Evaluation of Rain Rate Measurement Methods, Sensors and Systems

Transcription

1 Evaluation of Rain Rate Measurement Methods, Sensors and Systems DIONISIS KANDRIS, GRIGORIS KALTSAS, CONSTANTINOS NOMICOS Department of Electronics, Technological Educational Institution (T.E.I.) of Athens, GR-12210, Athens, GREECE Abstract: - Rain rate is one of the physical quantities used for the description and analysis of climatic conditions. Its measurement can be performed by using various sensors and measuring systems. This paper presents the characteristics of the most popular measuring systems used worldwide which namely are tipping buckets, ground based radars, satellite systems, optical, acoustic, capacitance and pressure sensors and evaluates their operation. Key-Words: Rain Rate, Tipping Bucket, Radar, Distrometer, Satellite, Laser, Acoustic, Capacitance 1 Introduction Weather is of crucial importance for the evolution of many environmental and natural phenomena along with human activities, such as agriculture, transportation etc. That is why the weather forecast has always been a challenge for scientists. Weather forecast however, is based on the measurement and analysis of physical quantities such as air temperature, air humidity, barometric pressure, wind speed, wind direction, precipitation size, rain rate, and depth of snow. The measurement of such physical quantities is performed through the use of appropriate sensors and measuring systems [1]. For some of them, the principle of operation remains unchanged during time. For instance, the barometer was invented in 1643 for the measurement of atmospheric pressure. Modern barometers which are used nowadays have similar structure with that of former ones but with refinements. Other systems utilized in the past are no longer used, or hardly ever used. A typical example of such a system is that of the large kites used to carry in the air weather measuring equipment in the late 19th and early 20th centuries. Weather balloons have taken their place. On the other hand, there are some other systems, such as Doppler radar, which were invented in recent times. Among the most important physical quantities related with weather conditions, is that of rain rate. The rain rate is calculated by measuring the time interval between each measured rainfall increment of 0.01 inch. When there is rainfall within the archive period, the highest measured value is reported. When no rainfall occurs, the rain rate will slowly decay based on the elapse time since the last measured rainfall. 2 Rain Rate Measuring Systems The measurement of rain rate is performed by using alternative methods and systems which are presented in the following sections of this paper. 2.1 Tipping Bucket The tipping bucket is considered to be the most widely used weather instrument for monitoring rain rate [1]. A gauge of such a kind is shown in Fig.1. Fig.1 External and internal form of a tipping bucket The typical structure of such a unit is shown in Fig.2. It consists of a collector funnel, a filter and a tipping bucket mechanism. The operation is based on filling and dumping two buckets, one mounted on either end of a beam that pivots around a central axle in just like a playground seesaw. Rainwater gathered in the lid funnel of the gauge drops down into the upper bucket through a small-diameter hole in the lid's funnel-shaped bottom. When the bucket has received the proper amount of water, it empties itself by pivoting around the axle and transferring

2 the water into the base. This action raises the other bucket into position and the cycle repeats. Holes in the base allow the water to drain away. very difficult to site since they should be neither sheltered nor exposed. TB estimates suffer from significant errors if based on time scales less than 10 to 15 minutes. The errors are more significant at low rain rates. Moreover, a rain gauge makes what is virtually a point measurement. Since rain fields can have strong gradients particularly in convective storms a series of point measurements may give a poor reflection and an underestimate of the real rainfall. Fig.2 Structure of a tipping bucket mechanism The bucket is properly calibrated so that to tip when precipitation of 0.01 has been collected. Each time a bucket tips, an electronic signal is sent to a recorder. More precisely, as soon as a bucket empties its contents, it moves a magnet past a magnetic reed switch and causes it to close. In this way each tip is marked by a reed switch closure and stored on a data logger or transmitted by a telemetry system. Therefore, the rainfall for a certain time period is calculated by the simple multiplication of the number of marks on the recorder by 0.01 inches. The division of the product by the corresponding time period results