You are on page 1of 10

130 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 17

An Optical Disdrometer for Measuring Size and Velocity of Hydrometeors


MARTIN LÖFFLER-MANG
Institut für Meteorologie und Klimaforschung, Forschungszentrum Karlsruhe, Universität Karlsruhe, Karlsruhe, Germany

JÜRG JOSS
Osservatorio Ticinese, Swiss Meteorological Institute, Locarno, Switzerland

(Manuscript received 15 September 1998, in final form 6 May 1999)

ABSTRACT
The characteristics of a prototype optical disdrometer are presented. Particles are detectable in the diameter
range from 0.3 to 30 mm having velocities of up to 20 m s 21 . Advantages of the new system are (i) it is easy
to handle, robust, and low cost, allowing a cluster of instruments to investigate the spatial and temporal fine-
scale structure of precipitation; (ii) it provides reliable detection of the range of small drops; and (iii) it allows
the possibility of snow measurements. Results of rain measurements are compared with data from a Joss–
Waldvogel disdrometer and a Hellmann rain gauge. Furthermore, some snow measurements are presented and
compared with results of a research spectrometer. The overall agreement is good. The repeatability of particle
size estimation was checked in the diameter range between 1.4 and 8.0 mm and yielded a standard deviation
of less than 5%. For drop velocities the standard deviation varies between 25% (0.3-mm drops) and 10% (5-mm
drops). The optical disdrometer can also serve as a present weather sensor, detecting and differentiating among
rain, snow, drizzle, graupel, hail, and the absence of precipitation.

1. Introduction niques. On the one hand, there are single particle coun-
For years scientists from many fields of research have ters, such as the forward scattering probe (Knollenberg
been interested in measuring the size and velocity of 1981), phase-Doppler instruments (Bachalo 1980; Dom-
particles. A large number of drop sizing instruments are nick et al. 1993), or extinction probes (Hauser et al.
described in literature. They can be divided into several 1984; Grossklaus et al. 1998). On the other hand, there
groups, depending on the physical principle used. are instruments that probe smaller or larger particle col-
Impact techniques are the basis of the first group of lectives by making use of Fraunhofer diffraction (Gerber
instruments. Early work with a filter method was done 1993; Lawson and Cormack 1995; Löffler-Mang et al.
by Diem (1956). A disadvantage of this method was the 1996; Löffler-Mang 1998) or by measuring the back-
large effort needed to evaluate the measurements. An- scatter of radar waves (Sheppard 1990; Rogers et al.
other instrument in this group is the well-known Joss– 1993; Löffler-Mang et al. 1999).
Waldvogel disdrometer (Joss and Waldvogel 1967). It In this paper the characteristics of a prototype optical
is widely used as a reference instrument for rain inves- disdrometer, based on single particle extinction, are pre-
tigations. sented. Particles are detectable in the diameter range
A second group is based on imaging techniques. between 0.3 and 30 mm, having velocities of up to 20
Some examples of these are the optical array probe m s21 . Attributes of the system we present are as fol-
(Knollenberg 1970), the three-dimensional holography lows.
(Borrmann and Jaenicke 1993), the pluviospectrometer 1) It is easy to handle, robust, and low cost; therefore
(Frank et al. 1994) based on a video camera, the video it is possible to install networks of disdrometers, thus
distrometer (Schönhuber et al. 1994) with two line-scan making ‘‘point’’ measurements more representative
cameras, and the particle spectrometer (Barthazy et al. and investigating small-scale variability.
1998), to mention just a few. 2) Small drops are reliably detected. This is of interest
A third group uses a large variety of scattering tech- for investigating scavenging and chemical effects.
Furthermore, by choosing different versions of the
optical system the measuring range of the sensor may
Corresponding author address: Martin Löffler-Mang, PMTech AG, be modified to also estimate the size of drizzle drops
Am Storrenacker 1a, D-76139, Karlsruhe, Germany. with diameters down to 0.1 mm.
E-mail: loeffler-mang@pmtech.de 3) Estimates of the size and velocity of snowflakes can

q 2000 American Meteorological Society


FEBRUARY 2000 L Ö F F L E R - M A N G A N D J O S S 131

TABLE 1. Specifications of the optical disdrometer.

