You are on page 1of 11

Atmospheric Research 77 (2005) 18 28

www.elsevier.com/locate/atmos

Calibration of tipping bucket rain gauges


in the Graz urban research area
Vilmos Vasvari*
Institute of Urban Water Management and Landscape Water Engineering, Stremayrgasse 10,
A-8010 Graz, Austria
Received 31 March 2004; received in revised form 10 December 2004; accepted 16 December 2004

Abstract
The Institute of Urban Water Management and Landscape Water Engineering of the Graz
University of Technology (Austria) operates a hydrological research area in the City of Graz. In this
urban research area precipitation and runoff data are collected by order of the municipality of Graz. At
present precipitation data are measured by seven tipping bucket rain gauges. Comparative measurements have shown a deviation between the recorded and the actual precipitation intensity. This made
the institute calibrate the rain gauges periodically. In the middle of the 1990s, the development of a field
calibration kit was started. Based on the experiences with the first field calibration kit, a microprocessor
controlled device was developed. With this calibration device, the tipping bucket rain gauges are
calibrated at regular intervals. In this paper the calibration process and the current results for seven rain
gauges are discussed. The calibration process is dynamic calibration and uses a peristaltic pump. Not all
of the tipping buckets investigated underestimate the rain intensity in the whole measuring range.
Several rain gauges have a positive relative deviation, not exceeding 22%, in the low intensity range up
to 0.5 mm/min. Positive deviation can be explained by retention of water in the buckets between tips.
The reason for the negative deviation is the loss of water during the tips. It leads to the underestimate of
the actual intensity. The largest relative deviation in the range of underestimate exceeds 30%. In the
range of extreme intensities, the larger buckets (5 cm3) show a lower relative deviation than the smaller
(2 cm3) buckets. The gauge characteristic can change in favourable or unfavourable directions after
several years. Therefore, the calibration of tipping buckets is recommended at least every 2 to 3 years.
D 2005 Elsevier B.V. All rights reserved.
Keywords: Precipitation measurement; Tipping bucket rain gauge; Calibration; Field test

* Tel.: +43 316 873 8882; fax: +43 316 873 8376.
E-mail address: vasvari@sww.tugraz.at.
0169-8095/$ - see front matter D 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.atmosres.2004.12.012

V. Vasvari / Atmospheric Research 77 (2005) 1828

19

1. Introduction
Tipping bucket rain gauges are the most popular recording rain gauges used by many
national weather and hydrological service agencies. High accuracy of recording low-tointermediate intensity rainfalls, a superior mechanism for actuating circuits, suitability for
remote recording and reliability distinguish tipping bucket rain gauges. This type of gauge
produces rainfall data in digital form which can be readily processed by computers. On the
other hand tipping bucket rain gauges are known to underestimate the rainfall at higher
intensities because of the loss of rainwater during the movement of the bucket. At low
intensities the recorded intensity may exceed the actual intensity (Marsalek, 1981).
Previous publications discussed the non-linearity of tipping bucket rain gauges, the
reasons of this phenomenon, and the need of dynamic calibration. Marsalek (1981) dealt
with the surface tension of tipping bucket material at volumetric calibration, and with the
duration of bucket movement as a function of actual intensity. Durations range between
0.3 and 0.6 s for the instruments analysed.
Comparing the calibration results of 21 rain gauges, Luyckx and Berlamont (2001)
found that the tipping time has an approximately constant value and it amounts to 0.45 s in
the intensity range between about 15 and about 200 mm/h for all the tested devices. In this
middle region of the calibration curve, the error is proportional to the amount of tips in a
given time period and to the tipping time. The relationship between the rain gauge
resolution and the slope of its regression line allows calculating the slope of the regression
line and therefore one simple calibration test is enough to determine the complete
calibration curve of any tipping bucket rain gauge. However the latest investigations by
Vaes and Berlamont (2004) showed that the tipping times tend to have an exponential
variation with the actual intensity which results in a slight improvement of the calibration
for actual intensities up to 540 mm/h.
Niemczynowicz (1986) investigated tree different types of tipping bucket rain gauges
and assumed a simple power equation to fit the measured data, where the correlation
coefficients amount to over 0.999. Comparing the errors in calculated rainfall intensity
when using a constant bucket volume for linear and non-linear calibration, Niemczynowicz found that the error varies in the intensity range between 0 and 5 mm/min from about
19% to about 10%.
To reduce the independency of the measurement of the actual rain intensity, a siphon
can be installed between the bucket and the funnel, below the funnel outflow. However, as
both siphon and bucket are discrete samplers, the relative size of the two volumes is
important. Overgaard et al. (1998) investigated the optimal ratio of the effective volume of
the siphon to the volume of the bucket. They found that the optimum size of the siphon
volume has to be less than the half size of the bucket. Considering the dynamic volume of
the siphon, the effect on the volumetric accuracy should be assessed based on a volume of
the siphon of 35 to 45% of the bucket volume.
Labarbera et al. (2002) studied the statistical influence of systematic mechanical errors
on, for example, the disaggregated rain data, the Gumbel distribution of rainfall extremes,
and the depthdurationfrequency curves and came to the conclusion that these errors
affect substantially the derived statistics. Furthermore the equivalent sample size was
derived, which quantifies the equivalent number of calibrated data that would be needed to

