OF DAMS Dr. J. M. Reger School of Surveying and Spatial Information Systems University of New South Wales UNSW SYDNEY NSW 2052 J.Rueger@unsw.edu.au ABSTRACT After a brief review of the origin and early days of the technique, the present role of geodetic deformation measurements is discussed. The design of geodetic measurement schemes is then considered, followed by a review of geodetic measurement, analysis and reporting techniques. An overview of the important discussions, that need to take place between engineers and surveyors in the design phase, follows. This covers the definition of the engineering needs and the resolution of surveying issues. 1 INTRODUCTION To get a better understanding of the term geodetic deformation measurements, it is useful to have a brief look at the origin and history of this dam monitoring method. In Switzerland, the construction of (mainly concrete) water storage dams (for electricity generation) started in earnest in the 1920s. Since the safety of dams is very important for the people living downstream, it is understandable that the dam engineers wanted to know more about the behaviour of dams than the earlier monitoring methods (levelling, clinometers (tiltmeters) and optical alignment) could provide. The structural deformation caused by changes in reservoir (water) levels and (air) temperature give an excellent insight into the quality and, in consequence, the safety of a dam. Beginning in 1921, Swiss National Mapping was contracted by a number of dam owners to determine the deformation at a number of points distributed over the dams (Lang 1929). The increased interest in dam deformation measurements after the failure of the St. Francis Dam (12 March 1928) in the USA lead to the publication by Lang on the geodetic method developed in Switzerland. The geodetic method was proposed by H. Zlly, then chief of geodesy at Swiss National Mapping. Originally, two to three reference points (survey pillars, downstream) were used to intersect the object points (survey marks placed in the downstream face of the concrete dams). The stability of the reference points (observation pillars) themselves was checked by resection from close (relocation points) and distant targets. Horizontal directions were measured with (precise vernier) theodolites. Only one distance was taped. The deformations were obtained by semigraphic means, using the differences of values measured in two consecutive epochs. The settlement of the structure was either monitored by levelling runs across the crest and along the base of the dam or, less often, by zenith angles from the reference points (pillars) to the object points (targets on the dam). Figure 1 (after Fig. 18 in Lang 1929) shows a typical example of one of the early measuring schemes for the Schrh-Dam in the Wgital Valley, about 40 km NE of Zrich (Switzerland). This concrete gravity dam was built in 1924, is 112 m high, 156 m long and at 900 m above sea level. 19 object points (targets) were installed in four rows (horizontal profiles) and six columns (vertical profiles). The measurements were taken from the three reference points (observation pillars). (Pillar movements of 0.7 mm were noted in these early measurements.) The points A to C, E to H in Figure 1 were all used to check the stability of the pillars (by graphical resection). The marks A, B, F, G, H are relocation marks at close range. The more distant marks C, E, J (as well as the rays to the other pillars) were used to orientate the arcs of directions on the three pillars (using weighting according to distance). Further information on the dam may be found on the world wide web at www.swissdams.ch. Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 2 Figure 2 (after Fig. 37 in Lang 1929) shows a three-dimensional representation of the lines of horizontal and radial deformations in six vertical and four horizontal profiles of the Schrh-Dam in the Wgital Valley (Switzerland) between the zero measurement in May 1925 and second filling and the second emptying of the dam. The axonometric diagram assumes a vertical plane for the initial measurements in May 1925 (reservoir at 860 m above sea level) and zero deformations at the abutments. The 14 mm maximum deformation in October 1928 before the emptying of the dam was the same as that of the initial filling (in October 1926). After lowering the reservoir in March 1929 by 40 m, the dam moved a maximum of 2 mm upstream. This means that the first filling caused an irreversible deformation of a maximum of 12 mm. Subsequently, the dam showed an elastic behaviour (2 mm for a 40 m change in water level). Such diagrams can only visualise the structural movements between two (or, as here, three) epochs of measurements. A number of features of the early geodetic measuring schemes in Switzerland and elsewhere have withstood the test times: The Freiberger ball (16.53 mm diameter, according to Lang (1929)) centring of instruments and targets and the corresponding brass centring bolts (with protective cover) are still used on some dams. (Since about 1973, the KERN pillar centring plates with a diameter of 158 mm were used on many new dams.) The basic layout of monitoring targets (in a grid pattern on the downstream face of concrete dams) has found wide acceptance, for example with six vertical profiles and four horizontal ones. Marks near the abutments are important if abnormal behaviour is to be detected. The measuring precision of directions (in two faces) achieved in the early surveys was excellent with better than 1" in most cases and has not changed since then. Figure 1: Typical example of an early geodetic deformation measuring scheme: Schrh Gravity Concrete Dam in the Wgital Valley, about 40 km NE of Zrich (Switzerland). Two of the 19 object points are labelled (1a, 6a). Pillars 1 to 3 are the reference points, the marks A, B, F, G, H are relocation marks and the marks C, E, J are (distant) targets used for orientation purposes (after Fig. 18 in Lang 1929) Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 3 The (up to four) relocation targets near pillars for the check of the stability of the pillars are not always installed on newer schemes even though they are useful to monitor the pillar behaviour against the surrounding ground. The use of concrete pillars (0.5x0.5x1.15 m at the time) was very successful. Today one prefers round double skin concrete pillars which minimise mechanical damage of the inner core. The inner pillar is shaded and, thus, moves less with the sun. Lang (1929) noted that 20 to 30 rays can be measured within the same arc of horizontal directions because of the good stability provided by concrete pillars. Each reference point should have clear sights to at least four orientation marks. He also suggested that, in an arc of horizontal directions, 50 % of rays to reference points should be measured first, followed by all object points and, then, the rest of reference points. Two arcs of directions should always be measured (for full geodetic measurements). The early Swiss experience was that measuring 4 arcs does not improve the results. Some features of the early schemes however changed over the years: The (two to three) observation pillars were originally placed at about mid height of the dam. Today, more observation pillars are installed up- and downstream of the dam. This requires that some pillars are higher than the dam. The distant targets (used originally for the orientation of arcs on the pillars) have been dropped in newer schemes. The targets on the dam featured white vertical lines (straight or conical) on a black base. Later, this was changed to concentric circles (see (4) in Fig. 20). The targets that were inserted into the centring bolts of pillars originally carried vertical lines. Later designs featured concentric target patterns (see (3) in Fig. 20) or, even better, brightly coloured spherical balls (Fig. 27). Figure 2: Three-dimensional (axonometric) representation of the lines of horizontal and radial deformations in six vertical and four horizontal profiles of the Schrh-Dam in the Wgital Valley about 40 km NE of Zrich (Switzerland) between the zero measurement (May 1925) and the measurements immediately before (October 1928) and after (March 1929) the first draw down of the dam (after Fig. 37 in Lang 1929) For t he earl y geodet i c deformat i on measurements in the 1920s, National Mapping of Switzerland tested a number of ways on how to present the results of geodetic measurements on dams. A number of them are still being used today. Particularly useful and visual are the representations of deformation lines. Figure 2 gives an example (after Lang 1929) of an a xonome t r i c 3- D r e pr e s e nt a t i on. Unfortunately, such diagrams can only visualise the structural movements between two (or, possibly, three) epochs of measurements. The lateral and tangential movements have to be plotted separately. Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 4 The early publication by Lang (1929) contains many other useful suggestions for the design of monitoring networks and the execution of surveys. The author warns of lateral refraction (affecting lines of sight close to ground or structures) and the huge refraction problems experienced when measuring along the crest (caused by the bending of the up- or downstream winds over the crest and the strong temperature gradients associated with it). This is a reason why optical alignment on the crest is not considered suitable for highest precision. If optical alignment is to be used, Lang (1929) suggests to place (on the abutments) the instrument and reference target higher than the crest. 2 ROLE OF GEODETIC MEASUREMENTS IN THE MONITORING OF DAMS According to the Swiss Commissioner for Dam Safety (Biedermann 1996, 1997), the geodetic dam monitoring techniques have become less attractive over the years, firstly, because they require skilled personnel and, thus are expensive and time consuming and, secondly, because direct mechanical measuring devices such as pendulums and wire alignment systems (that can be operated by less skilled dam based staff) have become available. The geodetic techniques are still very important since they produce absolute data and connect the localised dam based measuring devices to the dam's foundations and the area surrounding the dam (and, possibly, slopes along the reservoir). These days, the geodetic measurement scheme provides the foundation for the measurements in the case of an abnormal behaviour of the dam and is measured infrequently. To assure the safety of a dam, three elements are necessary (Biedermann 1996): safe state-of- the-art design of the construction, monitoring of the structure and an emergency concept. It could be argued that, in line with the first requirement, the dam should be measured during the first filling and emptying to test if the actual deformations agree with the expected deformations. The monitoring must be able to detect damages, constructive deficiencies and threats to safety so that an abnormal event can be detected and responded to (Biedermann 1996). The Swiss Commissioner for Dam Safety (Biedermann 1996) suggests that the monitoring of dams be carried out as follows: Visual Inspection ( once a week). Since not all threats to the safety of a dam and reservoir can be captured by measurements, a visual inspection by persons familiar with the structure is essential. The visual inspection must cover the dam, its surroundings and, if necessary, the slopes along the reservoir Measurements ( once per month) of key indicators of the behaviour of the dam, its underground and its surroundings (including slopes along the reservoir, if necessary). This is acceptable since abnormal behaviour of structures and terrain usually develop slowly. The knowledge of the radial displacement at one or more points along the crest is sufficient for this purpose. On concrete dams, these frequent measurements are typically carried out by on-site personnel with direct measuring devices, such as pendulums, wire alignment systems, clinometers, extensometers, etc. Such measurements are very precise, simple to make and cost effective. Automatisation of this type of measurements and on-line recording and inspection is easily possible, if so desired or required. Periodic Safety Examination (large dams: every 5 years, small dams: when required). Biedermann (1996) suggests that a reduced measuring program of the installed geodetic network be measured during these five yearly safety checks of the dams. Ideally that should be at full reservoir and at the same time of the year since seasonal effects are often more pronounced that changes caused by water level in the reservoir. It is further suggested to measure the complete geodetic network every 15 years when the reservoir is empty. It follows that the Swiss authorities see the role of geodetic measurements mainly as a measuring base from which more specialised measuring schemes can be developed should an abnormal behaviour of the dam be detected by other means (visual inspection, direct frequent measurements of key indicators). A three- dimensional network is required to determine radial and tangential movements, settlements and rotations of the dam as well as deformation of the surrounding terrain. In the case of an emergency, measurements are taken more frequently and the monitoring scheme might be extended. Important is that a fundamental geodetic network, covering all relevant parts of the reservoir (dam, abutments, foundation, surroundings, and, if required, slopes along the Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 5 reservoir, unstable rock masses and (in other parts of the world) unstable glaciers) has been installed, is being maintained and measured infrequently (e.g. every 15 years) to provide a reference (zero-epoch). If the dam experiences snow fall in winter, the design of the full geodetic measuring scheme should ensure that the measuring points of the network are accessible in winter as far as possible. Naturally, dams with galleries at different levels (ideally extending into the abutments on both sides and equipped with wire alignment or survey precision traverses and levelling networks) and vertical shafts (with pendulums, ideally extended into the foundation with inverted pendulums) are best suited for measurements under snow cover. Figure 3: Principle of a geodetic deformation measurement network for concrete dams. Inside the dam the network comprises three pendulums and one inverted pendulum as well as three surveying traverses in galleries at three levels. The interior network is in a vertical plane. The exterior survey network is essentially in a horizontal plane at crest level with a number of targets on the crest (connecting the pendulums to the exterior net) and a number of reference points up- and down-stream. The exterior network is measured by directions, zenith angles, distances and/or GPS. Some levelling lines at the crest level are also shown, as is an additional survey network at a lower level. (After Biedermann 1985) Figure 3 (after Biedermann 1985) shows the basic layout of a geodetic deformation measurement network for straight or curved concrete dams. It features measurements in two planes. In a vertical plane (may be curved), the network comprises three pendulums and one inverted pendulum as well as three