to the rain rate for the specific time period. The simplicity of the tipping bucket gauge design provides a low cost solution. However, it is recognized that tipping bucket gauges suffer from several sources of uncertainty both systematic and random [2]. For instance, the funnel and tipping bucket mechanism should be cleaned periodically because very often accumulation of dirt and insects on the tipping bucket adversely affects the calibration and birds build nests in the funnel. That is why in remote and unattended locations dual tipping-bucket rain gauges are usually utilized as shown in Fig.3. Such platforms offer increased reliability given that the two essentially collocated gauges normally give measurements in very close agreement if there are no problems. However, tipping buckets are not free from bias problems related both to the intensity of the rainfall and the wind field around the gauge. In urban areas and mountain valleys, in particular, gauges can be Fig.3 Dual tipping-bucket gauge platform 2.2 Radar One major limitation of the measuring technique of many rain rate measuring systems such as tipping buckets is the wind dependent collection efficiency. It prevents quantitative measurements at wind exposed sites and is the main reason for the lack of data over sea. Another challenge for realistic monitoring of precipitation is its fractal spatial and temporal distribution. This calls for a hierarchy of in-situ and remote measuring techniques covering the extremely wide scale range of precipitation. An important element of such a nested observing system is a calibration strategy as most of the remote sensing techniques require some empirical parameters for their pertinent retrieval algorithms. Ground based weather radars are considered to fill the gap between in-situ sensors and satellitebased sensors with global coverage. Presently global rain information is provided by infra-red and passive microwave satellite instruments, with the Precipitation Radar (PR) only recently being added to the armory of available sensors [3]. However, as rain varies on such short spatial and temporal scales there is a need to incorporate all available measurements. Dual-frequency altimeters provide an estimate of the quantity of raindrops within the atmospheric column, through the observed attenuation of radar pulses. Use of a simple

3 threshold on the attenuation generates the familiar geographical patterns of precipitation, including their seasonal movement Because of their small footprint and fine spatial sampling along-track there is the potential to recover information on the size and structure of rain cells. Fig.4 shows MRR-2 a vertically pointing Doppler radar operating at 24.1 GHz with a sensitivity of 0.01 mm/h rain rate and a useful height range up to 3km. Fig.5, two planes of light transect the measurement area. As a rain drop falls into the sensor unit, it occludes each plane of light, thereby producing two side image shadows that are recorded by cameras nested within the instrument. The two orthogonal projections provide 3D raindrop shape information that is used to describe the raindrop. Fig.5 A rain drop optical image recording system Fig.4 A ground based weather radar However, apart from their high cost radars have other certain drawbacks [4]. Radar does not measure rainfall, but reflected energy, and there is no unique relationship between these two quantities. Since radar normally scans at an elevation angle of at least 0.5, it scans above the surface of the Earth with the volume scanned increasing with the square of its distance from the radar. Such scanning may be subject to obscuration due to the local topography, and the data collected will not be confined to reflections from rain. The intensity of the energy returned to a weather radar depends on the size of each rain drop. Furthermore, the number of rain drops at each size (for example, in a cubic meter of air) can vary significantly from cloud to cloud. This can make radar estimation of rainfall rate very difficult. There may also be clutter, energy reflected from the ground or insects, etc., and reflections from other forms of precipitation such as hail or melting particles. Nevertheless, radar can provide excellent spatial resolution over a wide area typically a circle of at least 100km radius. The structure of such a system is illustrated in Fig.6. An outdoor electronic unit receives video signals from cameras, pre-processes raw data and runs data acquisition and plane alignment software. The data collected are transmitted via TCP/IP to the indoor PC which performs the final data processing. Data processing enables the calculation of certain features such as rain rate. Moreover, the appropriate arrangement of the two cameras in different planes allows