20

V. Vasvari / Atmospheric Research 77 (2005) 1828

achieve the same statistical uncertainty introduced by the influence of systematic


mechanical errors in uncalibrated data sets.
However, although most authors ascribe the underestimation of tipping bucket rain
gauges to lost rain water, the overestimation at low intensities remains unexplained.
In the urban research area of Graz, precipitation measurements are carried out by a
network of seven automatic recording tipping bucket rain gauges, three of which have
been operating for 8 years (Hable, 2000). The network of rain gauges is used to measure
the total amount of precipitation and the amount of storm events within the urban drainage
area. Therefore it is important to calibrate the gauges carefully to ensure correct
measurements of both high intensities and total volumes. The knowledge of precipitation
contributes to an optimal operation of sewers.
The aim of the present investigations was to carry out control measurements on these
rain gauges to find out deviations between recorded and actual rain intensities and to
correct the recorded data. The deviation is presented in the form of calibration curves
serving as instrument characteristics. By means of these curves or correction factors, the
measured intensities can be corrected to approximate the actual values. Because of the
large amount of data, the correction is possible only by using software.

2. Calibration process
In the following section the rain gauge being calibrated, the calibration device and the
calibration process itself are briefly described.
2.1. Rain gauges
The rain gauges used in the Graz research area consist essentially of four components:
collector funnel, tipping bucket, data recorder and collecting receptacle. The collector
funnel area of the two different types of gauges used in Graz varies between 200 cm2 and
500 cm2. Accordingly the volume of the tipping buckets amounts to 2 cm3 and 5 cm3,
which corresponds to a precipitation depth of 0.1 mm. The data recording interval is timevariable, the time of each individual tip is registered as a binary signal in permanent
memory. The rainwater measured ends up in the collecting receptacle (Fig. 1).
2.2. Calibration device and process
A simple calibration device developed especially for field calibration consists of a
peristaltic pump with transparent hose. The calibration equipment serves for adjustment
and continuous supply of a constant flow rate. The central part of the device is a MCP-360
peristaltic pump. The pump with tubing of different diameters was calibrated first in the
laboratory in spite of the available rating curves given by the manufacturer of the pump.
The tubing of smaller diameter is suitable for intensities up to 1 mm/min and the larger one
allows intensities up to 7 mm/min. The flow is fed via transparent tubing to the funnel of
the rain gauge. The water flows into the tipping buckets and it is collected in a receptacle
below (Fig. 1). For the calibration, tap water is used. During the measuring process the

V. Vasvari / Atmospheric Research 77 (2005) 1828

21

Fig. 1. Sketch of the calibration device in action (after Bergmann et al., 2001).

electronic system records the signals produced by the tipping buckets. The flow rate of the
pump and the measured intensity of the simulated precipitation in two independent
measuring processes are recorded. The flow rate can be converted into a certain intensity
considering the funnel area (200 cm2 or 500 cm2).
Since only the revolutions per minute can be adjusted directly but not the pumping rate
of the pump, the revolutions per minute in steps of 10 are chosen. Because tips of whole
number only are measurable, the duration of measurement is determined by the number of
tips. The number of tips varies, according to the setting, between 5 and 15.
2.3. Evaluation of the calibrating process
The results of the measured data evaluation are presented in graphical and in tabular
forms. The actual intensity i 0 calculated from the flow rate and the recorded intensity i
determined from the frequency of the tipping bucket are determined. Furthermore the
relative deviation d [%] of the recorded intensity i from the actual intensity i 0, related to
the actual intensity i 0, is calculated by the following formula:
d