surveying precision traverses in galleries at three levels. If the dam is straight, a wire alignment system can replace the surveying traverses in the galleries. Ideally, the galleries should extend into the abutments to provide an additional absolute reference. As shown, an absolute reference is provided in the interior network by an inverted pendulum reaching into the underground. The basic exterior survey network in Fig. 3 is in a horizontal plane at crest level with a number of pillars on the crest (connecting the pendulums to the exterior net) and a number of reference pillars up- and down-stream. The exterior network is measured with electronic tacheometers (by directions, zenith angles and distances) and/or GPS. Some levelling lines at the crest level are also shown, as is an additional survey network at a lower level. Essential is that the interior and the exterior networks are interconnected as shown by the large circles in Fig. 3. The interior network (vertical plane) and the exterior network (horizontal plane) are connected at three points on the dam's crest. The levelling lines are connected to the reference points of the other survey measurements. As mentioned before, the network might have to be extended if the reference points shown are still in the influence zone of the dam and/or if the ground near the Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 6 dam and/or the slopes along the reservoir need to be monitored too. With the Global Positioning System (GPS) the network can easily be extended to cover a wider area without requiring inter-station visibility. It follows from the earlier considerations in this section that at least one pendulum and the inverted pendulum be measured every month, the interior network (pendulums, traverses in horizontal galleries) every five years and the full network every fifteen years or so. Figure 4: Wire alignment system installed in the parapet of a gravity dam. Pulley and weight are seen on the right. The measuring point is on the left. The measuring microscope is attached to the big bolt seen above the wire. After Biedermann (1997, Fig. 2.10). Since most dams in Switzerland are concrete dams, the discussion above (and Fig. 3) are tailored for this type of dam. Embankment dams (deck type or fill type) and small concrete dams do not have internal galleries and vertical shafts. This means that the interior network shown in the vertical plane in Fig. 3 needs to be configured as an equivalent grid of object points on the downstream face of the dam. Rather than run traverses along berms, the object marks on the downstream face are connected to the reference points by standard surveying measurements (horizontal directions, slope distances and zenith angles) or, possibly, by satellite measuring techniques (e.g. GPS). That poses no problems for the (reduced) geodetic measurements every five years and the full measurements every 15 years. The remaining problem with embankment dams (or, generally, dams without galleries or shafts) is the simple measurement of critical parameters at monthly intervals. Biedermann (1997) suggests a wire alignment system along the crest. This is possible for straight concrete faced rockfill dams with crest/parapet wall, for example (if the settlements are not too large!). Since wire alignment systems are rarely seen in publications, the system installed within the parapet of the Rempen gravity dam in Switzerland is shown in Fig. 4. (More information on the dam can be found at www.swissdams.ch.) The wire is tensioned by a heavy weight through a pulley system. When not in use, the measuring points and the weight are protected by panels. Figure 5a: Simple measurement of angles from two reference points (Pillars 1, 3) to object points (42, 44, 46) on the crest of the Chapfensee/Parmort (gravity) dam in Switzerland. The upstream-downstream movements of the object points 42, 44, 46 are computed from the changes of the measured angles (and, as a check, from the measured angles ). After Egger & Walser (2005, Fig. 2.06-1). Alternatively, the optical alignment of one or more crest point(s) is feasible or the measurement of some distances from a pillar downstream to some object points on the downstream face of the dam or the intersection of one or few crest points by simple angle measur ement or cl i nomet er / t i l t met er observations on the upstream face. An arrangement of simple angle measurements is shown in Fig. 5a for the Chapfensee (Parmort) Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 7 Dam in Switzerland. The reservoir is located 4 km west of Sargans and features two dams. Since only the northern one (120 m long and 20 m high) is listed at www.swissdams.ch, it is assumed that Fig. 5a refers to the northern dam. Further information on the dam may be found on the world wide web at www.swissdams.ch. On slim and strongly curved arch dams (with no pendulums) simple vertical angles (or zenith angles) can be measured to bull's eye (circular) targets on the dam that face downwards. See Figure 5b. Because of the quasi-vertical line of sight, a change in the zenith angle is a measure of an upstream/downstream displacement of the dam (after multiplication with the height difference (assumed known) from the pillar to the mark). One-second theodolites with diagonal eyepiece are used for the purpose. According to Egger & Walser (2005), qualified personnel is required for this type of measurement that is simple in principle but not in practice. Figure 5b: Simple measurement of vertical angles from an observation pillar at the base of the dam to object points (11 to 16) on the downstream face of slim double curvature arch dams. The upstream-downstream movements of the object points 11 to 16 are computed from the changes of the measured vertical angles (and the known distance from the pillar). After Egger & Walser (2005, Fig. 2.06- 2). Some of these alternatives require some training in surveying of dam staff. The problems associated with optical alignments were already known in the 1920s. Biedermann (1997) cites some tests by the Swiss Federal Institute of Technology (ETH) in Zrich with optical alignment that exhibited errors of ten seconds of arc (10", 14 mm/300m) on a Swiss dam. Given the reduced precision (compared to pendulum observations in concrete dams) of the monthly simple observations on dams without galleries or shafts, Biedermann (1997) recommends geodetic measurements (of the dam face) at least once a year (rather than every 5 years). 3 DESIGN OF GEODETIC DEFORMATION SCHEMES The design and measurement of geodetic deformation schemes have been described many times. The reader may refer to the following publications for a more detailed discussion, for example: Swiss National Committee on Large Dams (1997, 1993, 1985), Egger & Keller (1976), Keller (1978), Kern (1971), Untersee (1951, 1975). Based on his extensive experience with deformation measurements, Egger (1997, 1993) gives some sound advice on terrestrial measurements of deformations. The monitoring scheme must be designed for a long service life of more than 50 years. The geodetic network must be as complete as possible and allow for later extension (in case of abnormal behaviour or new construction work). Close collaboration between engineers and surveyors is essential and should begin as early as possible in the preliminary design phase since the location of pendulums, galleries, wire alignment systems, etc. must allow connections to the geodetic measurement scheme. The measuring scheme should be flexible enough to allow the adoption of new measuring techniques later. The reference points must be close to the dam, both upstream and downstream, outside the influence zone of the structure, must have inter-visibility and, ideally, must be accessible all year round. Experience shows that one to two points in each scheme become unreliable with time or get destroyed (e.g. by construction activity, rock fall, avalanche). Egger (1997, 1993) considers four reference points an absolute minimum. The reference points feature normally deeply anchored double-walled concrete pillars with a forced centring device that is protected from the elements and vandals. Brackets for the attachments of umbrellas are helpful. In exposed locations, the pillar must be inside a concrete shelter that Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 8 protects from possible rock falls, avalanches, land slides and the like. Four relocation marks at close range are useful to check for pillar movements (by resection, against the surrounding terrain). If satellite measurements (e.g. using GPS) are planned, then there should be no obstructions above an elevation angle of 15-20 degrees and no reflecting surfaces nearby. Unfortunately, the water in the reservoir is a perfect microwave reflector. Figure 6: Connection of the horizontal position of the pendulum to the exterior geodetic network. The pendulum is 'measured in' from 'C'. The point 'B' is established vertically above 'C' by optical plumbing. After that, 'B' can be measured in from the pillar 'A'. The object points are placed so that they provide information on the behaviour of the structure (and its surroundings, if required). The object points can be pillars on the crest of the dam (with forced centring system), brackets (with forced centring system), bolts (with forced centring system) in the ground. Object points must be able to be surveyed with conventional surveying techniques. Since these include distance measurements, EDM reflectors must be able to be fitted (temporarily) to the object points and the object points should be accessible all year round. If object points must be established on the downstream face of concrete dams (because the dam lacks pendulums and galleries), they are usually circular targets on brass bolts and observed by the traditional intersection method. Egger (1997, 1993) does not favour inaccessible reflectors in dam walls since they become dirty and the reflecting surfaces deteriorate with time, particularly in a humid climate. Usually, the deformation measurement scheme also includes some object points ('bench marks') that are only determined in height (by levelling). Figure 7: Observation platform on the downstream face of the Gigerwald arch dam in Switzerland. (Photo by Kern & Co Ltd Aarau, Page 62, Swiss National Committee on Large Dams 1985) Object points located inside the galleries (of concrete dams) provide the most accurate vertical and horizontal movements. Unless the traverses in the galleries extend into the abutments, they must be connected to the exterior network to provide absolute deformation information. The preferred method is to connect the gallery traverses to the pendulums and the top (anchor points) of the pendulums to survey pillars on the crest. Figure 6 shows an early version of this approach (Egger & Keller 1976). The connection of the pendulums to the outer network can be simplified if the anchor points Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 9 of the pendulums are at (or near) the crest level. Egger (1997, 1993) considers the alternative method of connecting the galleries to the outer network through openings (see Fig. 7) in the downstream face less desirable. As in tunnel work, the measurements from the outside to the inside of the dam usually suffer from large horizontal refraction effects. Figure 8: CERN forced centring system with central 30 mm bore. Fits (directly) reflectors and targets (shown) fitted to Taylor-Hobson spheres as well as the DISTINVAR invar measuring device. Standard surveying equipment can be attached (and locked!) in the centring device with adaptor plates fitted with the customary 5/8 inch Whitworth thread on top (see Fig. 25). The Taylor-Hobson target with concentric circles shown in the figure is not part of the device proper. For the terrestrial measurements, theodolites, electronic distance meters and levelling instruments are typically used, the former two often combined in the form of electronic tacheometers. Typical theodolite precisions are 0.7" for directions, 1" for zenith angles and (0.1 mm + 0.7 ppm) for the Kern Mekometer ME5000 precision distance meter. Egger (1997, 1993) quotes levelling with 0.1 mm/station. For the internal traversing networks, invar wires are/were often used, recently in the form of the DISTINVAR originally developed by CERN (European Organization for Nuclear Research). According to Egger (1997, 1993) the DISTINVAR achieves 0.02 mm in routine measurements. The DISTINVAR requires the CERN centring system (with a 30 mm diameter centring cylinder) which is commercially available in its or i gi nal or si mpl i f i ed f or ms ( see www.geodesie.com). The original CERN centring system (in three parts to allow to make the central cylinder vertical) is made out of aluminium alloys and shown in Figure 8 (after CERN 1974). Measurements with motorised electronic tacheometers with automatic target recognition allow to measure under computer control to reference and object points that are equipped with EDM reflectors. If instruments are installed permanently (in weather proof and vandal proof observation buildings), continuous measurements are possible. Remote control is available if the communication links are installed. Continuous monitoring systems are very costly since they require substantial installations and maintenance and permanently tie down an expensive instrument. Egger (1993, 1997) notes that, ' in certain circumstances, this is entirely justifiable'. During the measurement of the network (called one epoch of measurements), a predetermined observation plan should be followed and the observations should be carried out according to the rules of good professional practice as it applies to precision surveys. The observation procedures have to be designed in such a way that obvious errors are detected on site and that they can be remedied on the spot. That might require some preprocessing of data in the evenings. (Should errors be found later in the office during processing, it is too late to take additional measurements.) It is customary, to use always the same equipment (in the same orientation) on the same reference and object points since this eliminates some systematic errors in the deformations that are being determined. If electronic data recording is being used, daily back-up of the data to other recording media should be carried out. Irrespective of the type of equipment used, some aspects are always important (Egger 1997, 1993): height of instruments, targets, reflectors, antennas above mounting plate atmospheric parameters (temperature, pressure, humidity), weather, reservoir level Misalignments, scale errors and eccentricities of instruments, targets, reflectors, staffs, antennas Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 10 Figure 9: Geodetic deformation network of the Mattmark embankment dam in Switzerland (height: 120 m, length: 770 m). The points marked 210 (and 600, 650) are concrete pillars from which the measurements to the object points (marked by circles) are taken. The reference points 2,4,7 and 9 are presumed to be outside the stress zone. Some targets (11 -21) mounted on rock faces were originally used for the orientation of arcs. The rows of points (numbered 651 to 958) were used to investigate some abnormal behaviour and are not part of