Sensor head
Size d 5 225 mm, h 5 200 mm
Weight 5 kg
Laser diode wavelength 780 nm
power 3 mW
Light sheet size 30 mm 3 1 mm 3 160 mm
Measuring area 48 cm 2
Measuring range diameter 0.3–30 mm (0.1–10 mm)
velocity 0.1–20 m s21
Electronics
Size 250 mm 3 260 mm 3 150 mm
Weight 5 kg
Power supply 10–40 V DC or 100–230 V AC
Power consumption 10 W
A/D–Converter resolution 12 bit
sample rate 50 3 10 3 s21
Memory capacity (internal) 1 month of rain
Access, interface PC via RS 232

be obtained. This information is useful for ‘‘present


weather sensors’’ and for interpreting results from
weather radar systems in wintertime, especially in
alpine regions, where hydrometeors in the radar vol-
ume usually consist of snow.
Section 2 describes the instrument: general attributes,
the optical part of the sensor, the sensor housing, the
data analysis, and the accuracy achieved. In section 3
results from field operations are presented and compared
with data from other instruments. Finally, ideas for the FIG. 1. Signals of particles falling through the light sheet. (a) Small
future, instrumental development, and planned appli- and large particles, (b) raw signal from the sensor, and (c) inverted
cations are outlined in section 4. and amplified signal after thresholding for measuring purposes.

2. Measuring system shows the signals of two particles of different size. The
amplitude of the signal deviation is a measure of particle
a. Attributes of the instrument
size, the duration of the signal allows an estimate of
The disdrometer consists of an optical sensor within particle velocity. An appropriate concept to detect the
a housing and some appropriate electronics including start and the end of a signal is implemented in the soft-
solid state memory, which allows at least one month’s ware.
recording of rain data. Attributes of the instrument are Geometrical considerations show that the effective
summarized in Table 1. The following sections explain width of the light sheet depends on the particle size. To
the measuring system in more detail. be completely in the light sheet, larger particles have a
smaller region in horizontal direction. Therefore, to es-
timate concentration, the effective width for each par-
b. Optical sensor
ticle is taken into account.
The basis of the instrument is a commercially avail-
able sensor, producing a horizontal sheet of light (30
c. Sensor housing
mm wide and 1 mm high, 160 mm long). The light sheet
is produced by a 780-nm laser diode with a power of Two different protections have been tested. At first,
3 mW. In the receiver the light sheet is focused onto a a housing (Fig. 2a) of a shape similar to a Hellmann
single photodiode. The transmitter and receiver are rain gauge was analyzed for the rain measurements.
mounted in a housing for protection (see section 2c). In Then, for snow measurements a tunnel-like housing was
the absence of drops the receiver produces a 5-V signal used (Fig. 2b).
at the output of the sensor. Particles passing through the The Hellmann housing has been tested; for example,
light sheet cause a decrease of this signal by extinction the effects of wind were investigated in great detail by
and therefore a short reduction of the voltage. The volt- Nespor (1998). Furthermore, the Hellmann housing has
age decrease depends linearly on the fraction of the light a rather small outside dimension, producing a minimum
sheet blocked. Figure 1 (upper part) schematically of disturbance for rain, though only at vertical incidence.
132 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 17