i  i0
 100 %
i0

The correction factor c [] is calculated by the formula:


c

i0
i

22

V. Vasvari / Atmospheric Research 77 (2005) 1828

By means of the correction factor c, the actual intensity i 0 can be directly determined from
the recorded intensity i as follows:
i0 c  i

where the correction factor c = c(i) is given as a function of recorded intensity (Vasvari,
1995).
2.4. Correction of the measured precipitation data
The correction of the measured data is carried out by means of the calibration
curves (described below) as follows. In the recorded data file the times of each tip are
recorded. One tip, a row in the data file, is equivalent to a volume of 2 cm3 or 5 cm3
depending on the type of the tipping bucket and corresponds to a precipitation depth
of 0.1 mm. From the interval between two successive tips, an intensity value can be
derived. The determination of intensity is erroneous at the beginning of a rainfall
event because the initial degree of bucket filling is unknown. The previous rainfall
event does not have to end with a tip. One of the buckets can be partly filled and
therefore this volume in the bucket represents a surplus at the first tip of the next
rainfall event.
The measured intensity is corrected to the actual intensity by means of the calibration
curve and then the corresponding bucket volume and the corrected precipitation depth,
respectively, are calculated. The times of the tips remain unchanged.

3. Results of calibration
In the study two different makes of tipping bucket rain gauges were tested. The
instruments included Horvath (Austria) rain gauge (collector area 500 cm2) and Young
(USA) tipping bucket rain gauges (collector area 200 cm2). The former have been
working since 1995, the latter were installed in the summer of 2003. From the
calibration point of view the intensity ranges are defined as follows: 0 to 0.5 mm/min.
low, 0.5 to 3.0 mm/min. high, and over 3.0 mm/min. extreme. It is known that an
intensity of about 1 mm/min (60 mm/h) already represents an extreme intensity. Under
alpine conditions, like in Styria, a precipitation event with an intensity of about 3 mm/
min has occurred. Therefore the rain gauges in this region have to be able to measure in
this and even in higher intensity ranges. That means the rain gauges have to be
calibrated also in the extreme intensity range.
3.1. Calibration curves and relative deviation
The first step after the calibration is the preparation of calibration curves. It is known
that a non-linear relation exists between the flow rate and the rate of a tipping bucket
gauge. That is because a variable quantity of water is lost during the tipping time (Calder
and Kidd, 1978).

V. Vasvari / Atmospheric Research 77 (2005) 1828

23

Calibration results are expressed in the form of calibration curves (Fig. 2), which are
usually assumed to follow the power function:
i0 a  ib

where a and b are the calibration parameters. If the correlation between the measured data
and the power function is not satisfactory, as it can be seen in Fig. 2, other functions,
mostly polynomials of the second order, can also be used. This function can be expressed
as:
i0 a  i2 b  i

The calibration parameters determined for the seven rain gauges are presented in Table 1,
in which r is the correlation coefficient. Polynomials of the second order have a better fit
for the measured data especially in the high intensity range.
Besides the power function and the polynomial of the second order which have a high
correlation coefficient but their fit can be piecewise not satisfactory, the linear interpolation
between the measured points is an obvious method. This method is a little bit more
complicate and time-consuming but using software the overhead is negligible. For the
correction in practice the replacement of the calibration curve by a number of regression
straight lines is recommended. In this way the accuracy of the data correction can be
increased. To obtain an adjustment as good as possible, 6 to 8 ranges and straight lines are
defined (Fig. 3). It is important to choose straight lines which have common starting and
end points. The reason for the approach of regression lines is that the error of estimation
based on one of the functions mentioned below can amount to 5% for the investigated
gauges. This deviation can appear in the high intensity range but also in any other ranges.
The advantage of the approach is the better fit to the measured points. However this
5.0

actual intensity [mm/min]

4.5
4.0
3.5
3.0
2.5
2.0

Klusemanngasse
Kepler Gymnasium
Andritz
St. Peter
Hohensinner
Lang
TU Graz

1.5
1.0
0.5
0.0
0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

3.0

3.5

4.0

recorded intensity [mm/min]


Fig. 2. Calibration results (recorded intensity versus actual intensity) and calibration curves using polynomials of
the second order (for parameters, see Table 1).