the continuing observation scheme. After Kgi (1978, Fig. 1). avoidance or elimination of other systematic errors affecting the measurements. Ideally, the reservoir level should be kept constant during the measurements of one epoch of data. Even so, daily readings of all direct measurement devices (e.g. pendulums) and of water level should occur. Simultaneously with the geodetic measurements, all other types of monitoring devices should also be read to allow the correlation of the results. Having discussed some important principles of geodetic deformation measurements, it is appropriate to look at some practical examples. Figure 9 depicts the inner (on the dam) and the outer geodetic networks of the Mattmark embankment (fill) dam high in the Swiss Alps. The dam is in the Canton Valais, 17 km east of Zermatt and 23 km ENE of the famous Matterhorn mountain. It has a height of 120 m, a length of 770 m and a volume of 10.4 million cubic metres. The maximum reservoir level is at 2197 m above sea level. Observations (directions, zenith angles and, these days, distances) are taken from the points 210 (double-walled concrete pillars). (The pillars 600 and 650 were only used temporarily.) Four of the nine observation stations (2, 4, 7, 9) are reference points on the sides of the valley and likely stable. Pillar 10 is on the dam to improve the geometry of the network. It is subject to deformations like the other points on the dam. Object points are on both faces of the dam and on the crest. Those on the upstream face are often under water. According to Gilg (1985), there is a total of 98 object points on the dam. Not all of them are shown in Fig. 9. An additional 11 object points are in the drainage gallery in the base of the dam and not shown. There are also 35 (levelling) bench marks in the monitoring Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 11 scheme. (Only some are shown in Fig. 9.) More information on the Mattmark Dam can be found in Kgi (1978), Gilg et al. (1982), Gilg (1985) and at www.swissdams.ch. Biedermann (1997, Fig. 2.8) gives an updated plan of the Mattmark network: There is now a second observation pillar on the dam, the targets on the rock faces are no longer shown nor are the rows of dense points for the temporary special observation scheme. The modern internal network of a gravity concrete dam in Switzerland is shown in Figure 10. The Panix (Pigniu) Dam was commissioned in 1989, is 53 m high and 270 m long. The reservoir is 32 km east of Chur in the Canton of Grisons and at an elevation of 1450 m above sea level. This dam monitoring scheme follows the recent trend and has no object points on the dam's faces. There are only three survey pillars on the crest to connect the interior net (of Figure 10: Interior geodetic deformation network of the Panix (Pigniu) gravity dam in Switzerland. 1: survey pillars on crest to connect the interior to the exterior network. 2: wire alignment, 350 m long. 3: pendulums. 4: inverted pendulum. 5: alignment reading points (one in each block and in the abutments). 6: pendulum reading points. 7: floating intermediate deformation network of the Panix gravity dam in Switzerland. After Biedermann (1997, Fig. 2.7). Figure 11: Interior measuring scheme of the Gigerwald Dam in Switzerland showing the four normal and four inverted pendulums as well as the three galleries with the traversing surveys. The lower diagram in Fig. 11 shows the uppermost gallery and its traverse stations at 32 m intervals (length of invar wires). Three pillars ('A') on the crest connect the uppermost traverse to the exterior network as do three pillars ('B') on little balconies in the downstream face. After Biedermann (1997, Fig. 2.6). Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 12 pendulums and wire alignment) to the exterior one. The up/down stream deformations are measured by the pendulums and the wire alignment, the sideway movements by the pendulums. It has to be assumed that the vertical movements are monitored by levelling in the gallery and/or on the crest. For completeness, the geodetic monitoring network of the Swiss Gigerwald double curvature arc dam is shown in Figures 11 and 12 even though the layout has been shown many times before (Keller 1978, Egger & Keller 1976, Egger 1993 1997, Biedermann 1997 1993). The Gigerwald Dam was commissioned in 1976, is 147 m high and 430 m long. It is about 15 km SW of Chur, Switzerland, and at 1335 m above sea level. Figure 11 depicts the internal measuring scheme of four normal and four inverted pendulums as well as three galleries with traversing surveys (originally: angles by theodolite and distances with invar wires). The lower diagram in Fig. 11 shows the uppermost gallery and its traverse stations at 32 m intervals (length of invar wires). The traverse stations in the galleries consist of steel brackets attached to the walls carrying a simplified version of the CERN centring system (Fig. 8). The DISTINVAR measuring device fits the CERN centring directly. The rest of the survey equipment uses appropriate adaptors. Keller (1978) shows most of the equipment used on the dam in the late 1970s. The internal network is connected to the external network (shown in Fig. 12) by three pillars (see Fig. 13) on the crest (connecting to three points of the uppermost traverse using the procedure Figure 12: 'Complete' exterior measuring scheme of the Gigerwald Dam in Switzerland (Scale: Distance 2-3 = ~395 m). 1: Observation stations. 2: reference points (presumed stable). 3: Pillars on crest (603, 611, 621). 4: Pillars on platforms in the downstream face (207, 211, 215). 6: Rock monitoring points (1F to 5F). 7: Close range relocation marks. 8: point number and elevation. 9: reciprocal and one-way measurements of directions, zenith angles and distances. Tamina is the name of the river in the valley. After Biedermann (1997, Fig. 2.6). Since 1997, the network was extended by two stations about 550 m upstream. (Refer to Fig. 4 in Egger & Walser 2005) Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 13 outlined in Fig. 6) and by three pillars on platforms in the downstream face (Fig. 7 shows one of these connection pillars denoted by B in Fig. 11.) Figure 13: KERN Mekometer ME5000 precision distance meter on Pillar 603 of the Gigerwald Dam (Aeschlimann 1988, Kern 1988). The top of the pillar with the KERN pillar centring plate is normally protected by a steel cover fitting onto the steel ring visible on the outer pipe. Figure 12 shows the 'complete' exterior measuring scheme of the Gigerwald Dam in Switzerland. (Fig. 4 in the 2005 publication of the Swiss National Committee on Dams shows that, since 1997, the Gigerwald Dam network was extended by two reference stations, about 550 m upstream from the dam.) The dark lines indicate the primary geodetic network between the four reference points (1, 2, 3, 4) and the three pillars on the crest. The additional rays connect the three pillars in the downstream face (207, 211, 215) and the rock monitoring points (1F to 5F). Measurements are only taken to the rock monitoring points but not from them. According to the photos in Keller (1978) all marks of the outer network on the Gigerwald arch dam are equipped with the Kern pillar centring plates of 158 mm diameter (see Figs. 13 and 24). The rock monitoring points feature centring plates that are attached to heavy duty steel brackets. Keller (1978) gives also the diagram of a 'reduced' exterior measuring scheme of the Gigerwald Dam, with a selection of measurements from Pillars 211 and 1. This example of a complete and a reduced measurement scheme is useful when selecting reduced schemes on other dams. 4 RESULTS OF GEODETIC MEASUREMENTS: ANALYSIS AND REPORTING In the early days of geodetic dam deformation measurements (in the 1920s), pocket calculators, and computers in general, were not available; least squares adjustments had to be carried out by hand using logarithm tables. No wonder that the analysis of dam deformation measurements was carried out with semi-graphic means that was much faster and sufficiently accurate. Today, most survey data are recorded digitally. The subsequent processing of the data is done on personal computers. Whereas the early geodetic measurements on dams involved only horizontal direction, levelling and, possibly, zenith angle observations, today's data are likely to include electronic (and/or invar wire) distance measurements and, increasingly, GPS measurements. EDM measurements require some preprocessing to account for ambient temperature, pressure and humidity. Also, the earlier derivation of the deformations from the change (between measuring epochs) of the raw observations was replaced by the execution of least squares network adjustments of two epochs and the derivation of the deformations from the change in coordinates between epochs. This evolution from simple differences of measurements to changes in coordinates (determined by least squares network adjustments) has some serious consequences, that should not be overlooked. As with all geodetic network adjustments, all observations must be reduced to a common datum or coordinate system before the network can be adjusted. In the case of dam deformation networks, a local x-y coordinate is usually adopted, typically with a false origin to avoid similar x and y coordinates in the project area. Essential is that the reference elevation of the horizontal coordinate system is clearly defined and that all distance measurements are reduced to this reference elevation (being a spherical surface). The reduction of measured slope distances to the reference elevation is not trivial; it can be based on measured zenith angles or the elevation of the terminals of the lines (Reger 1996). Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 14 Since surveying instruments are levelled at each set-up (so that the direction of vertical axis coincides with that of the plumbline), any deviation of the local vertical (plumbline direction) from the normal to the datum surface must be considered. In alpine areas these 'deviations of the vertical' can be significant and, if ignored, can affect the coordinates and precisions obtained by network adjustments. It might be necessary to model (in the least squares adjustment) the deviations of the vertical or to correct the measurements before the adjustment. In most cases, the horizontal measurements are adjusted (in the local x-y coordinate system) separately from the height (difference) measurements. Typically, levelling data and height differences (computed from zenith angles and slope distances) are used in the latter. So, essentially, the adjustments of the data occur in 2+1 dimensions. If the geodetic deformation measurements include GPS data, then the network adjustment becomes more involved. Schneider & Wiget (1997) suggest combined terrestrial-GPS network adjustments to obtain realistic results. These authors report the use of a true 3D hybrid adjustment in a Cartesian (or geocentric) coordinate system and the (additional) solution of the two components of the deviation of the vertical at each point. To evaluate if the dam has experienced deformations between two successive epochs, the two epochs are adjusted in one common adjustment. This adjustment will have to be repeated a number of times to statistically test the stability of the reference points and to identify the object points that have moved statistically between epochs. All statistical tests should be carried out at 95% confidence level. Software for the rigorous deformation analysis can be obtained from a number of sources. The results of a rigorous deformation analysis are best displayed in 2+1 dimension. Then, the 95% confidence ellipses of the horizontal (2D) deformation vectors can be plotted together with the deformation vectors (using the same origin for ellipse and vector). Any vector that crosses the ellipse is significant. Figure 14 gives an example (Welsch et al. 2000). The vertical movements can be shown in a similar diagram together with the error bars at 95% confidence level. The reporting of the results is described in detail by Egger (1997). This author states that 'the main aim must always be the production of an objective and complete record of the survey' and mentions that results that are difficult to explain are often the first sign of an abnormal behaviour of the dam. Egger (1997) suggests that the horizontal and vertical point movements (presumably between two consecutive epochs) are presented in graphical as well as in tabular form. It should be added that the diagrams and tables should clearly indicate the movements that are statistically significant and those that are not. As shown in Fig. 14, it might be appropriate to show only si gni fi cant vect ors i n t he graphi cal represent at i on of t he epoch-t o-epoch movements. Separate diagrams of the long term movements might also be useful or required. Here, it might be better to show all epoch-t o-epoch changes si nce smal l (insignificant) changes can add up to significant ones with time. Figure 14: Deformation vectors of a dam deformation network, with 95% confidence ellipses of the vectors. Only the significant vectors are shown. (after Welsch et al. 2000, Fig. 9.3-2) Figures 15 to 17 give examples of the graphical representation of the long term deformation of a large embankment (fill) dam in Switzerland. All results shown are based on geodetic measurements. The dates and reservoir levels of the epochs shown in the figures are reproduced in Table 1. As mentioned above, the Mattmark Dam is located in the Swiss Alps, 23 km ENE of the famous Matterhorn mountain. The dam has a height of 120 m and a length of 770 m. The maximum reservoir level is at 2197 m above sea level. The construction was completed in 1967 and the first filling occurred in 1969. The survey network is shown in Fig. 9. Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 15 Epoch Date Water Level (m) 22 6.1971 2134 23 10.1971 2194 24 6.1972 2132 25 9.1972 2190 26 6.1973 2143 27 10.1973 2196 28 6.1974 2135 29 10.1974 2193 30 6.1975 2143 31 10.1975 2197 52 6.1977 2158 53 10.1978 2196 55 6.1980 2140 56 10.1981 2196 58 6.1983 2145 Table 1: Mattmark Embankment (Fill) Dam, Switzerland. Dates and corresponding reservoir levels of the measurements shown in Figs. 15 to 17. (after Gilg 1985) Figure 15 shows the horizontal movements of five points on the crest. Evidently, the crest points move towards the centre of the dam as well as downstream. Also, some plastic movement continues 16 years after completion of the dam. An elastic behaviour of about 2 cm between full and empty reservoir can be observed. Figure 16 (after Gilg 1985, Fig. 4.3.3-5) shows the vertical and downstream displacements of the Mattmark Embankment (Fill) Dam, Switzerland, from June 1971 to June 1983. The data are from geodetic measurements of the object points (306 to 312) in the central profile M. Note that the construction was completed in 1967 and the first filling in 1969. Since the object points on the upstream face can only be measured at low water levels, they exhibit a continued settlement and downstream movement. (Point 309 moves 248 mm down and 92 mm downstream over 12 years.) The marks on the downstream face show an additional reversible horizontal movement of about 20 mm between full and empty reservoir. Figure 17 (after Gilg 1985, Fig. 4.3.3-6) depicts the vertical movements (settlements) of the survey marks in the drainage gallery of the Mattmark embankment (fill) dam, Switzerland, from June 1964 to June 1983. (Construction: 1961 - 1967, first filling: 1965 - 1969.) Gilg Figure 15: Mattmark Embankment (Fill) Dam, Switzerland. Horizontal displacements (from geodetic measurements) of the object points (110 to 510) on the crest from 1971 to 1983. The construction was completed in 1967 and the first filling occurred in 1969. See Fig. 16 for the movements in the central Profile M. (after Gilg 1985, Fig. 4.3.3-4) Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 16 (1985) reports that the maximum settlement reached 1.85 m by January 1967. The first complete filling of the reservoir (by the northern autumn of 1969) caused a further settlement of 0.30 m. The following 13 filling cycles added another settlement of 0.20 m. In 1985, the drainage gallery settled at about 10 mm/year with no end of the settlements in sight. The report on geodetic measurements must satisfy a disparate group of readers, such as the administrators, technical and legal staff of the owner of the dam as well as civil engineers, dam specialists and chief surveyors (Egger 1997). The findings and conclusions should be expressed in such a way that they are meaningful for all targeted readers. Following mainly Egger (1997 1993), the essential components of the report on the geodetic measurements (of an epoch) can be listed as follows: (a) date and times of survey (b) external conditions (reservoir level, air temperature, weather) c) condition of the installed equipment, any deterioration or damage to the reference or object points, any new modification or extensions (d) instruments used, names of the observers Figure 17: Mattmark Embankment (Fill) Dam, Switzerland. Settlement of the object points in the drainage gallery in the base of the dam from 1964 to 1982. The construction was completed in 1967 and the first filling occurred in 1969. 1: alluvium, 2: bedrock (after Gilg 1985, Fig. 4.3.3-6) Figure 16: Mattmark Embankment (Fill) Dam, Switzerland. Vertical and downstream displacements (from geodetic measurements) of the object points (306 to 312) in the central profile M from 1971 to 1983. (See Fig. 15 for the location of the profile.) The construction was completed in 1967 and the first filling occurred in 1969. (after Gilg 1985, Fig. 4.3.3-5) Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 17 (e) number of measurements taken and progress of observations with time (observation program) (f) operating methods (g) results of measurements and their accuracy (incl. 95% confidence intervals) (h) geodetic analysis (interpretation) of the results (incl. significant deformations since the last epoch at 95% confidence level) (i) appendices Egger (1997 1993) adds that items (d), (e) and (f) are mainly for the benefit of the surveyors and assist the planning and execution of the measurements of the following epoch. 5 GEODETIC MEASUREMENT TECHNIQUES: PAST, PRESENT AND FUTURE The hardware and methodology of geodetic deformation measurements have changed considerably since the first intersection measurements of the 1920s (see Section 1). Because of the long service life of dams, the type of survey equipment, the observation procedures and the analysis techniques are likely to change over the lifetime of the structure. Accordingly, the geodetic network has to be designed as flexible as possible. For an up-to-date listing of all geodetic measuring tools currently considered for dam deformation measurements, the readers are referred to a recent and detailed publication of the Swiss Committee on Dams (2005). The summary given here is derived from this publication as well as from Welsch (2000) and others. Theodolites The theodolites used in geodetic deformation measurements have changed over the years but the precision of the directions and zenith angles measured with them very little. The Wild T3 precision theodolite was already foreshadowed in Lang (1929) and was a marked improvement (for the measuring comfort) on the vernier theodolites used earlier. (According to Dedual (2005), Series 1 was in production from 1927 to 1934 and Series 2 ('NT3') from 1935 to 1957.) This theodolite allowed the removal of the bottom plate of the tribrach and the attachment of a centring ball (Freiberger ball centring, diameter ~16.5 mm) underneath for direct centring in the pillar bolts that are referred to as 'Wild Bolts' by ANCOLD (1983, Fig. 4.1.2 e). Wild started the manufacture of the precision electronic theodolite Wild T2000 in 1983. The Wild T3000 was manufactured between 1989 and 1997 and the Leica T2002 from 1988 to 1996 (Dedual 2005). The standard versions of these instruments do not feature the Freiberger ball centring for the ('Wild') bolts used since the 1920s. These instruments are not more accurate than the earlier Wild T3 or Kern DKM3, but they feature dual-axes level sensors (important for steep sights) and digital recording. Some of these instruments are available in motorised versions with or without automatic pointing (to EDM prisms only). Leica is advertising the (motorised) electronic theodolite TM5100 as the replacement of the earlier T3000. These two instruments feature the same telescopes. The angle encoders of the TM5100 are accurate to 0.5" and the servo motors have a positioning accuracy of 0.7" (Leica 1997). It is not known if the TM5100, as the T3000 it replaces, can be accurately attached to the Kern pillar centring plates using the system shown in Fig. 24. Sadly, the motorised video theodolites (Kern E2SE and Wild TM3000V), that were available in the 1990s (Reger 2003) and, theoretically, permitted automatic measurements to circular targets on downstream dam faces, are no longer available. Since at least one electronic tacheometer is presently (2006) sold with a built-in CCD camera, there may be hope that instruments capable to measure to non- reflective targets such as the target bolts shown in ANCOLD (1983 Fig. 4.1.2 d), do emerge (again). Egger & Walser (2005) quote a typical precision of directions measured with precision theodolites as 0.7" and a typical centring precision of smaller than 0.1 mm. Electronic Distance Meters The first small electro-optical distance meters (precision 10 mm) became available in 1969 (Reger 1996). The first precision distance meter (Kern Mekometer ME3000, see Fig. 7) entered the market in 1973. The newer Kern Mekometer ME5000 (see Fig. 13) was released in 1986. Both types of Mekometers were used widely for the deformation measurements of (concrete) dams. Sadly, none of the dedicated precision distance meters is still in production. Stand-alone electronic distance meters measure slope distances only. The complementary angle measurements must be obtained with a theodolite. No wonder that today's surveyors prefer to use instruments that can measure in Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 18 three dimensions (horizontal directions, zenith angles and slope distances). These combined instruments are called electronic tacheometers (or, sometimes, 'total stations'). Egger & Walser (2005) quote a typical precision of slope distances measured with precision distance meters as (0.1-0.2 mm + 1 ppm) and centring precisions of less than 0.1 mm. The ppm term is likely to include the uncertainty of the temperature, pressure and humidity measurements that are required to correct the measured distances. In the geodetic deformation network of the Mangrove Creek Dam (NSW), Reger (1995) demonstrated that the ppm term can be significantly reduced if the local scale parameter method is used (and the temperature, pressure and humidity are ignored). Figure 18: Red acrylic reflectors (road delineator, 80 mm diameter) on UNSW carrier with rotating vertical axis fitting the 5/8 inch thread of the object points on the downstream face of the Mangrove Creek Dam in NSW. (Reger 1994, Reger & Sippel 1994) To achieve their rated precision, distance meters (as well as manual and robotic electronic tacheometers) must measure to the (glass prism) reflectors suggested by the respective manufacturers. However, there are cases where the full precision is not required. In such cases, cheaper reflectors might be used once their performance with the intended distance meter has been tested. (Investigate errors in angle measurement and change of reflector constant with distance AND with angular mispointing of reflector, horizontally and vertically.) The (red) acrylic reflector shown in Fig. 18 was successfully used by UNSW with the instrumentation shown in Fig. 19 and some other Wild electronic tacheometers of that time. Electronic Tacheometers Electronic tacheometers have been around since the early 1970s. The early instruments recorded data on telex paper tape or audio cassette tapes. The second generation of electronic tacheometers followed in 1977/8 (Reger 1996). By the mid 1980s, all manufacturers concentrated on this type of instrument. The Wild TC2000 was manufactured from 1983 to 1987 and the Wild TC2002 from 1990 to 1997. Currently, Leica offers the TCA1800, TCA2003, TC2003 and the TDA/TDM5005 for more precise work. (The TDA and TCA models are motorised and can point automatically to reflectors.) Data recording is now slowly changing from PC-cards (PCMCIA-cards) to 'CompactFlash' cards (CF- cards). Figure 19: Early type of surveying robot (Wild TM3000D and DI3000, owned and operated by UNSW) on a reference point of the Mangrove Creek Dam (NSW), automatically measuring to other reference points and the object points (Fig. 18) on the downstream face. Note the (stainless steel) pillar centring system with 5/8 inch Whitworth thread that allows direct mounting of most (if not all) surveying instruments presently on the market. (Reger 1994, Reger & Sippel 1994) Robotic Electronic Tacheometers Robotic electronic tacheometers are equipped with servo motors on the two axes and with automatic pointing systems. Some types can find reflectors autonomously all over the horizon, others need training (or an equivalent data file) to find the reflectors. Fine pointing is automatic in both cases. For the monitoring of dams, the robotic instruments that require an input data file with the approximate pointings is appropriate. This ensures that the Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 19 observation routine is the same in all epochs. Figure 19 shows one of the earlier robotic electronic tacheometers used by UNSW with students for deformation measurements on the Mangrove Creek Dam (NSW). Note the (simple) pillar centring with 5/8 inch Whitworth thread known from surveying tripods. Today, the robotic instruments are smaller, faster and more accurate, particularly as far as the measurements of directions and zenith angles are concerned. Leica (1997) quotes the precision of the automatic target recognition system (ATR) of the TCA2003 and TDA5005 as 1 mm for reflectors closer than 200 m and 2-3 mm for reflectors at 500 m. These instruments can be operated remotely if the appropriate software and communication links are available. Permanently installed robotic instruments are used for the continuous monitoring of dams on tectonic fault lines, in earth quake zones or on sites that are at risk because of tunnelling underneath the dam and reservoir (Brker 2006, Chrzanowski et al. 2006, Duffy et al. 2001, Wilkins et al. 2002). Krickel et al. (2001) measured the reference point network of a German dam (Dreilgerbach) with a robotic electronic tacheometer Zeiss Elta S10 twice, once with manual pointing and once with automatic pointing. The measurements with automatic pointing ('FineLock') were at least as accurate as the manual measurements. The automatic direction measurements (automatic pointing and software for automatic direction measurements) reduced the pure measuring time at instrument stations by 70%. Total field time at the dam was reduced by 20% using robotic measurements. Levelling Instruments For more accurate work in dam deformation measuring schemes, only precision levelling instruments with their corresponding invar staffs should be used. The original Wild N3 precision spirit level was in production from 1929 to 1970 and the updated version from 1973 to 1996 (Dedual 2005). The 10 mm invar staffs are usually 3 m long; 2 m long versions were produced for tunnels and dams as well as shorter ones for the levelling of pillars on dams, for example. The first automatic (compensator) levelling instrument (Ni2) was released by Zeiss (Oberkochen) in 1950. Special observation routines must be followed to overcome some aspects of these instruments. Pre-1984 automatic levels might be affected by the Earth's gravity field (Reger 2006). The most sophisticated automatic level was the Zeiss (Jena) NI002. Digital levels appeared in 1990. By 1992, the first precision digital level (WILD NA3000) and the first invar bar code staffs were available. The Zeiss DiNi12 is presently the most accurate digital level if inherent systematic errors are considered. Users of digital levels for dam deformation surveys should be well aware of the errors of digital AND automatic levels since all digital levels are also automatic levels. The illumination of staffs in galleries remains a problem. Egger & Walser (2005) note that it is essential that levelling runs be carried out independently forward and backward since redundancy is always poor. These authors also note that precision levelling on dams, though simple in principle, is full of surprises. Satellite Signal Receivers Satellite signal receivers using GPS(USA), Glonass(Russia) and/or Galileo(Europe) satellite signals can achieve a precision of about 10-20 mm for the coordinate vector between stations irrespective of observation and processing methods (Egger & Walser 2005). Better precisions can be achieved but require a much larger measuring effort and more sophisticated processing softwares (e.g. Bernese). In small deformation networks, extending 1-2 km in both (horizontal) coordinates, GPS can presently not achieve the precisions of traditional techniques because of a number of systematic errors that cannot easily be modelled (Egger & Walser 2005). Satellite techniques have, however, a number of advantages: stations do not have to be inter-visible and the method is largely independent of weather and day-time. Satellite techniques are suitable to extend the dam monitoring networks to reference points further away in geologically stable ground and to monitor unstable terrain near and along the reservoir. Swiss National Mapping started using GPS in 1989 mainly for the extension of older dam monitoring networks. Schneider & Wiget (1997) give excellent advice on the procedures to be followed for this purpose. Reflecting surfaces (including the water surface in the reservoir) near satellite antennas are a particular difficult problem. Schneider & Wiget (1997) discuss 3D and 2+1D adjustments of combined terrestrial- satellite data and demonstrate the precision of horizontal GPS vectors (maximum length about 4 km) of 1.5 mm (vertical component: 3.0 mm) in the combined adjustment with horizontal directions (1"), zenith angles (1.3") and slope distances (0.3 mm + 0.5 ppm). Since satellite techniques provide the Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 20 results in a 3D