by thresholding to detect the start of particles and their


maximum value. The shadow signal duration and the
time between two particles are also recorded. The last
three quantities are stored for each particle for further
calculation of distributions (size, velocity, energy, etc.)
as well as integral values (i.e., rain rate, radar reflectiv-
ity, liquid water content). The time needed for this anal-
ysis of one particle is less than 1 ms.
The optical sensor was originally designed for ex-
tinction measurements with signal durations of more
than 2 ms. For snowflakes this duration is exceeded;
therefore, the device needs no correction. The size of a
particle is calculated from the maximum reduction of
the signal (reflecting the blocked fraction of the light
sheet). The particle is assumed to be spherical with a
diameter corresponding to the width of the maximum
blocked area. For single snowflakes, which may have
rather complicated shapes, this assumption is less well
fulfilled than for raindrops. But for the ensembles mea-
sured during 60 s or more, the stochastical variation is
reduced.
Calibration is needed to determine the characteristics
of the internal electronics of the sensor. Special care is
needed for signals of less than 2-ms duration. Note that
typical sizes and velocities of rain drops lead to signal
durations between 0.4 and 1.1 ms. Because of limited
bandwidth of the off-the-shelf electronics used, the re-
duction of the measured signal amplitude (as compared
to the theoretical amplitude) has to be compensated. For
rain and hail measurements, the device was therefore
calibrated with particles of known size, falling at ter-
minal velocity. Twenty-seven different sizes of glass
spheres, ethanol, and water drops in size ranges from
0.275 to 4.29 mm were used. Signals of particles were
measured in the laboratory after a free fall in air of a
10-m height. From each size some hundred particles
were analyzed individually, then the median voltage for
the particle size was calculated. Thus an empirical re-
lation between size and voltage was obtained. This re-
lation is used for the evaluation of raindrops and hail.
Note that the calibration also takes into account the
FIG. 2. (a) Front and top views of optical disdrometer in Hellmann oblateness of larger raindrops.
rain gauge housing. (b) Side views of optical disdrometer in tunnel
housing. The particle velocity is calculated from the signal
duration. The signal starts with the particle entering the
light sheet and ends when the particle has completely
In the first housing the light sheet is folded by two exited the light sheet. The distance of influence of a
mirrors to keep the instrument size small. The tunnel particle is given by the light sheet thickness of 1 mm
housing makes it easy to assemble the sensor and to plus the particle diameter (see Fig. 1). The ratio of this
adjust the light sheet. Finally, at first approximation the distance and the signal duration yields the particle ve-
sensitive measuring area is independent of wind speed locity.
and direction. Because of the possibility of coincidences of particles
in the light sheet, a correction is applied to estimate the
real number concentration (Raasch and Umhauer 1984).
d. Data analysis
For normal rain, coincidences are not a problem. But
The analysis of the signal (Fig. 1, middle panel) con- for strong rain and for rain with a large number of drops,
sists of the following steps: removal of the DC part, coincidences may increase to a value that is not neg-
inversion, amplification, and filtering (see Fig. 1, lower ligible but is so small as to be easily corrected. Prob-
panel). Then a fast A/D conversion is done, followed abilities of coincidence were calculated for three ex-
FEBRUARY 2000 L Ö F F L E R - M A N G A N D J O S S 133