24

V. Vasvari / Atmospheric Research 77 (2005) 1828

Table 1
Calibration parameters of the different tipping bucket rain gauges
Rain gauge

Manufacturer

Hohensinner
Lang
TU Graz
Andritz
St. Peter
Kepler Gymnasium
Klusemanngasse

Horvath
Horvath
Horvath
Young
Young
Young
Young

Power function

Polynomial of the second order


2

1.1668
1.0282
1.0029
1.1980
1.3128
1.1485
1.1735

1.0974
1.1028
1.0785
1.1214
1.0985
1.1727
1.1142

0.9966
0.9977
0.9991
0.9971
0.9978
0.9968
0.9981

r2

0.0656
0.0669
0.0747
0.2127
0.1757
0.1940
0.1757

1.0787
0.9869
0.8916
0.9322
1.1103
0.9444
0.9736

0.9996
0.9975
0.9974
0.9994
0.9993
0.9981
0.9990

approach cannot characterise the general behaviour of the instruments and their
weaknesses and does not provide for the comparability of the instruments.
On the basis of the actual and the measured intensities, the correction factors c and the
relative deviations d were determined for all rain gauges. Fig. 4 shows the relative
deviations versus the actual intensity of the seven rain gauges. It can be seen that the three
rain gauges with the larger buckets (Lang, Hohensinner, TU Graz) in the high intensity
range (i N 1 mm/min) have lower relative deviations than the four rain gauges with the
smaller buckets. For the new rain gauges, the relative deviations exceed 20%. In the
range of extreme intensities, the relative deviation can increase even over 30%. In the
case of low intensities, the recorded intensity may exceed the actual intensity. In this range,
the largest relative deviation amounts to 22%. For intensities higher than 1 mm/min, the
recorded intensity is always lower than the actual intensity. The courses of the curves for
all seven rain gauges are quite similar. The curves have a general decreasing, but not

7.0
i0 = 1.9639i - 3.363
r2 = 1

actual intensity [mm/min]

6.0

i0 = 0.0669i2 + 0.9869i

i0 = 1.2773i - 0.1808

r2 = 0.9975

r = 0.9996

5.0
4.0
i0 = 1.3025i - 0.2355
r2 = 0.9995

3.0
2.0

i0 = 1.3613i - 0.3557
r2 = 0.998

1.0

i0 = 1.0836i - 0.0972
r2 = 0.997

0.0
0.0

i0 = 0.9042i
r2 = 0.9993

1.0

2.0

3.0

4.0

5.0

6.0

recorded intensity [mm/min]


Fig. 3. Approximation of the calibration curve by straight lines (rain gauge Lang). The error bars show the
estimated errors of the measurements.

V. Vasvari / Atmospheric Research 77 (2005) 1828

25

30
St. Peter
Andritz
Kepler Gymnasium
Klusemanngasse
Lang
Hohensinner
TU Graz

20

relative deviation [%]

10
0
-10
-20
-30
-40
-50
0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

actual intensity [mm/min]


Fig. 4. Relative deviation versus actual intensity.

monotonous, trend. From 0 up to 1 mm/min, the curves of relative deviation are steeper
than in the intensity range between 1 and 4 mm/min.
3.2. Time-dependent gauge characteristics

30
2003

20

1998

10
0
-10
-20
-30
0.0

1.0

2.0

3.0

4.0

actual intensity [mm/min]

5.0

relative deviation [%]

relative deviation [%]

For two rain gauges that were calibrated in 1998 as well as in 2003, it was also possible
to compare the curves of the relative deviations at different dates. Changes of the
calibration curves and therefore the curves of the relative deviations in 5 years are shown
in Fig. 5.
For the rain gauge Hohensinner in the low intensity range, a good agreement can be
seen. In the high intensity range, the relative deviation improved in 5 years, i.e. the error of
underestimation of the rain gauge decreased. For the rain gauge Lang, the accuracy in the
10
2003

1998

-10
-20
-30
-40
-50
0.0

1.0

2.0

3.0

4.0

5.0

actual intensity [mm/min]

Fig. 5. Relative deviation versus actual intensity for the Lang (left) and Hohensinner (right) rain gauges.