measurement of fall velocity, which is of special importance for mixed phase events. Another advantage of such a system is that only the inner part of the measurement inlet is taken for data processing while the outer part which is influenced by splashes from the edges is ignored. On the other hand the optical path may be obscured due to the presence of insects, dirt, or raindrop on mirror or faceplate or by raindrops hanging from awning D Video Distrometer The 2D-Video-Distrometer is a precipitation gauge which measures the size of rain drops by working on the basis of video cameras [5]. Its name comes from the fact that such a device records two, side view optical images of each raindrop. As it is shown in Fig.6 2D video distrometer structure

4 Fig.7 illustrates the outdoor electronic unit and the sensor unit of 2D distrometer. Fig.7 2D video distrometer control and sensor units 2.4 Satellite Rain Rate System Understanding the Earth s climate requires knowledge of the global distribution of precipitation since precipitation is a key factor in the hydrological cycle, which in turn controls the circulation of the atmosphere and ocean. Furthermore, measuring precipitation at sea is extremely difficult. Satellites provide part of the answer, particularly with respect to global monitoring [6]. Either microwave or infrared methods can be applied. Microwave methods have major advantages over infrared rain rate algorithms. First, microwave based satellite systems can see through the tall convective clouds to the surface whereas the infrared methods use the cloud top temperatures as a guide to surface rain rate. Second, they can see through high cirrus cloud shields that make up a large part of tropical convective cloud clusters. The data process provides rain rate images similar to that shown in Fig.8. Fig.8 Example of a rain rate image The emissivity of oceans is very low, although it does increase with increasing frequency. Clouds in the atmosphere appear warmer than the ocean surface because they emit more microwave radiation than the ocean. Cloud drops and raindrops are generally the largest contributors to upwelling radiation over ocean areas. If the rainfall is light and the storm is only mildly convective (so that ice particles are not found in the cloud tops), the brightness temperatures will increase with increasing rain rate. However, in convective storms where ice is found in the upper regions, scattering plays a major role. Ice particles scatter the higher frequencies more readily than the lower frequencies since ice particles are closer in size to the 85 GHz wavelength. Large storms with a high degree of vertical development not only generate high rain rates, but also have large numbers of ice particles in their tops. The larger the vertical development of the storm, the higher the rain rate and the number of ice particles in the upper atmosphere. The ice particles will scatter the upwelling 85 GHz. Rain rate algorithms over oceans take advantage of the reflective ocean background and can use any of the lower frequency channels to determine the ocean rain rate. The ability to use the 19 and 37 GHz channels enables algorithms to sense rainfall more accurately by "seeing" through the cloud to near the surface. These types of algorithms are referred to as emission-based algorithms since the lower frequency channels are sensitive to emission from the rain drops. Over land, the surface is highly emissive with a very low reflectivity. The land surface overwhelms any effect clouds or rain will have on the upwelling microwave radiation. Since the 19, 22, and 37 GHz channels can see readily to the surface in most situations, they become useless when trying to detect rain over land. The only useful channel is the 85 GHz that cannot see the surface when there is significant amount of cloud/liquid water present in the atmosphere. Therefore, land rain rate algorithms rely heavily on the 85 GHz channel to determine the rain rate. This type of algorithms is referred to as scattering-based algorithms because they use the scattering signature by ice particles at the top of the clouds to determine the rain rate at the surface. They are more unreliable than the emission-based algorithms described in the above paragraph and have a closer relation to the infrared retrievals. The 85 GHz has a penetration depth of only a couple of kilometers which is a little better than the infrared method. However, microwave methods can still see through the cirrus cloud shield whereas the infrared cannot. Satellites are limited by the relatively low frequency with which they can observe a specific region, and by their large spatial footprint. Rainfall can vary on such small spatial and temporal scales,