geocentric Cartesian coordinate system and the terrestrial data are essentially in local 2D+1D coordinate systems, the deviations of the vertical, the difference between ellipsoidal and geoidal heights and the scale difference between terrestrial distance measurements and the satellite measurements most be taken into account. For these reasons, combined ('hybrid') adjustments require experts. Photogrammetry Typically, aerial photogrammetry provides precisions of 0.01% and 0.015% of the flight path above ground of horizontal and vertical coordinates, respectively (Flotron 1997). With much overlap, a precision (in cm) of 0.05% of the photo scale can be achieved, that is 2 cm for a photo 1:4000. Usually, photogrammetry is too expensive (and often not accurate enough) for the monitoring of the dam proper. However, photogrammetric techniques are ideal for the monitoring of terrain movements near or along the reservoir. If centimetre precision is required, monitoring points have to be marked. If precisions of decimetres are sufficient, no monitoring points need to be marked. Naturally, some (marked) reference points around the area of interest must be available. Fryer & Barlett (1989) reported on the use of terrestrial photogrammetry for the monitoring of the Chichester Dam in NSW. The dam is 254 m long and 43 m high. A precision of 3-8 mm was achieved in all components of the object point coordinates. Laser Scanners Laser scanners became available about 10 years ago. They are instruments that can measure distances, zenith angles and direction in a grid pattern over structures and terrain (Reger 2003). The measurements are fast and do not require reflective targets on the structure or terrain. Reflectors must be placed on a few reference points so that the data can be transformed later into the dam coordinate system. The maximum range of these devices varies greatly between brands and models, typically 100 m to 350 m. Precision gets worse with distance: 5 mm/50 m and 25 mm/200 m are quoted (Reger 2003). Since laser scanners provide 3D models of the area surveyed, they have similar applications as photogrammetry. Schulz & Zogg (2006) discuss laser scanner measurements (with a Leica HDS 3000) of a concrete dam in Switzerland. The instrument was set up 100 m from the dam. The dam was scanned (in 30 minutes) with a point density of about 0.20 m. These authors measured four epochs. The plots of the coordinate differences (i-th epoch minus reference epoch) show values of up to 5 cm near the abutments. The authors attribute these differences to measuring errors (and not to dam deformations). They quote relatively long distance, humidity or wetness of the concrete, angle between the measuring beam and the structure (ideally 90) and inaccuracy of the reference system used as likely sources of the discrepancies. They conclude that, for the time being, the conventional measurements are to be preferred for the monitoring of the dam proper. Laser scanners can be replaced by the cheaper motorised reflectorless electronic tacheometers, which may have longer range and better precision but are much slower. They need to be programmed to allow scanning. Because of the time involved, such instruments usually scan the structure with a much lesser density of points. Laser Trackers Laser Trackers contain a tracking laser interferometer as well as angle encoders on the mirror that keeps the laser beam on the prism (Reger 2003). The tracker follows an EDM prism mounted inside a steel (Taylor-Hobson) sphere as it is moved by an assistant from one measuring point to the next. If survey marks are to be measured, they have to have a conical top so that the sphere can be placed in it. (The CERN centring system has this facility built in.) Starting point for measurements is a reference point on the instrument. The 3D accuracy is about 0.03 mm per 10 m. The range of the trackers is typically 35 m. Some trackers feature an additional precision distance meter on board so that a measurement sequence does not have to be restarted after a beam interruption. Laser trackers are used in metrology but could be used to replace theodolites and the DISTINVAR in the gallery traverses in concrete dams. More likely than not, this would require some changes to the traversing networks. Hanging and Inverted Pendulums Since pendulums are the measuring instruments of choice inside concrete dams, they are briefly mentioned, even though they are not usually operated and installed by surveyors. These devices are used to measure horizontal displacements. The accuracy is 0.2 mm. The tension in the wires varies between 200 N and 2000 N. Manual and automatic read-out devices are available. Some good advice on installation and operation can be found in the Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 21 recent publication of the Swiss Committee on Dams (2005). Both types of pendulum equipment are commercially available (e.g. see www.huggenberger.com, under 'inclination'). Wire Alignment The alignment with horizontal steel wires allows to detect the horizontal movements in galleries and, if it is the only option, on the crest of (concrete) dams. Jakob (1969) and Milev (1985) summarised Bombcinsij' s development of a wire alignment system for the Kuibyschew Dam in the former USSR. A 1 mm diameter wire was used over 600 m, at 600 N tension, with an alignment precision of 0.3 mm. At least three floating supports (see Milev 1985 for photo) were used. The offsets were measured at 20 positions. Deumlich (1976) reported on a wire alignment over 825 m in the former USSR (Krasnojarsk), with float support every 60 m, a wire diameter of 1.2 mm, a tension of 2000 N (200 kgf) and an alignment precision of a few 0.1 mm. For the Kirov Dam in Kirgiskaya the wire was supported by floats in a water channel along the dam. This eliminates sag, reduces the influence external influences on the wire and increases accuracy to 0.1 mm (Deumlich 1976). The Swiss Committee on Dams (2005) quotes the accuracy of wire alignment as 0.2 mm and states that spans of 200 m are achievable with sags of less than 0.20 m. In Switzerland, the wire is tensioned at one end through a pulley system. Longer lengths require intermediate supports that allow free lateral movements. Float suspension as in inverted pendulums can be used. The measuring precision is independent of length and refraction; automatic reading and data transfer are possible. Air currents in galleries should be minimised since they can affect the wire's position. No commercial equipment seems to be available. But installations in Switzerland (4 gravity concrete dams) date back many years. The dam ('Schrh') shown in Figs. 1 and 2 was retrofitted with a wire alignment system on the crest in 1973. The system on the crest of another dam ('Rempen') is shown in Fig. 5. Photos of the key components of a mechanical alignment system in a dam gallery can be found in the publication by the Swiss Committee on Dams (2005) together with advice on installation and use. CERN in Geneva is developing a new generation of a wire alignment system for the new Compact Linear Collider (CLIC). They use a 60 m long carbon fibre wire of 0.32 mm diameter to measure horizontal and vertical displacements in real-time, using capacitive pick-offs. Tension is applied by 6 kg weights (60 N). The repeatability is 1 micrometre. In its present form (see www.fogal.fr) this device is not rugged enough, and does not have a large enough measuring range for dam measurements. However, the potential of measuring two components (also replacing levelling and hydrostatic levelling) in real-time is attractive if the instrumentation can be transplanted from clean laboratories to not so clean and dry dam sites. Hydrostatic Levelling Hydrostatic levelling has been in use for a long time. Commercial manual and automatic equipment for deformation measurements is available from Freiberger Przisionsmechanik (ww.fpm.de), for example. The resolution of these systems is 0.01 mm. According to the Swiss Committee on Dams (2005), the measuring units have to be connected by three tubes (3-4 mm diameter for the tubes carrying the water and 6-7 mm for the air tubes). The water used has to be degassed. Unequal temperatures along the tubes are another problem. Inside the dam, hydrostatic levelling can replace precise levelling if properly handled. It has the advantage, that the readings can be made automatically and be communicated to a control centre elsewhere. In particle physics laboratories, hydrostatic levelling systems with resolutions at the micrometre level have become very popular. (See www.emp-winterthur.ch and www.fogale.fr, for example.) Invar Distance Measurement There are two modern distance measuring instruments that are based on invar wires. The DISTINVAR was developed in the 1960s by the European Organization of Nuclear Research (CERN) in Geneva. It is commercially available (including an electronic readout version) from the company Geodes i e I ndus t r i el l e (www.geodesie.com). The Distometer was developed by the Swiss Institute of Technology and marketed through KERN & Co Ltd. (I do not know if the Distometer is still available commercially.) Both devices are depicted in Keller (1978), use wires of 1.65 mm diameter and can measure lengths from 1 to 30 m with a measuring range of 50 mm (DISTINVAR) and 100 mm (Distometer). The instruments measure only the changes of length between epochs. Since the lengths of the wires are fixed, the point spacing (in dam galleries for example) Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 22 must be standardised to avoid the need to buy many different lengths. For absolute measurements, a calibration of the system is required. It is advisable to have 2-3 wires of the same length and to measure a number of stable reference lines near the dam before and after the dam measurements. (Fig. 9 shows the location of a tunnel with a calibration facility for invar wires.) Further advise on the measurements with these devices can be found in Egger & Walser (2005). Optical Alignment The optical alignment (across the crest of a dam) qualifies as a simple technique. Dedicated equipment is available from the firm Freiberger Przisionsmechanik in Freiberg, Germany (www.fpm.de). As expected, this equipment does use the Freiberger Ball centring for the alignment telescope and the fixed target (on both ends of the alignment axis). The ball diameter is, however, larger than the 16.53 mm (Lang 1929) of the traditional centring bolts advertised (in the past) by Wild and Kern. Freiberger Przisionsmechanik (www.fpm.de) also supplies special marks for the (road on the) crest and an adjustable target that allows to measure the offset of ground marks from the line. The accuracy of the method depends on the length of the alignment axis and the horizontal refraction errors. As mentioned before, refraction errors of 14 mm at 300 m have been found experimentally (Biedermann 1997). Egger & Walser (2005) suggest that optical alignment is obsolete and is better replaced by other techniques. Optical Plummets The accuracy of optical plummets is given as 0.5 to 1 mm by Welsch et al. (2000, p. 59). Although surveying instrument manufacturers do sell specialised zenith (up) and/or nadir (down) plummets, plumbing upwards can be done equally (if not more) precise with theodolites that have a diagonal eyepiece. If the dam is equipped with vertical shafts, then wi re pendul ums are t he preferred instrumentation for many reasons. Optical plummets (or theodolites or electronic tacheometers fitted with diagonal eyepieces) may have to be used to connect the top gallery to the crest (if the pendulum attachment point is not at crest level). Such a case was discussed in connection with Fig. 9. Centring Systems Freiberger Ball Centring (~16.5 mm diameter) In the 1920s, the Swiss dams were equipped with brass centring bolts with a bore diameter to fit the Freiberger Ball centring (diameter 16.53 mm) of the theodolite 'without constraint but with a tight fit' (Lang 1929). The Hildebrand (vernier) theodolites supported this pillar centring as did the Wild T3 precision theodolites afterwards. (The T3 owned by UNSW has a centring ball of 16.52 mm diameter.) Wild supported this pillar centring kit (Fig. 20) at least until 1986. The centring repeatability is given as 0.02 mm when used with the Freiberger Ball attachment of theodolites (or tribrachs). Untersee (1951, 1975) shows how the Wild T3 fits the pillar bolt (1) shown in Fig. 20. According to the Leica-Geosystems web page (September 2006), the products shown in Fig. 20 are no longer available. Figure 20: Special Wild equipment for monitoring and deformation surveys: Wild pillar centring bolt (1), with ~16.5 mm bore and protective cap, spot bubble for levelling bolt during attachment to pillar (2), bullseye target for centring bolts (3), bullseye target for vertical dam faces (4). (after Wild 1986) For all other Wild theodolites available in 1986 (e.g T2, T2000, T2000S, TC2000) Wild suggested a special tribrach for centring on pillars with the Freiberger Ball (diameter ~16.5 mm) centring (Fig. 21). This arrangement is subject to two random centring errors, one between the instrument and the dish of the tribrach and one between the ball and the bolt, and a systematic one (ball in centre of tribrach dish). Naturally, this centring system also allows to fit most non-Wild/Leica equipment onto pillar bolts. Sadly, the equipment shown in Fig. 21 is no longer listed in Leica's 2000 'accessories' booklet nor on the current Leica web page. There is some concern about the rotational stability of the set-up shown in Fig. 21 when modern motorised theodolites are inserted; the angular accelerations and decelerations of these robotic instruments Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 23 introduce torsional forces that may cause slipping of the footscrews' disks on the pillar. This rotational slipping would depend on the weight of the instrument, the torque applied by the servo motor (around the vertical axis) and the smoothness of the pillar surface. Kern supported the same bolt system (diameter ~16.5 mm) with the pillar bolt system shown in Fig. 22. The centring bolt has a protecting cap. The set includes three brass bolts for the feet of the trivet (Fig. 23). Two of these have flat surfaces, one has a groove to provide torsional stability. For example, the reference points in the network shown in Figure 9 are fitted with the centring kit of Fig. 22. Figure 21: Wild GDF24 Freiberger ball centring device (diameter ~16.5 mm) with detachable base plate and three discs for the foot screws. A small scale allows to measure the height of the tribrach over the bolt. With this tribrach, all instruments that fit the Wild tribrach dish can be mounted on the bolt shown in Fig. 20. (after Wild 1986) Figure 22: Centring bolt and three support bolts (for the feet of the trivet) concreted into the top of an observation pillar. Central bore of bolt fits 16.53 mm diameter Freiberger ball centring. (after Kern 1971) To use the Freiberger ball centring bolts with KERN instruments, the trivet shown in Fig. 23 had to be mounted over the bolt. The trivet features a spot bubble (1) to level the top plate, a short centring rod (2) that can be lowered into the bolt and a lever (3) to lock the ball and socket top. Again, the