26

V. Vasvari / Atmospheric Research 77 (2005) 1828

low and in the high intensity ranges as well got worse. In the low intensity range, the
degree of overestimation whereas in the high intensity range the degree of underestimation
increased.
The influence of relative deviation on the precipitation depth of long periods (e.g. the
annual precipitation depth) is presented in Table 2. The measured annual precipitation data
from 1998 to 2002 were corrected with the calibration curve of 1998, whereas those of the
year 2003 were corrected with the actual calibration curve.
For the rain gauge Hohensinner, the values of deviation of annual precipitation are
roughly constant over the observation period. This happens because most precipitation
events occur in the low intensity range, where the deviation has not changed considerably.
The improvement of deviation in the high intensity range has no substantial effect on the
annual precipitation depth. The deviation of the annual precipitation depth essentially
corresponds to the mean relative deviation of the intensity between 0 and 0.5 mm/min.
In contrast to this, for the rain gauge Lang there has been a shift of the curve in the
direction of the positive relative deviation. Since the frequency of rain is high in this range,
the deviation appears clear in the annual precipitation depth. It can also be observed in Fig.
5. The calibration in 1998 shows in the low intensity range (up to 0.5 mm/min) a relative
deviation between 0.4% and 3.6%. It is equivalent to the mean deviation of the annual
precipitation depths. In this intensity range the relative deviation increased over 10%
which is reflected in the deviation of the annual precipitation depth increased to 13%.
In Table 2 the difference between corrected annual precipitation and annual
precipitation measured in the collecting receptacle excluding values with lost data
amounts to about 3%. The reason for this difference is the evaporation in the collecting
receptacle in which the volume of collected water is measured every 2 weeks. In 2000 for
the gauge Hohensinner, where the precipitation depth in the collecting receptacle is more
than the recorded after correction (difference 1.4%), it can be attributed to reading errors
at the collecting receptacle.
Table 2
Comparison of the annual precipitations recorded and corrected between 1998 and 2003 for the Lang and
Hohensinner rain gauges
Rain gauge

Year

Annual precipitation [mm]


Recorded

Lang

Hohensinner

1998
1999
2000
2001
2002
2003
1998
1999
2000
2001
2002
2003

933.5
(845.9)
(766.1)
(548.5)
892.6
(549.5)
853.3
(814.0)
768.6
514.8
(690.7)
607.4

Corrected
924.5
(838.1)
(754.1)
(537.4)
881.4
(485.7)
919.4
(873.5)
820.0
546.5
(733.7)
659.7

The values in parenthesis represent lost data.

Relative
deviation (%)

Annual precipitation [mm]


