3.2 Instrumentation

Figures 3.4, 3.5 and 3.6 show in detail the measurement positions which have been arranged for this study. They are numbered P1 to P7. Table 3.2 lists the applied instrumentation for the reference measurements.

Table 3.2: Instrumentation for reference measurements at the Auditorium.

position		quantity	instrument	sampling
P1		wind velocity (3d)	Solent Research Ultrasonic Anemometer	21 / s $\star$
P2		horizontal rain intensity	Young tipping-bucket rain gauge 52202	2 / min
P3		horizontal rain intensity	Young tipping-bucket rain gauge 52202	2 / min
P3		duration of hor. rain	rain indicator	2 / min
P3	$\star\star$	raindrop spectrum	disdrometer (Parsivel M300 by PMTech)	2 / min

$\star$ One-minute averages of 21 samples per second are logged.

$\star\star$ Operational from October 1999.

From 11 December 1997 to 7 July 1998

Table 3.3: Instrumentation for measurements at the west façade of the Main Building.

position		quantity	instrument	sampling
P4		wind velocity (3d) at 75 cm from façade surface	Solent Windmaster 1086M Ultrasonic Anemometer	1 / s
P4	$\star$	driving rain intensity	driving rain collector TUE-I	2 / min
P5		driving rain intensity	driving rain collector TUE-II	2 / min
P6	$\star\star$	driving rain intensity	driving rain collector TUE-II	2 / min

$\star$ Rectangular catchment area of TUE-I: $A_{\text{catch}} =0.830 \times 0.635 = 0.527$ m$^{2}$.

$\star\star$ Operational from 2 March 1998.

From 17 July 1998 to 16 September 1998

position		quantity	instrument	sampling
P4		wind velocity (3d) at 125 cm from façade surface	Solent Windmaster 1086M Ultrasonic Anemometer	1 / s
P4		driving rain intensity	driving rain collector CTH	2 / min
P4		driving rain intensity	driving rain collector DTU	6 / min
P5		driving rain intensity	driving rain collector TUE-II	2 / min
P6		driving rain intensity	driving rain collector TUE-II	2 / min

From 16 September 1998 to 31 December 1999

position		quantity	instrument	sampling
P4	$\star$$\star$	wind velocity (3d) at 50 cm from façade surface	Solent Windmaster 1086M Ultrasonic Anemometer	1 / s
P4		driving rain intensity	driving rain collector TUE-II	2 / min
P5	$\star$	driving rain intensity	driving rain collector TUE-I	2 / min
P7		driving rain intensity	driving rain collector CTH	2 / min
P7		driving rain intensity	driving rain collector DTU	6 / min
P6		driving rain intensity	driving rain collector TUE-II	2 / min

$\star$ Till 25 February 1999, the catchment area of TUE-I was circular: $A_{\text{catch}} =$ 0.444 m$^{2}$. After that, its catchment area was reduced to $0.250 \times 0.250 = 0.0625$ m$^{2}$.

$\star$$\star$ From 14 September 1999 onwards the anemometer distance is 124 cm.

Reference quantities are measured at positions P1, P2 and P3, i.e. above or on the roof of the Auditorium of the TUE. The reference wind velocity is measured on the mast on the Auditorium at 45 m height (position P1). The reference horizontal rain intensity is measured on two locations on the roof of the Auditorium: a position at the north side (P2) and one at the south side (P3) of the Auditorium roof were selected to investigate possible spatial differences of horizontal rain intensities. At position P3 a rain indicator was also installed in order to measure the duration of horizontal rain and to switch the wiper of the TUE-II driving rain gauges (see section 3.2.5).

Driving rain is measured at positions P4, P5, P6 and P7. Positions P4, P5 and P7 are adjacent to each other, at (approximately) the centre of the west façade of the Main Building. Position P6 is at the north edge of the façade (its centre is at 1.88 m from the edge). All façade positions are situated 39 m from ground level (87% of the building height). The driving rain gauges were placed behind specially arranged openings in the façade. These openings were formed by removing the original glass windows and replacing them by a plywood board with an opening of 827 $\times$ 735 mm$^{2}$.