treme examples, considering drops with diameters be- 3. Results and comparison with standard
tween 0.3 and 5.5 mm. For an extremely strong strat- instruments
iform rain (with size distribution parameter N 0 5 8000 a. Rain measurements
m23 mm21 and rain rate R 5 100 mm h21 ) it is 10%,
for a most intensive ‘‘drizzle’’ (N 0 5 30 000 m23 mm21 , First measurements of raindrop spectra were con-
R 5 30 mm h21 ) 9%, and for an extreme convective ducted from May to July 1997. The optical disdrometer
shower (N 0 5 1400 m23 mm21 , R 5 300 mm h21 ) it is was placed 50 cm from a Joss–Waldvogel disdrometer
5%. Similar maximum values of coincidence probability on the roof of a small building in the Forschungszen-
are found in snow. trum Karlsruhe. To facilitate comparison of two in-
struments, the results of the optical disdrometer were
calculated in time intervals of 1 min for the 20 classes
of drop size of the Joss–Waldvogel disdrometer. For
e. Accuracy illustration a convective event on 21 May 1997 in the
early morning hours was chosen. For each instrument
the 10-min mean number density was calculated as a
A number of effects influence the accuracy of deter-
function of drop diameter (Fig. 3). There is good agree-
mining size and velocity. The homogeneity of the light
ment between the measurements of the optical and the
sheet was checked with a 1.0-mm-diameter wire. For a
Joss–Waldvogel disdrometer in the diameter range
mean signal of 130 mV, a variation of less than 65 mV from 0.7 to 2 mm.
was found when moving the vertical wire through the The case, however, shows differences in the range of
whole area of the light sheet. The noise superimposed small drops. Strong winds and/or acoustic noise may
on the signal is on the order of 63 mV. The A/D con- inhibit detection of the lower end of the drop size with
version has a resolution of 12 bits (4096 steps) for 10 the Joss–Waldvogel disdrometer (this device originally
V and a sample rate of 50 3 10 3 s21 . The sample rate was designed for the determination of Z–R relations).
limits the accuracy of the estimate of signal duration; This part of the spectrum has less influence on rain rate
that is, it mainly influences velocity errors. and radar reflectivity, but may be important for cloud
The repeatability was checked with two different sizes physics aspects and washout of pollutants by rain.
of steel balls (5.52 and 8.03 mm in diameter) and with For larger drops the concentration is rather low.
a small glass sphere (1.45 mm). The same particle was Therefore, the standard deviation is too large for a safe
thrown 100 times through the light sheet with a velocity interpretation (see error bars). Note also that the instru-
of approximately 1 m s21 . In a first experiment the po- ments did not measure the identical drops.
sition was always in the middle of the sheet near to the For 1 h the time series of the rain rate, calculated
transmitting sensor. Then, in a second experiment, the from the drop size distributions, is presented for both
particles were randomly passed through the whole mea- disdrometers in Fig. 4. It shows a strong convective
suring area. The first experiment yielded for all three event in the afternoon of 5 July 1997 between 1600 and
particle sizes a standard deviation of size determination 1700 CET with a duration of three-quarters of an hour,
below 3%, the second was smaller than 5%. reaching rain rates of 20 to 30 mm h21 . The quantitative
When using the 20 size intervals of the Joss–Wald- agreement between the two disdrometers is quite good,
vogel disdrometer to classify drops, the overall error in considering the accuracy of the instruments (around
estimating the diameter in the whole range of the in- 10% for the rain rate).
strument does not exceed 6100 mm plus 65%. For the The intercomparison of daily rain sums, measured by
velocity measurements of raindrops, the errors are with- both disdrometers and a Hellmann rain gauge, are com-
in 25% for the smallest drops (0.3 mm) and 10% for pared in the histogram of Fig. 5. For 10 days of the
time span between May and July 1997 data of all three
largest drops (5 mm).
measuring devices are available. For most of the days,
When calculating integral values such as rain rate and
the daily sums agree within 1 mm of rain amount. In 8
radar reflectivity, the instrumental errors are reduced by out of 10 cases the Joss–Waldvogel disdrometer showed
averaging, as verified by simulations. The stochastic slightly higher (10%) values than the optical disdro-
variation caused by the quantization of rain drops may meter and the Hellmann rain gauge. The agreement be-
exceed the instrumental errors. Its magnitude depends tween the last two instruments is not surprising because
on the sample size, an attribute valid for all drop sizing the optical disdrometer and Hellmann gauge also have
instruments (Smith et al. 1993). nearly identical housings with similar wind effects.
For these considerations misadjustment of the optics, When looking at daily rain sums, the observed sample
water or dust on sensor windows, drop sorting, or a of drops is large and stochastic variations caused by the
poorly defined measuring area caused by high horizontal quantization of rain by drops become negligible. Then
wind speed were not taken into account. These error the observed variations reflect systematic instrumental
sources can be reduced to a negligible size by proper differences. The variations of rain amount measured
installation and maintenance of the instrument. with the optical disdrometer are similar to those of the
134 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 17

FIG. 3. Mean spectral number density vs drop diameter, 0611–0620 CET 21 May 1997;
comparison of optical disdrometer (solid line) and Joss–Waldvogel disdrometer (dashed line).

Joss–Waldvogel disdrometer and the Hellmann rain imately 29 000 measured drops of 5 July 1997 is plotted.
gauge, known to be in the order of 10% when measuring Each drop results in a point of the diagram. The dashed
daily rain sums. Nespor (1998) discussed the influence line in Fig. 6 also shows the empirical relation from
of the wind. Atlas et al. (1973) after the measurements from Gunn
The optical disdrometer measures size and velocity and Kinzer (1949). The measurements scatter signifi-
of single particles, allowing a velocity–size correlation. cantly around the empirical curve. This scatter is caused
In Fig. 6, velocity versus drop diameter for the approx- by turbulence of air close to the ground influencing the

FIG. 4. Time series of rain rate during 1600–1700 CET 5 Jul 1997; comparison of optical
disdrometer (solid line) and Joss–Waldvogel disdrometer (dashed line).
FEBRUARY 2000 L Ö F F L E R - M A N G A N D J O S S 135

FIG. 5. Daily rain sums of 10 days during May–Jul 1997; comparison of optical disdrometer,
Joss–Waldvogel disdrometer, and conventional Hellmann rain gauge.