measured in the collecting receptacle

1.0
1.0
1.6
2.1
1.3
13.1
7.2
6.8
6.3
5.8
5.9
7.9

899.7
900.6
791.1
547.0
855.9
(579.1)
901.8
909.5
831.5
539.1
(823.9)
599.5

V. Vasvari / Atmospheric Research 77 (2005) 1828

27

4. Discussion and conclusions


For the rain gauges investigated in this study, the use of a power function as a
calibration curve would usually yield an insufficient adjustment especially in the range of
high intensities. Therefore, for a higher accuracy the calibration curves are composed of
regression straight lines. In this way the most accurate correction of the data can be
achieved assuming a correct calibration.
Through the analysis of the relative deviation of the rain gauges, it can be realized that
the tipping buckets underestimate the rain intensity but not necessarily in the whole range
of the measurement. In the range of the low intensities, the relative deviation may be
higher than zero and it can achieve maximal 22%. However, in case of extreme intensities,
the relative deviation can exceed 30%.
If the relative deviation is positive (i.e. the recorded intensity is greater than the actual),
it can be a result of either less water than the nominal volume coming into the bucket or
water remaining in the bucket after the last tip. After an accurate static calibration, the
former reason can be ruled out.
The reasons for the reserved water are surface tension and the adhesion on the surface
of the bucket. The surface tension depends on the temperature of the liquid and is lower for
tap water than in case of rain water (Marsalek, 1981). Therefore tap water causes a smaller
error at calibration. The volume of the adhesive retained water depends on the bucket
material. The older buckets made from metal can retain more water trough corrosion,
which raises the specific surface of the bucket. Plastic buckets are more water-repellent.
On the other hand, the turning moment of the retained water in the lower compartment has
to be overcome by adding some more water to the upper compartment. The ratio of the
time elapsed from the start of the bucket movement to the point in time when the
compartment being emptied does not receive any more water from the funnel to the
reduction in the rainfall depth increment is decisive to whether the relative deviation
moves into the positive range.
If the relative deviation is negative (i.e. the recorded intensity is less than the actual), it
originates from the water loss during the tipping time. In the range of high intensities,
during the bucket movement drops additional rain water in the bucket and in this way the
measured volume of water becomes more than the nominal bucket volume. In case of
larger buckets, the specific loss is lower than otherwise, therefore the greater buckets show
lower relative deviation at extreme intensities. The adhesive reserved water in this range is
negligible in comparison to the water loss.
It can be seen that, in the range of extreme intensities, the larger buckets have generally
lower deviations than the smaller buckets. Since both bucket sizes correspond to the same
precipitation depth, they tips with the same frequency, in the larger buckets occurs a
specifically lower loss than in the smaller buckets. That results from the measuring
principle of tipping buckets and cannot be avoided by adjustment.
For the investigated rain gauges the three regions of the calibration curves described by
Luyckx and Berlamont (2001) could not recognised. The slope of the calibration curves is
commonly greater in the low intensity range than in case of high intensities.
It could also be seen that the gauge characteristic can change in favourable or
unfavourable directions after several years. In spite of periodical cleaning of the buckets, a

28

V. Vasvari / Atmospheric Research 77 (2005) 1828

variation of the bucket wettability through surface oxidation or contamination by


impurities during the control periods can occur. Therefore, the calibration of tipping
buckets is recommended at least every 2 to 3 years. Further investigations have to show
with which rate the gauge characteristics change and during which period of time
significant changes happen.
However, by means of the calibration curves, it is possible to describe the present
instrument characteristic, to take into account the sources of error and to convert the
measured precipitation values into the actual values.
Appendices
a
coefficient of power function
b
exponent of power function
c
correction factor
d
relative deviation
i
recorded intensity
i0
actual intensity
r
correlation coefficient
a, b
coefficients of polynomial second order

References
Bergmann, H., Breinhalter, H., Hable, O., Krainer, R., 2001. Calibration of tipping bucket hyetographs. Physics
and Chemistry of the Earth. Part C: Solar-Terrestrial and Planetary Science 26 (1012), 731 736.
Calder, I.R., Kidd, C.H.R., 1978. A note on the dynamic calibration of tipping-bucket gauges. Journal of
Hydrology 39, 383 386.
Hable, O., 2000. Die Eichung von mit Wippe ausgestatteten Niederschlagsmessgeraten am Beispiel des
hydrologischen Versuchsgebietes Annabach in Graz. Mitteilungsblatt des Hydrographischen Dienstes in
sterreich 79, 35 51.
O
Labarbera, P., Lanza, L.G., Stagi, L., 2002. Influence of systematic mechanical errors of tipping-bucket rain
gauges on the statistics of rainfall extremis. Water Science and Technology 45 (2), 1 9.
Luyckx, G., Berlamont, J., 2001. Simplified method to correct rainfall measurements from tipping bucket rain
gauges. Urban Drainage Modeling, Proceedings of the Specialty Symposium Held in Conjunction with the
World Water and Environmental Resources Congress. American Society of Civil Engineers, Orlando, FL, pp.
767 776.
Marsalek, J., 1981. Calibration of the tipping-bucket raingauge. Journal of Hydrology 53, 343 354.
Niemczynowicz, J., 1986. The dynamic calibration of tipping bucket raingauges. Nordic Hydrology 17,
203 214.
Overgaard, S., El-Shaarawi, A.H., Arnjberg-Nielsen, K., 1998. Calibration of tipping bucket rain gauges. Water
Science and Technology 37 (11), 139 145.
Vaes, G., Berlamont, J., 2004. The use of standard tipping bucket rain gauges to measure extreme rainfall
intensities. Proceedings of the 6th International Conference on Urban Drainage Modelling-UDM04. Institute
for Urban Water Management, Dresden, Germany, pp. 591 598.
Vasvari, V., 1995. Kontrollmessungen zur Eichung der mit Wippe ausgestatteten Niederschlagsmessgerate.
sterreich 73, 55 66.
Mitteilungsblatt des Hydrographischen Dienstes in O

You might also like