University of Washington Disdrometer Data
The distribution of these data sets is funded by NASA's
Earth Science Enterprise. The data are not copyrighted;
however, we request that when you
publish data or results using these data, please
acknowledge as follows:
The authors wish to thank Dr. Sandra Yuter, Department of Atmospheric
Science, University of Washington for the production of these data and the
Distributed Active Archive Center(Code 610.2) at the Goddard Space Flight
Center, Greenbelt, MD, 20771, which archives and distributes them under
sponsorship of NASA's Earth Science Enterprise.
The condition of use on this data set is that an acknowledgement is
made that Dr. Sandra Yuter, Department of Atmospheric Science,
University of Washington is the source of the data. No co-authorship
is required or requested. If extensive use of these data is intended,
it would be valuable to correspond with Sandra Yuter to prevent
duplication of effort.
Description
Data are available from 0043 UTC 14 July 1999 - 1953 UTC 29 August 1999
representing 4327 minutes of measured rainfall and 3950 minutes of
rainfall > 0.1 mm/hr. Total accumulation was 418.7 mm.
The data file for the hour of 12 UTC on 14 July 1999 is missing and
not included in the data set.
All times are UTC.
Instrument is a standard Distromet Inc. Joss-Waldvogel disdrometer.
Instrument was located on the Aeromet building roof 3.7 m above
ground. Data were verified by comparing to drop spectra obtained
using the filter paper method [Rinehardt 1995].
Known Data Problems
Known problems with the data set are the loss of small drops from wind
effects and the noise from an air conditioner located about 15 ft from
the instrument. The noise from the air conditioner may have adversely
effected the ability of the acoustic-based instrument to detect drops
in the smallest two drop categories (< 0.5 mm diameter) as the
constant noise may have raised the noise floor slightly. Based on
discussions with Juerg Joss, the effect of this level of noise is hard
to quantify unless one has data collected in a truly quiet setting to
compare to. In the extreme case, in a very noisy environment, such as
the disdrometer deployment on Legan, where it was too loud to talk in
a normal voice and be heard, the disdrometer would only be able to
detect drops < ~1 mm in diameter. This is a previously documented
limitation of the JW instrument [Joss and Gori 1976].
A subsequent deployment of the same instrument to a quieter,
wind-sheltered spot on Kwajalein in summer 2000 has yielded many more
small drops than measured in 1999 during KWAJEX. Measurements in 2000
during periods when the instrument was exposed to SWerly wind indicate
that the wind effect (i.e. the divergence of small drops over the top
of the instrument associated with the blocked flow by the instrument
itself) appears to be the main culprit in the loss of drops < 0.5 mm
(smallest two drop categories).
It is recommended that users of these data utilize analysis techniques
that are less sensitive to the presence of small drops since the
number of drops < 0.5 mm in the KWAJEX data set are likely an
underestimate.
Quality Control
A quality control algorithm was applied to the raw data to remove
spurious data such as drips, insects and wind hits. The details of the
algorithm are in Parker and Yuter 2000. In brief, samples are removed
from the data set when over a 1 min sample there are any of the
following conditions: less than 25 drops, three embedded zeros between
categories containing drops, or a zero surrounded by two categories
each containing at least 30 drops.
A dead-time correction was applied to the processed data. The
particular correction applied is identical to the one provided by
Distromet Inc. in their DOS software and has been verified by Juerg
Joss. The correction formula in Sauvageot and Lacaux (1995) is
incorrect. Dead-time is built into the instrument so that splashes
(i.e. artificial drops) associated with a large drop hitting the
instrument are not counted. The dead-time correction is intended to
correct for loss of sensitivity of the acoustical instrument due to
noise associated with rain and to account for the loss of real small
drops that fall within the dead-time period. It will not correct for
losses due to non-rain ambient noise or wind effects. The correction
is intended to be correct to +/- 10%. To restate for clarity, under
ideal conditions when there is no wind and no ambient noise, the dead
time correction will correct the number of small drops to within +/-
10%. Under non-ideal conditions, as present during the KWAJEX
deployment, the number of drops <0.5 mm in size will be only
partially corrected i.e. only the noise of the rain and the dead time
are accounted for in the correction. The ambient non-rain noise and
wind are not accounted for in the correction. Hence the number of
small drops <0.5 mm is likely underestimated even after the dead-time
correction is applied.
Note further that the dead-time correction is not linear. The
correction applied to the raw counts obtained over 3 periods of 1 min
is different than the correction applied to the raw counts obtained in
a consecutive 3 min sample. Since the raw counts and hence processing
of the raw data are required to compute the dead-time correction, the
distributed data includes 3 versions of the processed data: 1 min
samples, consecutive 3 min samples and consecutive 5 minute samples.