At position P4 an anemometer was mounted on a boom, fixed to the façade. The length of the boom is variable. This enables measurements of wind velocities at a distance of 0.25-1.5 m from the façade, which will be useful for comparison with CFD simulations.

Table 3.3 shows the instrumentation applied on the west façade of the Main Building. The instrumentation at the central façade position changed over time. Three periods are distinguished. During the first period (December 1997 to July 1998) the driving rain gauges TUE-I and TUE-II were installed at positions P4 and P5 respectively (figure 3.7). The second period was a short period during which two new driving rain gauges were tested. One of them came from the Chalmers University of Technology (CTH) and the other was developed at the Technical University of Denmark (DTU). In the third period (from September 1998 onwards, figure 3.8), the positions of the TUE-I and TUE-II gauges were interchanged with respect to the first period. The shape of the catchment area of TUE-I was changed from a rectangle into a circle. The distance between the anemometer and the façade was reduced to 50 cm (figure 3.8). During the third period, the CTH and DTU gauges were located at position P7 (i.e. next to P4 and P5), so that their readings could be compared with those of the TUE-I and TUE-II gauges for the full-scale comparison test of driving rain gauges.

Figure 3.7: The two driving rain gauges on the west façade of the Main Building during the first measurement period from 11 December 1997 to 7 July 1998. From left to right: TUE-II (position P5) and TUE-I (P4). The ultrasonic anemometer has been mounted below TUE-I. Photograph by Ben Elfrink.

Figure 3.8: The four driving rain gauges on the west façade of the Main Building during the third measurement period from 16 September 1998 to 25 February 1999. From left to right: DTU (position P7), CTH (P7), TUE-I (P5) and TUE-II (P4). The ultrasonic anemometer has been mounted near TUE-II. Photograph by Ben Elfrink.

Figure 3.9: The driving rain gauge (type TUE-II) on the north edge of the west façade of the Main Building (position P6). Situation from 2 March 1998.

The driving rain gauge at the north façade edge (P6) was operational since March 1998, and was always of the TUE-II type.

The disdrometer became operational on 1 October 1999. Its measurements went on till 7 January 2000. Except for analyses of the disdrometer data, we will not take the other data after December 1999 into account.

Altogether, the full-scale measurements comprised a period of two years (December 1997 to December 1999) and resulted in an almost complete set of data on reference horizontal rain, reference wind velocity and driving rain obtained by the TUE-II gauges at P4/P5 and P6.

In the following subsections the used instruments will be discussed briefly. See [van Mook 1998a] for details.

3.2.1 Ultrasonic anemometers

The ultrasonic anemometer on the mast on the Auditorium (position P1) is suited for measurements of wind speed spectra. The sample rate is 168 per second; the output sample rate is 21 per s. From this data 1-min averages of the three velocity components $U_1$, $U_2$ and $U_3$ (figure 3.10) are calculated and logged. The ultrasonic anemometer at the façade position P4 is a simpler version. It can sample with a rate of 9 per second, but it is configured so that every second a sample is taken and logged. Major advantages of the used anemometers are: no moving parts (and thus no maintenance), accurate (error $\sim$3%), no calibration, and easy communication via an RS232 serial line.

Figure 3.10: Anemometer axis systems of the Solent Research at position P1 (left) and the Solent Windmaster at position P4 (right).

3.2.2 Tipping-bucket rain gauges

For the horizontal rain intensities two standard rain gauges of the tipping-bucket type are applied. Figure 3.11 shows one of the two tipping buckets. The horizontal catchment area is circular and equals 200 cm$^{2}$. The typical resolution $V_{\text{tip}}$ is 2 ml per tip, which corresponds to a rain amount of 0.1 mm with the given catchment area. Their calibration was done by ourselves, yielding:

series number	$V_{\text{tip}}$ [ml]	position
#413	1.90 $\pm$ 0.14	P2
#414	2.21 $\pm$ 0.11	P3

Figure 3.11: Rain gauge (left) and rain indicator (right) at position P3 (roof of the Auditorium). Photograph by Ben Elfrink.