height of the trivet above bolt can be read off a scale. The centring precision of the trivet was better than 0.1 mm (Kern 1971). The trivet shown in Fig. 23 can be used to mount all those instruments on Freiberger ball centring pillars, that fit the Kern centring system. This includes the precision distance meter Kern Mekometer ME5000. Figure 23: Kern trivet for centring on Freiberger ball bolts of ~16.5 mm diameter. The trivet features a spot bubble (1) to level the top plate, a short centring rod (2) that can be lowered into the bolt and a lever (3) to lock the ball and socket top. The height of the trivet above bolt can be read off a scale. (after Kern 1971) Kern Pillar Centring Plates About 1973, Kern introduced an alternative pillar centring system with their 158 mm diameter aluminium pillar centring plates. These plates were permanently attached (screwed) to observation pillars and, sometimes, object points. Important is that the top of the plate is horizontal after mounting (and that the pillar does not go off level with time). The Kern pillar plate is shown in Fig. 13 (and, later, in Fig. 24). The exterior network of the Gigerwald Dam shown in Figs. 11 and 12 is equipped with the Kern pillar plates. Having a number of monitoring schemes based on the Kern pillar plates lead to the problem of fixing other makes of instruments onto them. One way was to use a special Wild tribrach, that features the Kern centring at the base. This Wild GDF25K tribrach (Fig. 24) featured one fixed 'footscrew' and two adjustable ones so that the height of the tribrach dish did not change from one set-up to the next (provided that the tribrach is always mounted in the same orientation). In addition, the central disk in the base of this tribrach can be removed so that the (10 mm) Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 24 ball centring extension for the theodolites T2000, T2000S and T3000 can be used. These theodolites (like the T3 before) carry a central thread underneath that allows the attachment of the ball centring extension to the instrument body. This centring rod then goes directly into the central bore of the Kern centring plate and centres the theodolite (not the tribrach) to 0.2 mm over the pillar plate. This tribrach was available in 1986 (Wild 1986) but not any longer in 2000 (Leica 2000). Figure 24: Wild GDF25K tribrach with one fixed 'footscrew' and two adjustable ones. The (10 mm diameter) ball centring extension shown on the left is explained in the text. This tribrach allows to fit any Wild-tribrach compatible instruments to the Kern pillar plate or (Kern tripods for that matter). (after Wild 1986) Centring on fixed 5/8 inch Whitworth Thread Most current surveying instruments feature Wild-type tribrachs, which carry a female 5/8 inch Whitworth thread (with 11 turns over 25.4 mm, after British Standard BS 84) underneath for the attachment to tripods. Therefore, it suggests itself to equip the top of pillars with a secure (stainless steel) 5/8 inch Whitworth thread, usually centred on a solid and smooth stainless steel plate. The pillars of trigonometric stations in NSW are so equipped as are the observation pillars (and downstream face points) on the Mangrove Creek Dam in NSW, for example. Figures 18 and 19 show examples of the latter. I do not know of any investigations into the repeatability of the centring of instruments on such pillars or marks. The centring precision is certainly worse than the 0.02 mm of the Freiberger Ball centring and somewhat worse than the 0.1 mm centring precision of the Kern trivets (Fig. 23) on the pillar bolts (Fig. 22). This simple pillar centring system might be entirely appropriate where centring errors of a few tenths of a millimetre can be tolerated, namely where no short rays are involved and where no equipment of highest precision is being used. Since embankment dams typically show larger deformations than concrete dams, this simple pillar centring might be more appropriate for the former. Galleries in new concrete dams may have to be equipped with the CERN centring system discussed below (since the 16.5 mm bolts and the Kern pillar centring plates (158 mm diameter) are no longer commercially available). For robotic instruments with fast acting servo motors (large rotational accelerations) this simple centring system might provide more torsional stability than the system shown above. Figure 25: Adaptor for instrument tribrachs with 5/8 inch Whitworth thread that fits (and locks in) the 30 mm CERN centring system (see Fig. 8, from www.geodesie.com, 'Universal Fixation for Wild') CERN (30.00 mm) Centring System The system shown in Fig. 8 is the current version of the 30 mm cylindrical CERN centring system. It was redesigned to allow the centring cylinder to be levelled (e.g. made vertical). The centring accuracy is at the 0.01 mm level. Taylor-Hobson spheres (which can contain circular targets or EDM prisms) can be put into the conical surface at the top with a height repeatability of 0.02 mm (CERN 1974). This is of advantage if laser trackers are to be used. The CERN centring system is very sturdy since it has to serve the CERN DISTINVAR, which applies a tension of 150 N to the invar wire (against the centring cylinder). In deformation measurements, the CERN centring system has been used in galleries of concrete dams for the traversing network. The CERN centring system is commercially available (from Geodesie Industrielle in Geneva, Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 25 Switzerland, www.geodesie.com). Figure 25 shows how standard surveying equipment can be attached (and locked!) in the CERN centring device with adaptor plates fitted with the customary 5/8 inch Whitworth thread on top. The likely centring errors between tribrach and the 5/8 inch Whitworth thread have been discussed in the previous section. It is worse than the 0.01 mm centring between the CERN socket (Fig. 8) and the adaptor shown in Fig. 25. 6 DEFINING ENGINEERING NEEDS As early as possible in the design phase of a dam, the engineering needs should be communicated to the surveyor. Early discussions are essential since the positions of shafts for pendulum systems and the locations of galleries inside the dam have to interface with the geodetic deformation measurement scheme, for example. 1D, 2D or 3D? Although the actual radial (upstream- downstream) movements as well as the up-down movements (settlements) are of primary interest, the lateral (tangential) components also need to be known since an abnormal dam behaviour can include rotations and block shifts. The scheme must also cover any potentially unstable terrain around the dam and unstable slopes or rock formation along the reservoir. Since, the geodetic measuring scheme has to provide the reference for any future specialised measuring schemes established in response to abnormal behaviour, it should be as redundant and wide ranging as possible. The geodetic network on large dams should provide 3D information in an absolute sense, that is against stable terrain outside the influence zones of dam and reservoir. Magnitude of Anticipated Deformations The engineers must indicate the expected deformations and displacements of critical points on the planned dam for the different stages of the dam that need to be measured since the required measuring precision must relate to them. Measuring stages of interest are: total long-term deformation before and after first filling (possibly additional measurements at the 1/2 or 1/3 and 2/3 points of filling) before and after the first emptying (possibly additional measurements at the 1/2 or 1/3 and 2/3 points of emptying) seasonal changes (summer, winter; temperature often dominant influence) Once the expected deformations for the above events are defined, the required surveying precision can be estimated with a rule-of-thumb design equation. From the expected deformations of the listed events above, the smallest one is the minimal movement (dy) of object points that the geodetic measurements have to be able to resolve. The measuring precision (sy, in one epoch, one standard deviation) can be estimated conservatively as (Welsch et al. 2000 Eq. 1.3-3, Pelzer et al. 1987, Eq. 2.1-3) sy dy/5 (1) This formula considers that two epochs of measurements are needed to obtain a change in the position of an object point and that changes larger than dy must be significant at 95% confidence level. Pelzer et al. (1987) also give other rule-of-thumb advice that is useful when designing deformation networks. (For English translation, see Reger 1997). Based on Eq. (1), the surveyor can select appropriate measuring instruments and techniques and design a monitoring network. Using least squares adjustment programs, the expected precision (95% confidence error ellipses) of the deformation vectors between two epochs can be simulated (without actual observations). The 95% confidence ellipses of the critical object points must all be smaller than the (dy) specified. During the preliminary analysis of the geodetic network, the reliability and redundancy of the network design must be checked. The suggestions by Carosio & Dupraz (1997) are helpful in this regard. The potential loss of one or two reference points must be factored in. Density of Object Points Recent trends show that, wherever possible, the monitoring points are established inside the dam, where galleries are interfaced with pendulums in vertical shafts. Ideally, the pendulum measurements connect to horizontal wire alignments (rather than to surveying precision traverses) in galleries or along walls on the crest. So, ideally, no object points will be placed on the outside of dams other than on the crest. This approach suits gravity concrete dams and some concrete arch dams. See Figs. 6, 10, 11 and 12. Since concrete dams are poured in blocks, it is often required to place the Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 26 internal points in such a way that each block is monitored. For embankment dams, the traditional placement of survey marks in rows (horizontal profiles) and columns (vertical profiles) dates back to the 1920s and is well established and proven. Six vertical and four horizontal lines of object points (e.g. see Figs. 1 and 2) on the downstream face of dams with no galleries and no pendulum shafts are a starting point for the considerations. Naturally, the number of object points will depend on the size of the dam. Figure 9 shows an arrangement for a large dam (780 m long, 120 m high). The asymmetric layout of the point profiles in Fig. 9 is due to the increasing dam height towards the East. Short and (Potential) Long Term Needs The geodetic measurements will be more frequent during the construction of the dam and during the first filling and emptying of the reservoir. This is to check the actual deformations against the expected ('design') ones. After that, geodetic measurements (full network or reduced network) are less frequent unless, naturally, an abnormal behaviour is detected (by other means). Since the geodetic network provides the reference for any emergency measurements, it is sensible (Biedermann 1996) to measure a reduced network at least every 5 years (during safety checks of the dam) and the full network every 15 years (when the reservoir is emptied). The geodetic deformation network must be designed for the same working life as the dam and must be maintained and protected accordingly. Coordinate System, Grid Orientation It is customary to use a local coordinate system for the dam construction and the geodetic measurements. This is usually split into an X-Y horizontal coordinate system (sometimes with a false origin to avoid equivalent x and y values) and a vertical one, the latter based on heights above sea level. For the vertical datum, the reference height (elevation above sea level of the chosen X-Y plane coordinate system) should be clearly specified since all length measurements have to be reduced to this elevation. Depending on the elevation of the dam site, selection of sea level as height reference might not be optimal. For orientation, the symmetry axes of the dam are suitable. It should be noted that the extended geodetic network might involve GPS satellite signal observations and that proper transformation of the geocentric satellite coordinate system into the local X-Y & H system is not trivial. Reporting Format Section 4 gives details of the reporting of survey measurements and data. The engineers and surveyors should discuss and specify the extent and form of the reports, considering Section 4. The 'results of measurements' (see Section 4) might refer to the mean values of the observed data and/or the set of adjusted coordinates (of a new epoch). A list of the current coordinates of all point should always be supplied (with associated 95% confidence intervals and, possibly, standard deviations). The result of the actual deformation analysis (between the previous and the current epoch) should be reported. A graph similar to Fig. 14 shows visually the 95% confidence ellipses of all (horizontal) deformation vectors between two epochs as well as the significant deformation (vectors). The same information should also be given in table form. The height changes should be given in a similar form. Graphic displays are well suited for a visual representation of the long term behaviour of some or all object points. Diagrams such as the one shown in Fig 2, are beautiful but can handle only 2-3 epochs. The diagrams shown in Figs. 15-17 show very graphically (and in 2D in Figs. 15 and 16), how the displacements progress with time. To cover the whole dam, these diagrams would have to be duplicated for the remaining groups of points, if so desired by the engineers. Plots of 1D displacements with time are sometimes used but lack the geometric information and 2D aspects. As long as the deformations are given and plotted in 2D, there is no need to split the vectors into its radial and tangential components. Frequency of Data Collection (Epochs) Since the suggestions by the Swiss regulatory authority for dams minimise the cost (of the expensive geodetic measurements by specialist surveyors) to owners without increasing the safety risk, they are repeated here. Concrete dams with internal pendulums and horizontal galleries (equipped with wire alignment systems or geodetic precision traverses), the complete interior network (pendulums, alignment (wire or traverse), (possibly, together with a reduced exterior network,) should be measured every five years. The full interior and exterior network should be measured at least every fifteen years. (The pendulums are read at least every month by non-surveyors.) Note that these dams do not Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 27 carry survey marks on the upstream and downstream faces. Embankment dams (or concrete dams with no galleries and/or pendulums) with object points at least on the downstream face: On these dams, the monthly measurements have to be carried out with simple techniques (ideally by non-surveyors). Some simple techniques are outlined in Section 2. Since these simple survey techniques carried out by non-surveyors are less accurate than the pendulum and wire alignment readings, Biedermann (1997) proposes to measure the 'interior network' (of the dam points) and a reduced exterior net at least every year (rather than every 5 years) and the full interior and exterior network at least every 15 years. The Swiss Committee on Dams (2005) adds an annual (or more frequent) levelling to the list of required observations. If such dams are equipped with a permanent pillar with a permanently installed robotic el ect roni c t acheomet er for aut omat i c continuous (or daily, or weekly) measurements (with data communication to the dam operator), then the cycle of reduced geodetic measurements could go back to 5 years. Here, the robotic measurements replace the monthly pendulum readings by dam staff. Full geodetic measurements again every 15 years. For dams with heights of less than 10-15 m, the Swiss Committee on Dams (2005) suggests the measurement of the deformation once a year (rather than monthly as above) if only small and insignificant deformations are expected. However, monthly seepage and piezometer data are still required under these circumstances. 