FIG. 6. Drop velocity vs drop diameter for 5 Jul 1997: each point represents the velocity and
the size of a single drop (total number of drops: ø29 000); comparison with empirical curve of
Atlas (1973) after Gunn and Kinzer (1949): y 5 9.65 2 10.3e20.6D .
136 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 17

FIG. 7. (a) Spectral number density vs snowflake diameter for 5-h interval, 0900–1400 CET
1 Mar 1998; comparison of optical disdrometer (solid) and ETH particle spectrometer (dashed).
(b) Mean velocity of particles in size classes vs snowflake diameter for the period in (a), comparison
of optical disdrometer (squares), ETH particle spectrometer (crosses), and empirical curve (dashed
line, Locatelli and Hobbs 1974).

fall velocity and, probably more important, by instru- The cutoff at low velocities results from the maximum
mental limitations, such as the uncertainties caused by value of detected signal durations, which was 2.54 ms
quantization and thresholding. The cutoff at approxi- (127 A/D conversion time steps) for rain measurements.
mately 0.3-mm drop diameter results from the trigger Some hundred data points can be seen in the range
level used to detect the drops (a trigger level is necessary directly above the cutoff line. These signals were mainly
to separate drop signals from noise); the line-by-line produced by drops splashing on the housing and, there-
structure of the data points comes from the 20-ms time fore, having unrealistically low velocities. They could
resolution of the A/D conversion. be removed by plausibility considerations.
FEBRUARY 2000 L Ö F F L E R - M A N G A N D J O S S 137

FIG. 8. Ten-min means of radar reflectivity vs time, 0900–1400 CET 1 Mar 1998; comparison
of optical disdrometer (solid), ETH particle spectrometer (dashed), and ETH X-band Doppler radar
100 m above ground level (dotted).

b. Snow measurements ing instruments estimate the velocities clearly above the
empirical curve, especially for the larger snowflakes.
During February and March 1998, a number of rain
This could be simply because the snowflakes investi-
and snow events were investigated in Linthal in the
gated in Switzerland were different from those found
Swiss Alps. On 1 March 1998 a predominantly strati-
by Locatelli and Hobbs. On the other hand, if the ob-
form snowfall was recorded with the optical disdrometer
(with tunnel housing), the ETH particle spectrometer served snowflakes were more broad than high (the ver-
(Swiss Federal Institute, Zürich; Barthazy et al. 1998), tical extent of the snowflakes is shorter than the mea-
and vertically pointing X-band Doppler radar (Mosi- sured width), this would result in an overestimate of the
mann et al. 1993). For 5 h the echo top was detected fall speed of particles larger than the thickness of the
by the radar at a height of 2.7 km above ground level. light sheet. Its thickness in the optical disdrometer is 1
A tipping bucket measured 5 mm of precipitation. mm, in the ETH particle spectrometer only 0.15 mm.
For both disdrometers spectral number densities were This difference may contribute to the difference between
calculated as functions of particle size. Distributions the two instruments and the dashed relation (for den-
were integrated over 10 min. In addition, the mean par- drites according to Locatelli and Hobbs 1974). It would
ticle velocity for each size class was calculated assuming also explain why the speed of smaller snowflakes es-
a spherical shape. The particle size was assumed to be timated with the optical disdrometer comes closer to the
the maximum horizontal dimension of the snowflake empirical curve than with the ETH particle spectrom-
seen. During the 5 h of observation both instruments eter.
showed similar behavior. Finally, radar reflectivity factors were derived from
For the intercomparison the mean size distribution the particle size distributions as suggested by Smith
for the 5-h period is shown in Fig. 7a, the mean particle (1984), with the assumption that the radar cross section
velocities in Fig. 7b. The number densities (size) of the of an irregular particle is the same as that of a sphere
two instruments agree well in most size classes. Dif- of the same mass (then he used an artifice from Marshall
ferences occur at the small diameter end, where the and Gunn 1952). When using the equivalent diameter
optical disdrometer was not sensitive in the first two of melted drops, the dielectric factor has to be multiplied
classes caused by the trigger level used. This error has by 1.18 to compensate for the density of water with
been recognized and corrected. At the large diameter respect to ice, resulting in the value 0.208 instead of
end only the optical disdrometer measured a single 0.176. For the mass determination of snowflakes again
snowflake in the size class of 12–14 mm. Compared to a relation according to Locatelli and Hobbs (1974) was
an empirical velocity–size relation y 5 0.8D 0.16 (Fig. used: M 5 0.059D 2.1 . Averages over 10 min of radar
7b), taken from Locatelli and Hobbs (1974), the esti- reflectivity versus time are shown in Fig. 8, together
mated velocities show significant differences. Both siz- with reflectivities from the vertically pointing X-band
138 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 17