3.2.3 Rain indicator

The rain indicator is a device which only indicates whether there is precipitation or not. It consists of 6 sensor boards on which electrodes were deposited in a grid. A drop or snowflake which is lying on a board and touches the electrodes, is sensed by an electronic circuitry. In that case, heating resistors mounted on the rear of the sensor board are switched on in order to facilitate evaporation. We did not investigate the rate of evaporation. The sensor boards were manufactured by Conrad Electronics. Their size is $4\times4$ cm$^{2}$ and the distance between the electrodes is 1.5 mm. Note that the rain indicator can not distinguish between dew, rain, snow and other forms of precipitation. The rain indicator was installed on the roof of the Auditorium (position P3, figure 3.11), and therefore it detects horizontal rain. It serves two purposes: (1) it switches the wiper of driving rain gauges of type TUE-II on when it rains (see section 3.2.5), (2) it is used to measure the duration of rain events. The measurement of the rain duration by a rain indicator is more direct and more accurate than by a tipping-bucket rain gauge, which gives only a signal after a time when the bucket is full. The rain duration measurements will be used to obtain corrected horizontal rain intensity data from the tipping-bucket rain gauge data (see section 3.4.3).

3.2.4 Disdrometer

We call every device which measures raindrop spectra, a disdrometer. In literature one finds alternatives, like the spelling ``distrometer'' and the terms ``spectrometer'' and ``spectropluviometer''; sometimes ``disdrometer'' is restricted to the device developed by [Joss and Waldvogel 1967]. The following modern measurement methods exist (see [Laws and Parsons 1943] and [Wessels 1967] for the older filter-paper and flour methods):

• Joss-Waldvogel disdrometer [Joss and Waldvogel 1967]. This device consists of a sensor head with a catchment area of 50 cm$^{2}$. Drops impinging on the head cause a downward displacement which is sensed by a coil. The amplitude of the voltage pulse from the coil is a measure for the size of a drop. The signal-processing electronics classifies the pulses into 20 channels which correspond to 20 raindrop diameter ranges, as calibrated and specified by the manufacturer. The Joss-Waldvogel disdrometer is often used as a reference, because it was the first electronic instrument which was (and still is) commercially available,

• shadowgraph or imaging method. A laser beam, optically shaped into a rectangular or elliptical cross section with a relatively small height, illuminates a CCD or an array of photodiodes. Drops passing through the beam are sized by the shadow created on the light sensor. The principle was applied e.g. by [Knollenberg 1970] (with an array of photodiodes). The two-dimensional video-disdrometer of [Schönhuber et al. 1994] is equipped with two video cameras. In fact, the video cameras take pictures of the front and the side of each single raindrop, snowflake or other hydrometeor passing through the light sheet. This produces detailed information on size, shape and speed of hydrometeors,

• light attenuation method. Instead of sensing the whole shadow image of a raindrop, a single photodiode senses the total attenuation of the detected light. [Salles et al. 1998] describe a disdrometer (originally from 1969) which uses this principle and which was improved by them. Other such devices were developed by [Grossklaus et al. 1998] and by [Löffler-Mang and Joss 2000],

• Doppler radar method. In contrast to conventional weather radars, Doppler radars suited for raindrop spectrum measurements measure the spectral radar reflectivity in a relatively small air volume. The Doppler radar points vertically and therefore measures the falling velocity of the raindrops. With help of an assumed relation between drop size and terminal (falling) velocity (e.g. eq. 2.11), the raindrop mass flux spectrum through the horizontal can be calculated. Applications of this principle were described e.g. by [Sheppard 1990] and [Richter and Hagen 1997].