5 that satellite measurements are unable to resolve the diurnal variation, and regional rainfall differences become homogenized. Thus, there is a need for in situ measurements not only to resolve the spatial and temporal variation in rainfall but also to provide validation data for improving satellite products. Compared to a tipping bucket type rain gauge, the optical sensors have faster response, better time resolution given that the exact time when a rainfall starts can be specified and the ability to distinguish between rainfall and snowfall. 2.5 Laser Rain Rate Gauge Optical systems used for rain rate measurement are mostly based on optical forward scatter and back scatter techniques for analyzing water particles of different forms, like fog, rain or snow in the air [1, 7]. These two techniques are explained by using Fig.9 [8]. With reference to this figure, a laser beam that hits a small particle in the air is considered. The beam may also come from another light source like a LED. The particle is typically a very small water drop. A fraction of the laser light propagating will be scattered in all directions by a particle. The light that only has changed its propagation direction slightly is called forward scattered light and the light that has changed its direction around 180 degrees is called back scattered light. Fig.10 Forward scatter sensor structure Fig.11 Backward scatter sensor structure Fig.12 Water particles optical detection Fig.9 Forward scatter vs back scatter In an optical measurement system based on either of these two techniques there is a receiver and a transmitter. More precisely, a laser radiation source is used in order to generate a narrow, amplitude modulated collimated beam. The optical receiver consists of a lens, a detector and a phase locked amplifier. Fig.10 and Fig.11 illustrate the basic structure of a forward scatter and a backward scatter sensor correspondingly. Both these kind of sensors detect particles in a very small volume. The backscatter concept is the more compact since the transmitter and the receiver can be located close to each other in the same box. Fig.12 illustrates how water particles are detected by an optical rain rate system, while Fig.13 shows such a system. The disadvantage of the backscatter concept is that the optical power reaching the receiver is lower than that in the forward scatter case. This can be compensated by a receiver with very good signal to noise ratio. However the back scatter has simpler structure, lower power consumption and lower cost. Fig.13 Laser detection system 2.6 Acoustic Rain Gauge Precipitation is one of the major factors governing ocean climate. Despite the successful use of satellite-based observations in this area of remote sensing it is necessary to obtain in situ measurements to observe the high temporal and spatial variation in rainfall. One of the most promising new technologies for measuring rainfall at sea is an Acoustic Rain Gauge (ARG) [7, 9]. Rainfall measurement is achieved through sampling of the underwater sound spectrum between 500 Hz and 50 khz. More precisely, wind and rain generate noise within this specific

6 2.8 Pressure Rain Rate Gauge Another type of rain rate sensors uses a pressure- sensitive diaphragm to record the momentum of each impacting raindrop. The earliest designs used a flat horizontal membrane while more recently frequency range and furthermore the spectra generated are characteristic of the source, and louder than other sources (breaking waves, biology etc.) by several orders of magnitude. Thus, it is possible to discriminate as to the source of the acoustic signal, and use algorithms to make quantitative estimates of wind speed and rain rate. This sampling is carried out by the use of a sub-surface hydrophone, which avoids working at the hostile ocean-atmosphere interface. As soon as sound data are collected, sound intensity algorithms are used to classify the sound source and generate estimates of wind speed and rainfall rate. These algorithms are still experimental, and extended comparison with ancillary instrumentation is necessary to develop them further. investigators have proposed using a microphone with a spherical surface similar to that shown in Fig.15 [7]. Such an instrument can thus detect raindrops equally well from all directions. The terminal velocity of raindrops increases with size and wind speed, thus some calibration is required to infer mass of drop from the impulse it generates. A typical size for such a sensor is ~5 cm across, which gives a reasonable sensitivity in light to heavy rain. Because of the small surface area it is not reliable at very low rain rates, and there are significant errors in very heavy rains where the sensor cannot distinguish the impacts of individual drops. 