage and higher
time resolution.
RADAR
Advances in
Quantitative Areal
Precipitation Estimation by
Radar (proposed to the DFG)
Project Cluster
proposed by
Clemens Simmer, Susanne Crewell, Michael Griebel
Klaus Beheng, University Karlsruhe
Stephan Borrmann, Subir Mitra, MPI/University Mainz
Martin Hagen, DLR
Gerhard Peters, University Hamburg
Thomas Trautmann, Gerd Tetzlaff, DLR/University Leipzig
Peter Winkler, DWD
MRR contribute to AQUAradar SOP
SOP to be performed in southern Germany in an overlap area of 2
polarimetric radar (POLDIRAD and DWD radar in Hohenpeissenberg) A
wind profiler can measure and compensate the till now unknown error
source of vertical wind
•9-10 MRRs will provide better spatial
coverage with higher time resolution
(10 s): the volume distribution will no
longer have to be interpolated through
advection but can be measured directly
• Possibility of tracking rain shafts
•Is there any scaling behavior of rain?
Original goals seem achievable