In our choice for a disdrometer, the directness of the drop size measurement principle, the (in)sensitivity to wind, the resolution in time and space, the easiness of use and the price were the most important criteria. The two-dimensional video-disdrometer of [Schönhuber et al. 1994] would have been the most ideal instrument, because it measures the sizes and velocities of the raindrops in the most direct way. It can even be used on a façade to measure the raindrop spectrum of driving rain. Unfortunately, its price exceeded far too much our budget. The Doppler radar by [METEK 2000] has the advantage of measuring in a relatively large air volume (50-250 m$^{3}$) which improves the resolution, but at the time it was being improved so that it could measure vertical wind velocities (to correct the falling drop velocities). We chose the optical disdrometer of [Löffler-Mang and Joss 2000] (manufactured as ``Parsivel M300'' by [PMTech 2000]), although its development was not yet totally finished. Its main advantages are the direct measurement principle and its easiness of use.

The characteristics of the disdrometer are the following ([PMTech 1999], [Löffler-Mang and Joss 2000]):

• A 780 nm laser diode is the source of a horizontal light sheet, 27 mm wide, 1 mm high and 180 mm long. In the receiver the light sheet is focussed on a single photodiode. The decrease of the output voltage, due to a passing drop, depends linearly on the fraction of the light sheet blocked. The amplitude of the signal is a measure of the drop size; the duration allows an estimation of the drop velocity.

• After signal conditioning and A/D conversion, the data processing unit determines the amplitude and duration of every signal (i.e. of every particle). By the manufacturer, an empirical relation between amplitude and particle size was determined. With help of this relation, the data processing unit determines the size of the detected particle. The combination of the detected particle size and velocity is used to determine whether it is a raindrop or an other hydrometeor (e.g. snowflakes fall much slower than raindrops and have larger sizes).

• Every 30 s, the data processing unit reports the measurements by outputting (a) the mean particle velocity in each size class, (b) the particle number concentration per size class ($n(D_j)$ in m$^{-3}$ mm$^{-1}$), (c) a code for the precipitation classification, (d) the horizontal rain intensity, and (e) several control parameters. The applied size classes are tabulated in table 3.4. The codes for the precipitation classification are explained in the manual [PMTech 1999]. Data of approximately a month can be stored in the data processing unit.

• A correction [Raasch and Umhauer 1984] is applied to estimate the real particle number concentration spectrum when coincidence of particles in the light sheet occurs at higher rain intensities. [Löffler-Mang and Joss 2000] calculated that an intensive drizzle with $R_{\text{h}}=$ 30 mm h$^{-1}$ would yield an error of 9% due to coincidence; a convective shower with $R_{\text{h}}=$ 300 mm h$^{-1}$ would yield 5%. Thus, errors due to coincidences are unimportant for normal rain.

Table 3.4: The 25 particle size classes of the Parsivel M300 [PMTech 1999], indicated by the centre diameter, $D_j$, and the interval, $\Delta D_j$.

$D_j$	$\Delta D_j$	$D_j$	$\Delta D_j$
[mm]	[mm]	[mm]	[mm]
0.146	0.146	3.942	0.730
0.292	0.146	4.745	0.876
0.438	0.146	5.767	1.168
0.584	0.146	7.008	1.314
0.730	0.146	8.468	1.606
0.876	0.146	10.293	2.044
1.022	0.146	12.556	2.628
1.214	0.292	15.330	3.066
1.533	0.292	18.688	3.650
1.825	0.292	22.776	4.526
2.190	0.438	27.813	5.548
2.701	0.548	33.945	6.716
3.285	0.548

The disdrometer was delivered in December 1998, but it took a year of testing and improvements (by the manufacturer) before the disdrometer became really operational on 1 October 1999. It was installed approximately 1.8 m from position P3 (figure 3.12). We did not do any calibration or adjustments; we will compare the rain intensity calculated from the disdrometer data with those of the tipping-bucket rain gauges in chapter 5.

Figure 3.12: Disdrometer (Parsivel M300) at position P3 (roof of the Auditorium).

3.2.5 Driving rain gauges TUE-I and TUE-II

There is no standard for the design of driving rain gauges. In literature details of used driving rain gauges are rarely given; [Frank 1973], [Flori 1990] and [Osmond 1995] are the exceptions known to us. No references on a comparison of different types of driving rain gauges were found, and hence no reference dealing with the systematical error of a given driving rain gauge type. This subsection deals with the design of driving rain gauges which were developed at the TUE. Details of the gauges were previously reported in [van Mook 1998a].