7 RESOLVING SURVEYING ISSUES There are some aspects that the surveyor has to resolve in discussions with the engineers. Epochs All geodetic measurements taken at a particular time constitute one epoch of measurements. Since changes of the dam during the measurements of an epoch must be minimised, the water level should be kept constant. Depending on the size of the geodetic deformation network, the number of reference and object points and the sophistication of the measuring techniques, the measurement of one epoch may take a few days to two weeks. Accuracy Versus Precision ISO Uncertainty All survey measurements are affected by unavoidable random errors (the smaller the more 'precise' an instrument is). This repeatability of the observations is referred to as precision. Accuracy is a measure for the agreement with the 'true value' or the national standard for the corresponding unit of measurement. (Electronic distance meters and levelling staff should be calibrated against the national standard of length, for example.) Even if calibrated instruments are used, some systematic errors can affect the measurements. Some of these than be neutralised by smart field techniques, other (such as atmospheric refraction effects) not. The old rule that one should always use the same equipment on particular points and always measure the same way (Lang 1929) still applies. This ensures that some systematic errors are the same in all epochs and, thus, cancel each other when differences of two epochs are taken. Since the modern deformation analysis is based on network adjustments (where systematic errors inflate the output precisions) the deformation vectors might be more accurate that the confidence ellipses of the vector suggest. Recently, the International Organization for Standardization (ISO 1995) published a Guide to the Expression of Uncertainty where the uncertainty (of measurement) is defined as a 'parameter, associated with the result of a measurement that characterises the dispersion of the values that could be reasonably be attributed to the measurand'. It is suggested to use the 'expanded uncertainty' as defined by ISO (1995) in any new documentations and reports. Relative Versus Absolute Displacements The pendulums inside dams give displacements relative to their suspension points. If inverted (floating) pendulums are anchored deep below the dam in bedrock, then the pendulums can be referred to a quasi-absolute reference. Similarly, alignment measurements (wire alignment or precision traverses) in galleries give only quasi- absolute deformations, if the galleries extend deep into the side of the valley on both sides. Since dams can shift as a block, absolute displacements can only be obtained by anchoring the geodetic network well away from the influence zone of dam and reservoir. Presently, there is a tendency to extend existing monitoring networks (typically extending a few hundred metres around the dam, e.g. Fig. 1). Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 28 Setting-Out versus Monitoring Networks The geodetic deformation measurements require the installation of a number of double-wall concrete pillars near the dam site. The cost of construction of these pillars can be shared and the stability of the pillars increased (settling, ageing), if pillar locations can be found that serve the construction (setting-out) as well as the monitoring phase. Stable ground and solid protection are required. Costs Versus Accuracy and Frequency The cost s of geodet i c deformat i on measurements increase with the precision that is required to determine critical deformations with confidence, since higher precision requires more precise equipment and, often, specialist observers. In consequence, the engineer should specify the expected deformations at key points and the events that need to be measured. The measuring precision and equipment can then be selected by the surveyors. To keep costs down, the aim is to just meet the requirements. The frequency of the geodetic measurements should me minimised to keep costs at bay, as suggested above. Having wire pendulums and wire alignment systems installed increases precision, reduces the hours surveyors spend on the dam and easily allows conversion to continuous measurements. Design of the Survey Network The design of monitoring schemes is discussed in Section 3 in detail. It is the surveyors task to design the geometry of the network. Through computer simulations, the redundancy, reliability and precision of the deformations are checked. Since the network must service the dam for its life, it is essential that the network can cope with the loss of one or two reference points in its life. The performance in case of the loss of a reference point should be simulated and the network improved, if necessary. A robust, reliable and redundant design is required. Satellite measuring techniques might be planned to extend the net further away from the dam, without inter-visibility constraints. Reference and Object Points All points of the network, that need to be visited by surveyors, must be easily accessible all year round. Occupational safety must be observed. The stability of the points must be ensured by proper design and careful selection of the sites. Permanency of the points is essential considering the 50-100+ year service life of the dam. The height of the pillars must be optimised for convenient manual observations. The engineers must assist with the protection of the object and reference points from construction work and vandals. Solid steel covers might be required to protect the pillars from the elements and vandals. Centring Systems The centring precision is part of the design of the network. The centring system should give the required precision and be as simple as possible. Since some of the traditional pillar centring systems, that were previously supplied by Kern and Wild, do not seem to be available any more, a fixed 5/8 inch male thread on a stainless steel plate might be sufficient for embankment dams. The centring precision should be confirmed by experiment. The same applies to monitoring points on the downstream face of embankment dams. For the high precision traverses (and the connection to and in-between pendulums) in the galleries of concrete dams, more sophisticated centring systems might be required. The CERN system is commercially available (or can be manufactured in-house). Figure 26: 'Traversing' target (by Kern, attached to a Kern trivet) on a reference point of a dam deformation network (after Kern 1971, Fig. 19) 'Targets' for Geodetic Measurements Targets (see Fig. 26, on points that are being observed to) are used when only angular measurements (horizontal directions and/or zenith angles) are required (or if angular measurements to EDM reflectors are not accurate enough). In the early days of geodetic dam deformation measurements, long before the introduction of infrared distance meters in Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 29 1969 and of precision distance meters in 1973, only directions and zenith angles were measured between the reference stations. In consequence, the targets had to cater for angular measurements only. Wild (see (3) in Fig. 20) advertised the traditional circular target bolts (that cannot be levelled) as late as 1986 for pillars and monitoring points. Today, distances between reference stations have grown, requiring bigger targets. Traversing targets, than can be accurately centred and levelled, are preferred for manual angular measurements. Figure 26 shows a typical 'traversing' target. According to KERN (1971) this target is suitable for sighting distances from 20 m to 1 km. The eccentricity between the symmetry axis of the target and the cylinder of the forced centring does not exceed 0.3-0.4 mm. For shorter sights, smaller and better centred targets were sold by KERN. Figure 27: Spherical targets (bright orange, see arrow) on the object points on the upstream face of the Mattmark embankment dam in Switzerland. The numbers on the pillars serve for point identification during measurements. Note the (hanging) glacier visible at the top. One reference point (not shown in Fig. 9), situated in the path of falling ice masses, had to be abandoned due to safety concerns. (May 1970, photo by author) In the case of embankment dams with no pendulums or wire alignment systems, a large number of object points might have to be monitored from the reference points. Since it is very time consuming to turn the traditional targets (see (3) in Fig. 20) on a dam when the surveyor moves from one reference point to another, brightly coloured three-dimensional spherical target (Fig. 27) were some times used. Naturally, such targets again are only for angular measurements. Spherical targets were once sold by Huggenberger AG, but not any more if the information on their web site (www.huggenberger.com) is correct. Wild also supplied circular (bullseye') targets for the faces of concrete dams (see (4) in Fig. 20). These are no longer available but have withstood the test of times on many concrete dams. Bolts like that suit angular measurements, which, today, can only be carried out manually. Where possible, concrete dams should be equipped with internal networks with pendulums and survey systems in the horizontal galleries rather than with marks on the downstream face. (Retro-fitting of wire alignment and pendulum systems might be appropriate.) Bolts in concrete and rock faces may have to be used in new dams where pendulums and/or wire alignments cannot be installed. Unless they can easily be serviced and replaced, the attachment of reflectors in the faces of concrete dams is not recommended since reflectors get dirty, may get damaged (e.g. by dropped stones) and the reflective coating deteriorates with time and humidity. Ref l ect ors as ' Target s' : For distance measurements as well as for robotic angular measurements, EDM glass prism reflectors are required. The reflectors must be carefully levelled and pointed to the measuring stations, in two directions. The poor pointing of reflectors not only generates errors in distances but also errors in measured horizontal directions and zenith angles. (For details see Reger 1996.) Depending on the length of lines measured and the required measurement precision, 'precision' reflectors should be used. According to Leica (2000), the GPH1P prism with GZR3 carrier and GDF21 tribrach has a 3D system accuracy of 0.3 mm. Figure 28 shows an earlier version of the Wild GPH1P reflector. The quoted 'system accuracy' is likely to include the repeatability of the reflector constant, the centring of the reflector's axis relative to vertical and horizontal axes and the centring in the tribrach. The GPH1P reflector has a metal housing and carries a quality gun sight to assist with the pointing. This type of equipment should be used where such accuracy is required. Standard EDM reflectors are likely to have a Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 30 'system' accuracy' of typically better than 1 mm. Since the reflectors must face the instrument(s), all reflectors (and targets for that matter) must be re-pointed when the instrument is moved from one station to another. Figure 28: Precision EDM reflector (rotatable about two axes) on a ground monitoring point (after Egger 1997). Figure 29: Monitoring point on a large concrete block in an extreme avalanche slope of a continuous measurement system installed at the Nalps Dam in Switzerland. The continuous measurements monitor the length changes of lines (at different elevation) across the valley in the dam area whilst the new Gotthard base tunnel is being drilled underneath the dam. (after Brker 2006) If solely robotic measurements are carried out, only EDM reflectors, properly levelled and pointed are required. Although 360 reflectors are available, they introduce errors in angular and distance measurements. Depending on the orientation of the prism, errors of 2 mm and 3 mm in the horizontal and vertical coordinates measured with automatic pointing (Leica TCA instruments to Leica GZR4 360 reflectors) were reported by Favre & Flach (1999). 360 prisms are not recommended for the measurements between reference points (unless the required precision is not better than, say, 5 mm). If EDM is used to measure to the object points, there must be as many prisms available as the maximum number of reference and object points visible from any point occupied by the instrument (e.g. reference points). Consider Fig. 9 in this context. The manual and robotic electronic tacheometers might be able to measure successfully to cheaper reflectors, such as the one shown in Fig. 18. If cheaper options are considered, the achievable precision must be determined by experiment and the inherent angle/distance errors must be determined as functions of the incidence angle (vertical and horizontal) as must be the change of the reflector constant with distance. (See Reger (1996) for details.) Note that the reflector shown in Fig. 18 is unidirectional and must be rotated each time the surveyor moves the instrument to a new observing station. Since reflectors deteriorate if left on structures for years, they are normally not left on dams between the measurements. However, if robotic electronic tacheometers are installed for around-the-clock measurements, then the reflectors must be permanently installed. Depending on what is to be monitored, these reflectors may have to be mounted on natural terrain, rock or man-made structures. Additional problems arise if the site experiences snow fall in winter. Figure 29 shows that the supports of permanent prisms can be substantial, if protection from avalanches has to be provided. Changes in Geodetic Measurement or Scheme Because of the long life of dams, changes in technology and the ever shrinking service life of electronic surveying equipment, the instrumentation used for the geodetic measurements will necessarily change over the 50 to 100 year life of a dam. Whenever si gni f i cant changes i n pr ocedur es, instrumentation and/or network design are implemented, it is strongly recommended to measure one epoch with the old and the new arrangements so that the continuity of the deformation results can be demonstrated and assured. 