FIG. 9. Schematic concept for using velocity and size information to detect the different types
of hydrometeors.

radar measured 100 m above ground level. The results Future work may include aspects of instrumental im-
of the disdrometers agree well with each other, but provement as well as the application of the instrument
sometimes there is a discrepancy of up to 5 dB of the to obtain information on the precipitation process.
values measured with radar. The difference may be
caused by some wetness of the snow or even more by 1) Instrumental work:
changes of the particle type. First estimations have R investigate the value of the velocity information
shown that the use of different mass–size relations (re- to identify and eliminate edge effects,
lated to different particle types) has a strong influence R estimate effects of drops splashing on the housing,
on the radar reflectivity derived from particle size dis- and
tributions. For a single size distribution, variations of R test a modified sensor to extend the measuring
up to 20 dB are possible. Obviously there is a need for range toward small drizzle drops.
additional research on how to derive radar reflectivity 2) Applications:
factors for snowflakes. R collect more experience in measuring rain and
snow,
R make use of the combined velocity–size infor-
4. Summary and future development mation for detecting the type of hydrometeor (see
An optical disdrometer was presented and results Fig. 9),
from rain and snow were compared with data from a R analyze strong precipitation events and deduce a
Joss–Waldvogel disdrometer, a Hellmann rain gauge, measure for soil erosion by large raindrops, and
and a particle spectrometer. The overall agreement is R investigate the representativity of point measure-
good. ments by combining the results of a number of
The error in determining size in the whole range does identical instruments operated simultaneously.
not exceed 6100 mm plus 65%. This was verified when
comparing drop concentration in the 20 size classes of Acknowledgments. We would like to thank F. Fiedler
the Joss–Waldvogel disdrometer. These error limits are for making the development of the instrument possible.
expected to also be valid for size classes above those Furthermore, thanks to E. Barthazy for preparing the
recognized by the Joss–Waldvogel disdrometer. For the snow measurements with the ETH particle spectrometer
velocity measurements, the error for the smallest drops and Doppler radar, and to J. Handwerker for simulating
(0.3 mm) yields values of 25% and goes down to 10% instrumental errors. Additionally, we would like to ex-
for the largest drops (5 mm). For daily rain sums the press our appreciation for the helpful comments of three
standard deviation between instruments is around 10%. anonymous reviewers.
FEBRUARY 2000 L Ö F F L E R - M A N G A N D J O S S 139