The most widely used driving rain gauges consist of:

• a collector (a shallow tray) fixed to the wall of a building. The drops hit the tray, drip downwards and are collected by:
• a drainage channel, which leads the collected rain water to:
• a reservoir or a water flux gauge. A water flux gauge enables the measurement of the instantaneous driving rain intensity.

It is obvious that the catch efficiency (and thus the measurement error) of a gauge depends on size, shape and finish of the gauge surfaces. One should prevent that drops remain on the collector or in the drainage channel and evaporate, that drops splash out of the gauge, and that the shape of the gauge causes extra wind disturbances.

It is very difficult, sometimes practically impossible, to imagine realistic tests for evaluation of every error. The main concern though was the reduction of remaining and evaporating drops. This idea was brought about by observations at the façade of the Main Building of the TUE that much of the drops simply remain on the window glass, and do not drip downwards during many driving rains.

The following requirements were considered for the design of the TUE driving rain gauges:

1. estimated driving rain intensity range: 0.05 to 2.0 mm h$^{-1}$ at least,
2. sampling rate: 1 per min,
3. practical catchment area limited by the window size of the Main Building: approx. 0.5 m$^{2}$,
4. estimated maximum collected driving rain sum during 3 consecutive days: 5 mm,
5. short and straight drainage path to direct the collected raindrops into the water flux gauge,
6. hydrophobic coating to decrease the number of drops remaining on the collector or in the drainage path,
7. provisions preventing that the wind blows into the reservoir and preventing that the wind affects the desired flow of the drops into the reservoir (e.g. due to wind suction).

The first and fourth requirement are estimations obtained from hourly meteorological data of De Bilt (KNMI). The driving rain intensity range was calculated according to the principle of [Lacy 1965] and BS 8104 [BSI 1992]. The estimations appear to be quite good, as was confirmed by our measurements (see section 5.2.5; only two times in two years the 3-day driving rain sum exceeded 5 mm).

The first three requirements imply that the minimum amount of water, measurable within a minute for a driving rain intensity of 0.05 mm h$^{-1}$ and through a catchment area of 0.5 m$^{2}$ will be 0.5 ml. We use a balance, which can easily and accurately measure such small amounts of water.

The fourth requirement implies that, if the catchment area is 0.5 m$^{2}$, the driving rain gauge reservoir will collect at most 2.5 litres in three consecutive days.

Figure 3.13: Driving rain gauge TUE-I, with driving rain collector, wind deflector, reservoir (2 l) and balance. Left: front plate with the circular catchment area (0.444 m$^{2}$). Right: back plate and the inside.

Figure: Driving rain gauge TUE-II, with driving rain collector, wind deflector, reservoir (3 l) and balance. Left: front plate with the circular catchment area (0.492 m$^{2}$). Right: back plate and the inside. (See figure 3.15 for a detail of the wind deflector.)

Figure 3.15: Detail of the wind deflector of driving rain gauge TUE-II.

Figures 3.13 and 3.14 show the two types of driving rain gauges. Both types consist of a collector, a so-called wind deflector (to satisfy requirement 5 and 7, see figure 3.15), a reservoir and a balance (for the actual detection of the amount of water). Also for both types, all the inner sides have been coated with PTFE (teflon) to comply with requirement 6. The main difference is that driving rain gauge TUE-II is equipped with a wiper and driving rain gauge TUE-I is not. The wiper is basically a standard windscreen wiper for cars, and is automatically switched on by a rain indicator. The speed is approximately 1 rotation per 3 seconds; after every 5 seconds the wiper rests during 5 s to reduce wear and tear.