8 OUTLOOK The changes in dam deformation measurement during the last 80 years include a move towards direct measuring devices, such as pendulum and Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 31 wire alignment systems, to allow the (monthly) measurement of critical deformations with high precision and by non-surveyors. The geodetic measuring schemes are now measured less frequently but, often, extended (frequently with satellite techniques with no line-of-sight restrictions) to stable ground further away from the dam. Since the geodetic schemes provide absolute data, they serve as a reference whenever an abnormal behaviour of a dam requires repeated and, sometimes, extended measurements. In consequence, it is essential that the geodetic measurement schemes are well appointed and maintained. The pendulum and wire alignment systems cannot be installed in concrete dams that lack vertical shafts and horizontal galleries. Where an installation is feasible, retrofitting of such devices should be considered. (A wire alignment along the parapet of straight dams (fill-type and small concrete type) might be possible.) Otherwise, the other simple techniques proposed in Section 2 should be considered. Permanent installation of a robotic electronic tacheometer is always an option. Today, robotic surveying instruments are routinely used for the continuous monitoring of dam sites in tectonically active zones. The 'Guidelines for Dam Instrumentation and Monitoring Systems' (ANCOLD 1983) contain some information on the geodetic (surveying) aspects. Some of the recommendations should be revised, such as the requirement to measure angles six times (4.1.2.8 in ANCOLD 1983) and the proposal to use optical plummets (4.1.3.10 in ANCOLD 1983). The latter might be replaced by the simple zenith angle measurements shown here in Fig. 5b. The advice on the physical construction of survey monuments is sound, but often include the centring bolts shown in Fig. 20(1) that are, now, no longer commercially available. The diagrams of key centring components should show the critical dimensions (so that they can be manufactured in-house if the devices are no longer commercially available). The ANCOLD (1983) guidelines contain no suggestions on how to mount surveying equipment on the 'Wild' bolts of the type shown in Fig. 20(1). Some of the mounting procedures currently used in Australia may not be optimal and may cause a loss of centring accuracy. The 5/8 inch male Whitworth thread may be considered in future where the best possible centring accuracy is not required. The ANCOLD (1983) guidelines do not cover the analysis procedures. Since there is some doubt that rigorous deformation analyses are carried out on all Australian dams, some guidelines on that aspect might be helpful. Also, any future editions of the guidelines (ANCOLD 1983) should express the uncertainty ('precision') of data and results at 95% confidence level or in terms of the 'expanded uncertainty' as defined by the ISO (1995) 'Guide to the Expression of Uncertainty in Measurement'. 9 REFERENCES Aeschlimann, H. 1988. 'Reference Standard (Mekometer) ME5000'. Publ. No. 303 e, Kern & Co. Ltd. (now Leica-Geosystems), Aarau, Switzerland, 25 pages ANCOLD. 1983. Guidelines for Dam Instrumentation and Monitoring Systems. Australian National Committee on Large Dams, May 1983, 88 pages. Biedermann, R. 1985. 'Backfitting of Instrumentation'. In: Swiss National Committee on Large Dams (ed.), Swiss Dams - Monitoring and Maintenance. Published on the occasion of the 15th International Congress on Large Dams 1985, Lausanne, Switzerland, pp. 174-183 Biedermann, R. 1993. 'Introduction' & 'Surveillance des retenus'. In: Swiss National Committee on Large Dams - Working Group on Dam Monitoring (ed.), Mesure de dformations godsiques et photogrammetriques pour la surveillance des ouvrages de retenu, Wasser Energie Luft (eau nergie air), 85(9): 183-192 (in French and German) Biedermann, R. 1996. 'Die geodtische Deformationsmessung als Element der berwachung von Stauanlagen' (Geodetic Deformation Measurement as an Element of the Surveillance of Water Storage Dams). In: Brandsttter, Brunner, Schelling (eds.), Ingenieurvermessung 96 Beitrge zum XII. Internationalen Kurs fr Ingenieurvermessung, Graz, Austria, 9. - 14. September 1996, Ferd. Dmmler Verlag, Bonn, Vol. 1, pages B12/1- B12/9 (in German) Biedermann, R. 1997. 'Introduction' & 'The Monitoring of Dams'. In: Swiss National Committee on Large Dams - Working Group on Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 32 Dam Monitoring (ed.), The Geodetic and Photogrammetric Deformation Measurements of Dams, published on the occasion of the 19th ICOLD Congress, Florence, Italy, pp. 1 - 12 (Arbeitsgruppe Talsperrenbeobachtung, C/- Bundesamt fr Wasserwirtschaft, P. O. Box, CH-2501 Biel, Switzerland) Brker, F. 2006. 'Anspruchsvolle berwachungsaufgabe beim Project AlpTransit Gotthard' (Demanding Monitoring Task for the Gotthard Alp Transit (Tunnel) Project), Gomatique Suisse - Goinformation et gestion du territoire, 104(6): 314-320 (in German) Carosio, A., Dupraz, H. 1997. 'Evaluation, Accuracy and Reliability'. In: Swiss National Committee on Large Dams - Working Group on Dam Monitoring (ed.), The Geodetic and Photogrammetric Deformation Measurements of Dams, published on the occasion of the 19th ICOLD Congress, Florence, Italy, pp. 36-45, (Arbeitsgruppe Talsperrenbeobachtung, C/- Bundesamt fr Wasserwirtschaft, P. O. Box, CH-2501 Biel, Switzerland) CERN (European Organization for Nuclear Research). 1974. 'Le nouveau system de mires de rfrence SPS'. Internal CERN Document LAB.II/SU/8062/Int.74.5 (in French) Chrzanowski, A., Wilkins, R. 2006. 'Accuracy Evaluation of Geodetic Monitoring of Deformations in Large Open Pit Mines'. Proc., 12th FIG Symposium on Deformation Measurement, 22-24 May 2006, Baden, Austria, 10 pages (www.fig.net/commission6/baden_2006) Dedual, J. 2005. 'Web page on the History of Wild-Heerbrugg Ltd'. www.wild-heerbrugg.com, viewed on 18 July 2005 Deumlich, F. 1976. 'Bestimmung horizontaler Deformationen hydrotechnischer Bauwerke' (Determination of Horizontal Deformations of Hydrotechnical Structures). Vermessungstechnik, 24(5): 192 (in German) Duffy, M.A., Hill, C., Whitacker, C., Chrzanowski, A., Lutes, J., Bastin, G. 2001. 'An Automatic and Integrated Monitoring Program for Diamond Valley Lake in California', Proc., 10th FIG Int. Symp. on Deformation Measurements, 19-22 March 2001, Orange, California, USA, pp. K-1 to K- 23 (http://www.fig.net/figtree/com6_orange/) Egger, K., Walser, F. 2005. Sections on 'Geodetic Measurements' in Swiss National Committee on Dams (ed). Messanlagen zur Talsperrenberwachung Konzept, Zuverlssigkeit und Redundanz (Measuring System for the Monitoring of Dams Concept, Reliability and Redundancy) Published by the Working Group on Dam Monitoring, March 2005, 126 pages (available from www.swissdams.ch) (in German or French) Egger, K. 1997. 'Terrestrial Measurement of Deformation'. In: Swiss National Committee on Large Dams - Working Group on Dam Monitoring (ed.), The Geodetic and Photogrammetric Deformation Measurements of Dams, published on the occasion of the 19th ICOLD Congress, Florence, Italy, pp. 15-23, (Arbeitsgruppe Talsperrenbeobachtung, C/- Bundesamt fr Wasserwirtschaft, P. O. Box, CH-2501 Biel, Switzerland) Egger, K. 1993. 'Mesures terrestres de dformation'. In: Swiss National Committee on Large Dams - Working Group on Dam Monitoring (ed.), Mesure de dformations godsiques et photogrammetriques pour la surveillance des ouvrages de retenu, Wasser Energie Luft (eau nergie air), 85(9): 195-203 (in French and German) Egger, K., Keller, W. 1976. 'New Instruments and their Applications for Geodetic Deformation Measurements on Dams'. Transactions, 12th International Congress on Large Dams, Mexico, Reprinted by Kern & Co AG (now Leica-Geosystems), Aarau, Switzerland, 27 pages Favre, C., Flach, Ph. 1999. 'Stations totales motoriss - tat de la technique et perspectives d'avenir' (Motorised Total Stations State of the Technique and Future Trends). Mensuration Photogrammtrie - Gnie Rural, 97(3): 96-100 (in French) Flotron, A. 1997. 'Photogrammetric Measurement of Deformation'. In: Swiss National Committee on Large Dams - Working Group on Dam Monitoring (ed.), The Geodetic and Photogrammetric Deformation Measurements of Dams, published on the occasion of the 19th ICOLD Congress, Florence, Italy, pp. 31-36, (Arbeitsgruppe Talsperrenbeobachtung, C/- Bundesamt fr Wasserwirtschaft, P. O. Box, CH-2501 Biel, Switzerland) Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 33 Fryer, J.G., Bartlett, P. 1989. Photogrammetric Monitoring of Chichester Dam. In: Ogleby, C.L. (ed.), Proc., Symp. on Surveillance and Monitoring Surveys (SAMS'89), Melbourne, 9-10 Nov 1989, 74-85 Gilg, B. 1985. 'Mattmark', Section 4.3.3, In: Swiss National Committee on Large Dams (ed.), Swiss Dams - Monitoring and Maintenance. Published on the occasion of the 15th International Congress on Large Dams 1985, Lausanne, Switzerland, pp. 89-95 Gilg, B., Sinniger, R., Gavard, M. 1982. 'Long Term Measurements at Three Swiss Dams Mauvoisin, Grande Dixence and Mattmark', In: Transactions, 14th Int. Congress on Large Dams, 3-7 May 1982, Rio de Janeiro, Brazil, Vol. 1, Q.52, R. 54, 909-926 ISO. 1995. Guide to the Expression of Uncertainty in Measurement. Corrected and reprinted 1995, ISBN 92-67-10188-9, International Organization for Standardization, P.O. Box 56, CH-1211 Genve 20, Switzerland, viii+101 pages Kgi, R. 1978. 'Measuring Dam Deformations by Geodetic Techniques'. International Water Power & Dam Construction, April 1978: 45-48 Kern. 1988. 'Reference Standard (Mekometer) ME5000'. Written by H. Aeschlimann, Publ. No. 303 e, Kern & Co. Ltd. (now Leica- Geosystems), Aarau, Switzerland, 25 pages Kern. 1971. Przisions-Messausrstungen (Precision Measuring Equipment). Publ. No. 126 d, Kern & Co AG (now Leica-Geosystems), Aarau, Switzerland, 52 pages (in German) Keller, W. 1978. 'Geodetic Deformation Measurements on large Dams'. Edited and published by Kern & Co Ltd (Aarau) and Walter Schneider Consulting Engineers (Chur), Switzerland, 15 pages. Kinkel, B., Hlsmann, T., Lffler, C. 2001. 'berwachungsnetzmessungen mit dem Robot- Tachymeter Zeiss Elta S10' (Monitoring Network Measurements with the Robotic Tacheometer Zeiss Elta S10). Flchenmanagement und Bodenordnung (FuB), formerly: Vermessungswesen und Raumordnung (VR), 63(2): 86-93 (in German) Lang, W. 1929. Deformationsmessungen an Staumauern nach den Methoden der Geodsie (Deformation Measurements on Dams with Geodetic Methods), Verlag der Abteilung fr Landestopographie (L+T), Berne, Switzerland, 59 pages (in German) Leica. 2000. 'Accessories The Leica- Geosystems range of accessories for solving your individual survey task'. Publ. 710 883 en, Leica-Geosystems AG, Heerbrugg, Switzerland, 34 pages Leica. 1997. TM5100/TM5100A High Precision Theodolite for Industrial Metrology. Brochure U1-358-0EN, Leica-Geosystems AG, Unterentfelden, Switzerland Milev, G. 1985 Geodtische Methoden zur Untersuchung von Deformationen (Geodetic Methods for the Investigation of Deformations). Vol. 12, Vermessungswesen bei Konrad Wittwer, Konrad Wittwer Verlag, Stuttgart, Germany, xi+286 pages (in German) Pelzer, H. et al. (eds). 1987. 'Deformationsmessungen' (Deformation Measurements). In: Pelzer, H. (ed.), Ingenieurvermessung: Deformationsmessungen & Massenberechnung, Vol. 15, Vermessungswesen bei Konrad Wittwer, Verlag Konrad Wittwer, Stuttgart, Germany, pp. 37- 192 (in German) Reger, J.M. 2006. 75 Years of Change in Survey Technology, Survey Review, 38(300): 459-473 Reger, J.M. 2003. Electronic Surveying Instruments A Review of Principles, Problems and Procedures, Monograph No. 18, School of Surveying and SIS, University of NSW, UNSW Sydney NSW 2052, ix +155 pages Reger, J.M. 1997. Project Surveying. Unpublished lecture notes, School of Geomatic Engineering, University of New South Wales, x + 186 pages Reger, J.M. 1996. Electronic Distance Measurement An Introduction. 4th ed. Springer Verlag, Berlin-Heidelberg-New York, xix + 276 pages Reger, J.M. 1995. 'Control Networks for Dam Monitoring Measurement Options'. Proc., 3rd Symposium on Surveillance and Monitoring Surveys (SAMS'95), University of Melbourne, 1-2 November 1995, pp. 131-143 Overview of Geodetic Deformation Measurements of Dams ANCOLD 2006 Conference Page 34 Reger, J.M. 1994. 'Monitoring with Surveying Robots', International Journal for Surveying, Mapping and Applied GIS (Geodetical Info Magazine), 8(2): 58-63 Reger, J.M., Sippel, K.D. 1994. 'Practical Use of Surveying Robots with EDM Beam Pointing'. Proc., Commission 5, 20th FIG Congress, 5-12 March 1994, Melbourne, Paper TS509.2, 13 pages Schneider, D., Wiget, A. 1997. 'Deformation Measurement by Satellites'. In: Swiss National Committee on Large Dams - Working Group on Dam Monitoring (ed.), The Geodetic and Photogrammetric Deformation Measurements of Dams, published on the occasion of the 19th ICOLD Congress, Florence, Italy, pp. 23-30, (Arbeitsgruppe Talsperrenbeobachtung, C/- Bundesamt fr Wasserwirtschaft, P. O. Box, CH-2501 Biel, Switzerland) Schulz, Th., Zogg, H.M. 2006. 'Terrestrisches Laserscanning in der Geomatik' (Terrestrial Laserscanning in Geomatics), Gomatique Suisse - Goinformation et Gestion du Territoire, 104(8): 420-424 (in German) Swiss National Committee on Dams. 2005. Messanlagen zur Talsperrenberwachung Konzept, Zuverlssigkeit und Redundanz (Measuring System for the Monitoring of Dams Concept, Reliability and Redundancy) Published by the Working Group on Dam Monitoring, March 2005, 126 pages (available from www.swissdams.ch) (in German and French) Swiss National Committee on Large Dams. 1997. Working Group on Dam Monitoring (ed.), The Geodetic and Photogrammetric Deformation Measurements of Dams, published on the occasion of the 19th ICOLD Congress, Florence, Italy, pp. 1 - 62 (Published by: Arbeitsgruppe Talsperrenbeobachtung, C/- Bundesamt fr Wasserwirtschaft, P. O. Box, CH-2501 Biel, Switzerland) Swiss National Committee on Large Dams. 1993. Working Group on Dam Monitoring (ed.), Mesure de dformations godsiques et photogrammetriques pour la surveillance des ouvrages de retenu, Wasser Energie Luft (eau nergie air), 85(9): 181-260 (in French and German) Swiss National Committee on Large Dams. 1985. Swiss Dams Monitoring and Maintenance. Published on the occasion of the 15th International Congress on Large Dams, Lausanne, Switzerland, 260 pages. Untersee, V. 1975. 'Investigation of Spatial Deformations in Power Dams by the Geodetic Method'. (Translated by A.T.W.) Publ. No. Th 133e, Wild Heerbrugg Ltd, (now Leica -Geosystems), Heerbrugg, Switzerland, 14 pages. Untersee, V. 1951. 'Investigation of Spatial Deformations in Power Dams by the Geodetic Method'. Original publication (1 March 1951) by Geodetic Section, Swiss Federal Topographic Survey, Wabern near Berne, Switzerland (in German) Welsch, W., Heunecke, O., Kuhlmann, H. 2000. Auswertung geodtischer berwachungsmessungen (Analysis of Geodetic Monitoring Measurements), Herbert Wichmann Verlag, Heidelberg, Germany, 507 pages (in German) Wild Heerbrugg. 1986. 'Theodolite Accessories'. Publ. No. G1 279 e, Wild Heerbrugg Ltd, Switzerland, 18 pages Wilkins, R., Bastin, G., Chrzanowski, A. 2002. 'A Precise, Reliable and Fully Automatic Real Time Monitoring System for Steep Embankments', Paper presented at CIM 2002, 28 April - 1 May 2002, Vancouver, Canada, 8 pages (http://aca-ltd.com/publications.html)