Four different versions of the data have been provided to the DISC.
- Raw data files
Naming Convention: rdddmmyy.hh
where
dd=day
mm=month,
yy=2 digit year
and hh = hour
Description: These are the raw data files as
recorded by the Joss-Waldvogel disdrometer in the field. No quality control or
dead-time correction has been applied to the raw data. These files are intended
for use by PIs familiar with the JWD who have the software to process the data.
each line contains data in the following format:
"0426 2 3 18 78 85124117 50 8 2 1"
hhmm and then up to 20 counts for each of 20 size categories enclosed
in double quotes. Idiosyncracies of the format are that only 3
columns are used for the counts so that counts over 99 run into other
columns and that 0's are not included in the format for size bins
larger than the largest non-zero size bin. See Distromet Inc. manual
and software for further details.
The mean diameter of each of the 20 JWD size categories are as follows
based on the values in the Distromet Inc. software distributed with
the instrument (units mm): 0.359, 0.455, 0.551, 0.656, 0.771, 0.913,
1.116, 1.331,1.506, 1.665, 1.912, 2.259, 2.584, 2.869, 3.198, 3.544,
3.916, 4.35, 4.859, 5.373
- Quality controlled derived variables for 1 min averages when raining
These data files are for PIs interested in obtaining statistics on the
characteristics of the drop spectra for the _whole_ data set. For example,
average drop spectra and a Z-R relation have been obtained from these data
by Parker and Yuter (2000). Accumulated rainfall for the 1 min data set is
419 mm.
Naming convention: yymmdd.ND1wZR.qc
Description: Each line contains data in the following format for each 1 min DSD
1999 8 11 16 7 00 353.36 3705.68 1971.21 717.97 94.36 0.00 0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 12.39 0.41 49.87 8.70 90913.4
8 363
- year month day hour min sec
- N(D) - values for each of the 20 size categories, units mm^-3 mm^-1
- dBZ - reflectivity
- RR - rainrate, units mm/hr
- LWC - liquid water content, units mm^3/m^3
- delta - slope of exponential fit line units mm^-1
- No - intercept of N(D) for D=0, units mm^-3 mm^-1
- rawcounts -
- integer total number of drops hitting disdrometer in 1 min, used for quality control and computing uncertainty. This number represents
the raw counts recorded by the instrument, no corrections have
been applied.
NOTE: delta and No are _approximate_ fits computed using method of
Waldvogel (1974). It is recommended to use these _only_ qualitatively.
- Quality controlled derived variables for consecutive 3
min averages when raining
Naming convention: yymmdd.ND3wZR.qc
Description These data files are better suited than the 1 min data for obtaining
information on _individual_ drop spectra as the sample volume
associated with each spectra is larger. However, since only
consecutive 3 min data are used these data represent a subset of the
whole dataset. An example of use where 3 or 5 min data is preferable
to 1 min data are applications such as one by Chandrasekar who used
these data in an electromagnetic model to compute the distribution of
ZDR values. Accumulated rainfall for the consecutive 3 min average
data set is 406 mm.
- Quality controlled derived variables for consecutive 5
min averages when raining
Naming convention: yymmdd.ND5wZR.qc
Description: Same as 3) except for 5 consecutive minute periods. Accumulated
rainfall for the consecutive 5 min average data set is 394 mm.
Joss and Waldvogel 1967: Pure Appl. Geophys. 68, 240-246.
Joss and Gori 1976: Riv. Ital. Geofis., 3, 275-283.
Parker, W. S, and S. E. Yuter, 2000: Tropical open-ocean rainfall over
the Marshall Islands. in preparation.
Rinehart 1995: Preprints 27th Radar Meteor. Conf. 58-60.
Sauvageot and Lacaux 1995 JAS, 52, 1070-1083.
Waldvogel 1974, JAS, 31, 1067-1078.
The U.W. disdrometer data from KWAJEX may be accessed from this page,
 UW Disdrometer Raw Data Online
1,3 & 5 Minute Avg Data Online
or directly via ftp:
ftp disc2.nascom.nasa.gov
logon: anonymous
passwd:
cd data/KWAJ/surface_data/uw_disdrom/
The Principal Investigator for the disdrometer data is
Dr. Sandra Yuter
Research Assistant Professor
Department of Atmospheric Sciences
University of Washington
Box 351640
Seattle, WA 98195-1640, USA
yuter@atmos.washington.edu
Phone: (206) 685-1073
Fax: (206) 543-0308
For KWAJEX user services at the DISC contact:
For additional information, please contact the GES DISC User Services:
Last update:Thu Dec 4 14:40:11 EST 2003
|
KWAJEX
Home