Figure 3.16: Spray test on driving rain collector TUE-II: $\includegraphics[width=0.5em]{gen/m-rondje.eps}$ without wiping and $\includegraphics[width=0.5em]{gen/m-kruisje.eps}$ with wiping. The tested collector was not yet exposed to the outdoor climate. A spray intensity of 1 g m$^{-2}$ s$^{-1}$ equals to 3.6 mm h$^{-1}$.

The wiper serves to improve the driving rain intensity measurement: drops on the collector are forced to coagulate and drip down. Furthermore, the wiper keeps the surface of the driving rain gauge clean. In the laboratory the driving rain collectors were tested for their collection efficiency. By a plant sprayer drops were sprayed onto the collector and the collected amount of water and the sprayed amount of water were measured. Also the time of spraying was measured. Results of this test (figure 3.16) show that wiping significantly decreases the dependence of the collection efficiency on the total sprayed amount and the spray intensity. The real effect of wiping can only be found by full-scale measurements, because the drop spectrum and the intensity of the used spray is different from real rain.

3.2.6 Driving rain gauges CTH and DTU

Apart from the TUE-I and TUE-II gauges the full-scale comparison test of driving rain gauges included two additional gauges, one provided by the Chalmers University of Technology (CTH) and another one by the Technical University of Denmark (DTU).

Figure 3.17: Driving rain gauge CTH. The catchment area is $0.18\times 0.18 = 0.032$ m$^{2}$.

Gauge CTH (figure 3.17) can be considered as a traditional driving rain gauge. It has a small catchment area (0.032 m$^{2}$), is made out of perspex and the collected rain flux is measured by a tipping bucket with a tipping volume of 1 ml. One tipping in 20 min represents a driving rain intensity of 0.09 mm/h. The gauge is described in [Högberg 1998] and [Högberg 2002].

Figure 3.18: Driving rain gauge DTU. The catchment area is $0.46\times 0.46 = 0.21$ m$^{2}$.

Table 3.5: Overview of the applied driving rain gauges.

type	principle	material	catch area
	min. intensity
CTH	traditional collector with tipping bucket ( $V_{\text{tip}}$ = 1 ml) $\frac{1\text{tip}}{10\text{min}}$ = 0.19 mm h$^{-1}$	perspex	$0.18\times 0.18$ $= 0.032$ m$^{2}$
DTU	collector weighted by a strain gauge ( $\Delta m \approx 1.3$ g) $\frac{1.3\text{g}}{10\text{min}} \approx 0.04$ mm h$^{-1}$	stainless steel	$0.46\times 0.46$ $= 0.21$ m$^{2}$
TUE-I	traditional collector with reservoir (2 l) and balance ( $\Delta m = 0.1$ g) $\frac{0.1\text{g}}{10\text{min}}$ = 0.001 mm h$^{-1}$	teflon coating	0.444 m$^{2}$
TUE-II	as TUE-I but with a rotating wiper and a reservoir of 3 l $\frac{0.1\text{g}}{10\text{min}}$ = 0.001 mm h$^{-1}$	teflon coating	0.492 m$^{2}$

The DTU gauge (figure 3.18) was designed as an improved driving rain gauge. As for the design of TUE-II, the main concern was reduction of measurement errors due to raindrops which remain stuck on the collector surface and subsequently are not measured in a flux gauge. The solution of the DTU gauge is weighting the whole collector, i.e. including the drops on the collector. The collector of gauge DTU is suspended freely from a strain gauge (though horizontal movements are prohibited). The collector consists of a stainless steel tray with a net mounted on the tray to reduce raindrop bouncing. A reservoir is integrated in the collector and is self-siphoning with a capacity of approximately 300 ml. Details are described in [Kragh 1998]. Obviously, the reading of the strain gauge is sensitive to wind fluctuations. This is partially compensated by averaging the reading during each 10-min period. The driving rain sum over a 10-min period is calculated by the difference of the mass of the collector in two subsequent 10-min periods. Only positive differences exceeding a threshold value of 1.3 g and during periods of rain according to a rain indicator, are taken into account. The mentioned threshold value approximates the resolution of the strain gauge and A/D converter.

In table 3.5 the main characteristics of the driving rain gauges are summarised.