REFERENCES Their Formation, Optical Properties and Effects, P. V. Hobbs


and A. Deepak, Eds., Academic Press, 15–91.
Lawson, R. P., and R. H. Cormack, 1995: Theoretical design and
Atlas, D., R. Srivastava, and R. Sekhon, 1973: Doppler radar char-
preliminary tests of two new particle spectrometers for cloud
acteristics of precipitation at vertical incidence. Rev. Geophys.
microphysics research. Atmos. Res., 35, 315–348.
Space Phys., 11, 1–35.
Locatelli, J. D., and P. V. Hobbs, 1974: Fall speeds and masses of
Bachalo, W. D., 1980: Method for measuring the size and velocity
of spheres by dual-beam lightscatter interferometry. Appl. Opt., solid precipitation particles. J. Geophys. Res., 79, 2185–2197.
19, 363–389. Löffler-Mang, M., 1998: A laser-optical device for measuring cloud
Barthazy, E., W. Henrich, and A. Waldvogel, 1998: Size distribution and drizzle drop size distributions. Meteor. Z., 7, 53–62.
of hydrometeors through the melting layer. Atmos. Res. 47–48, , K. D. Beheng, and H. Gysi, 1996: Drop size distribution mea-
193–208. surements in rain—A comparison of two sizing methods. Meteor.
Borrmann, S., and R. Jaenicke, 1993: Application of microholography Z., 5, 139–144.
for ground-based in situ measurements in stratus cloud layers: , M. Kunz, and W. Schmid, 1999: On the performance of a low-
A case study. J. Atmos. Oceanic Technol., 10, 277–293. cost K-band Doppler radar for quantitative rain measurements.
Diem, M., 1956: Messungen der Größe der Regentropfen in natür- J. Atmos. Oceanic Technol., 16, 379–387
lichen Regen und bei künstlicher Beregnung (Measurements of Marshall, J. S., and K. L. S. Gunn, 1952: Measurement of snow
rain drop size in natural and artificial rain). Beitr. Naturk. Forsch. parameters by radar. J. Meteor., 9, 322–327.
Süddeutschland, 15, 75–90. Mosimann, L., M. Steiner, and W. Henrich, 1993: Prediction of snow
Domnick, J., C. Tropea, and R. Wagner, 1993: A miniaturized semi- crystal shape and riming by vertically pointing Doppler radar.
conductor fiber optic phase-Doppler anemometer (DFPDA) with Atmos. Res., 29, 85–98.
applications to liquid sprays. Meas. Sci. Technol., 4, 411–415. Nespor, V., 1998: Wind induced error of precipitation gauges. Op-
Frank, G., T. Härtl, and J. Tschiersch, 1994: The pluviospectrometer: erational use of radar for precipitation measurements in Swit-
Classification of falling hydrometeors via digital image proc- zerland. Final Rep. NRP 31, 91–101. [Available from the Swiss
cessing. Atmos. Res., 34, 367–378. Federal Institute of Technology Zürich, ETH Zentrum, CH-8092
Gerber, H., 1993: In-cloud measurements of effective droplet radius. Zürich, Switzerland.]
J. Aerosol Sci., 24, 583–584. Raasch, J., and H. Umhauer, 1984: Errors in the determination of
Grossklaus, M., K. Uhlig, and L. Hasse, 1998: An optical disdrometer particle size distributions caused by coincidences in optical par-
for use in high wind speeds. J. Atmos. Oceanic Technol., 15, ticle counters. Part. Charact., 1, 53–58.
1051–1059. Rogers, R. R., D. Baumgardner, S. A. Ethier, D. A. Carter, and W.
Gunn, R., and G. Kinzer, 1949: The terminal velocity of fall for water L. Ecklund, 1993: Comparison of raindrop size distributions
droplets in stagnant air. J. Meteor., 6, 243–248. measured by radar wind profiler and by airplane. J. Appl. Me-
Hauser, D., P. Amayenc, B. Nutten, and P. Waldteufel, 1984: A new teor., 32, 694–699.
optical instrument for simultaneous measurement of raindrop Schönhuber, M., U. Urban, J. P. V Poiares Babtista, W. L. Randeu,
diameter and fall speed distributions. J. Atmos. Oceanic Tech- and W. Riedler, 1994: Measurements of precipitation character-
nol., 1, 256–269. istics by a new distrometer. Proc. Conf. on Atmospheric Physics
Joss, J., and A. Waldvogel, 1967: Ein Spektrograph für Niederschlags- and Dynamics in the Analysis and Prognosis of Precipitation
tropfen mit automatischer Auswertung (A spectrograph for rain Fields, Rome, Italy.
drops with automatical analysis). Pure Appl. Geophys., 68, 240– Sheppard, B., 1990: Measurement of raindrop size distributions using
246. a small Doppler radar. J. Atmos. Oceanic Technol., 7, 255–268
Knollenberg, R. G., 1970: The optical array: An alternative to scat- Smith, P. L., 1984: Equivalent radar reflectivity factors for snow and
tering or extinction for airborne particle size determination. J. ice particles. J. Climate Appl. Meteor., 23, 1258–1260.
Appl. Meteor., 9, 86–103. , Z. Liu, and J. Joss, 1993: A study of sampling-variability effects
, 1981: Techniques for probing cloud microstructure. Clouds: in raindrop size observations. J. Appl. Meteor., 32, 1259–1269

You might also like