2.7 Capacitance Rain Rate Gauge An alternative system for measuring rain rate makes use of the fact that electrical capacitance changes by the presence of water. A system of such a kind is based on an inclined capacitance plate, similar to the one shown in Fig.14, which has interwoven electrical contacts in order to provide instantaneous rainfall rates [7]. Raindrops on the plate affect the capacitance between the two terminals; gravity and/or a heating circuit remove the drops from the plate. Due to the inherent structure simplicity such gauges are of low cost. However, they are found to be inadequately accurate in high humidity conditions. Furthermore, at sea environment sea spray can lead to a coating of salt on the surface. Moreover, the plate needs to be directed into the wind. Fig.14 Inclined capacitance plate 5 Fig.15 Pressure-sensitive diaphragm sensor Conclusion This paper presented the most popular methods, sensors and systems developed for the measurement of rain rate. This presentation can be utilized for both engineering and educational purposes. More precisely, satellites offer the ability of global coverage. However they are limited by their relatively low scanning frequency and their large spatial footprint which is an obstacle for accurate local monitoring. Therefore in-situ sensors have to be additionally used. Ground based weather radars are considered to fill the gap between in-situ sensors and satellitebased sensors but given that they do not measure rainfall but reflected energy, their measurements can be affected by unwanted reflections. Despite recent advances in the use of remote sensing techniques traditional rain gauges such as tipping bucket gauges, are still the most widely used in hydrological and climatological applications. However, they may fail to provide accurate measurements due to the intensity of the rainfall and the wind field and malfunctions of the tipping mechanism operation. Optical sensors using either cameras such the 2D distrometer or laser beam which is either forward or backward scattered can be utilized. The

7 data collected by using such systems provide a complete description of rain drops and rain variability. Nevertheless, the optical path may be obscured. The use of capacitance sensors is simple and of low cost but shows to have inadequate robustness in extreme weather conditions. Pressure sensitive gauges are considered to be sufficiently reliable only in medium rate rainfalls Finally, acoustic methods seem to be more promising although they have to be further improved. Concluding, it has been apparent that no one of the existing methods and systems described above can be generally adopted as the ideal for accurate and reliable rain rate measurements. Therefore, in order to achieve better results, scientists work on the further improvement of their performance [10] and the appropriate combination of their use. Acknowledgements: This work and its dissemination efforts have been funded by i) the Greek Operational Programme for Education and Initial Vocational Training (O.P. Education) in the context of action entitled Reformation of Undergraduate Studies Programs and ii) the research programme Archimidis I of the Greek Ministry of Education. Oceanic Technology, Vol.19, No.5, 2002, pp [6] E. N. Anagnostou, and C. Kummerow, Stratiform and convective classification of rainfall using SSM/I 85-GHz brightness temperature observations. Journal of Atmospheric Oceanic Technology, Vol.14, 1997 pp [7] G.D. Quartly, T.H. Guymer and K.G. Birch, Measuring Rainfall at Sea.Part I, In Situ sensors, Rain_pt1.pdf. [8] Comparison of optical sensors, [9] G.D. Quartly, T.H. Guymer, K.G. Birch, J. Smithers, K. Goy and I. Waddington, Listening for Rain, Theory and Practice, Proceedings of the Fifth European Conference on Underwater Acoustics ECUA 2000, Lyon, France, [10] J. Marendic-Miljkovic, M. Tasic, S. Rajsic, and Z. Vukmirovic, Precipitation Onset Detection with a Rain Sensor of Improved Sensitivity, Atmospheric,Environment, Vol.34, 2000, pp References: [1] J.A. Nystuen, J.R. Proni, P.G. Black, and J.C. Wilkerson, A comparison of automatic rain gauges, Journal of Atmospheric Oceanic Technology, Vol.13, 1996, pp [2] E. Habib, W.F. Krajewski, and A. Kruger, Sampling Errors of Tipping-Bucket Rain Gauge Measurements, ASCE Journal of Hydrologic Engineering, March/April 2001, pp [3] A.R. Rahimi, G.J.G. Upton and A.R. Holt, Dual-frequency links a complement to gauges and radar for the Hydrology, Vol.288, 2004, pp [4] G.W. Lee and I. Zawadzki, Errors in Rain Measurement by Radar due to the Variability of Drop Size Distributions, Sixth International Symposium on Hydrological Applications of Weather Radar, Melbourne, Australia, February [5] A. Kruger, and W.F. Krajewski, Twodimensional video distrometer: A description, Journalof Atmospheric and