The Use of GPR in Road Rehabilitation Projects

Transcription

1 The Use of GPR in Road Rehabilitation Projects Mara Nord Project in cooperation with: European Commission, Finnish Transport Agency, Swedish Transport Administration, Norwegian Road Administration, ELY center, Lapin Liitto, North Calotte Council, Rovaniemi University of Applied Sciences, Oulu University of Applied Sciences, Roadscanners, Road Consulting, Carement, Ramböll Sweden, Malå Geoscience, SwedIsh National Road and Transport Research Institute, SINTEF, 3D Radar, NCC Roads

2 The Use of GPR in Road Rehabilitation Projects Contents 1. 0BIntroduction Ground Penetrating Radar (GPR) Technology B2.1 General B2.2 Electrical properties affecting GPR wave propagation Dielectric value B2.2.2 Electrical conductivity Principles of different GPR systems B Impulse radar B2.3.2 Stepped frequency radar principles B2.4 Principles of GPR image B2.4.1 Reflection and polarity B2.4.2 Depth penetration, resolution and interface depth B2.5 Ground Penetrating Radar in road surveys B3. Survey Equipment B3.1 General B3.2 Ground penetrating radar quality requirements B4. Survey Planning and Performance B4.1 Measurement timing, planning and information B4.2 Implementation of measurements B4.3 Digital image capture during GPR road surveys B4.3.1 In general... 21

3 43B4.3.2 Digital video or still image capture equipment and data collection B4.3.3 Video data processing and linking to software projects B4.4 GPR surveys and reference coring B5. GPR Data Processing and Interpretation B5.1 In general B5.2 Preprocessing B5.3 Air coupled antenna data processing and interpretation B5.4 The ground coupled antenna data processing and interpretation B MHz ground coupled antenna data B MHz ground coupled antenna data B6. Reporting and Delivery of Results B6.1 General B B6.2 Printing to an image file B BLITERATURE... 33

4 PREFACE Ground Penetrating Radar (GPR) is a non-destructive ground survey method that can be used in assessing roads, railways, bridges, airports, tunnels and environmental objects. Its main advantage is the continuous profile it provides over the road structure and subgrade soil. GPR technique is becoming an increasingly important tool especially for planning structural evaluation of roads. Another important advantage of GPR in road surveys is that it does not essentially disrupt the flow of traffic during the survey. The Nordic countries have reached good quality of skills and accumulated the knowhow of GPR applications on roads over the last 15 years. However Finland, Sweden and Norway have slightly different practices in the use of GPR and that is why there is a need for common procedures now when companies are more and more operating across the borders. Also the level of knowledge, awareness and experience regarding the use of GPR in the Road Administrations vary in all three countries and there is a need to share the knowledge and develop procedures to ensure better quality of GPR services. To respond to these recognized needs Mara Nord, international cooperation project financed by Interreg IV A Nord, has been initiated among Finland, Sweden and Norway. In this project one goal was to produce common guidelines that can be used as a reference in procurement processes in all three countries. Johan Ullberg from Swedish Transport Administration has been in charge of the guidelines. Mara Nord project lead partner has been Rovaniemi University of Applied Sciences and experts participating have been Anita Narbro and Janne Poikajärvi. Other key experts have been Katri Eskola from Finnish Transport Agency, Kalevi Luiro from ELY-centre of Lapland, Finland, Per Otto Aursand and Leif Bakløkk from Norwegian Public Roads Administration, Rauno Turunen from Oulu University of Applied Sciences, Finland. This recommendations for the guidelines has been written by Timo Saarenketo and Pekka Maijala from Roadscanners Oy, Finland. Rovaniemi April 25th 2011 Timo Saarenketo Pekka Maijala

5 1. 0BIntroduction Nordic countries have been pioneers in ground penetrating radar (GPR) applications on roads. The first applications have been reported already in early 1980 s from Denmark and Sweden. In Finland subgrade soil and road structures have been analyzed with GPR already for almost thirty years and the method is a routine tool in Finnish Traffic Agency s rehabilitation projects. This GPR method description guideline for road rehabilitation projects has been made due to new procurement policies in Nordic road administrations which means that road condition surveys will be ordered through open competition. That is why it is essential that current practice is described precisely to make it also possible for new entrepreneurs to know how and at which level the ground penetrating radar surveys should be done and what is the default quality level for the results. There is also a need to specify separately the procedure that should be followed in surveys projects made in 2 dimensional (2D) and 3 dimensional (3D) GPR projects. This method description guideline should be used when doing data collection with the 2D and 3D GPR systems prior to rehabilitation planning. It should also be used when modifying output data of the interpretation and problem diagnostics on a compatible format that can be read with GPR data viewer software packages. A description of the testing arrangements and antennas accreditation is given as well. In this publication, there is also an exact description of how ground penetrating radar data collection should be done, how the collected data should be processed and interpreted and how the outcome should be linked to road registry data bases or other GIS systems used in Nordic Road Administrations. Pertaining to the parts and methods not described in this publication, the other national publications and guidelines related to road rehabilitation projects must be followed. Digital video or digital images collected during the GPR data collection have proven to be essential during GPR data analysis and interpretation and in quality assurance of GPR projects. In order to emphasize the importance of this the description how to collect the images or video(s) and the required quality of these are described in this publication.

6 This publication presents also a short introduction to GPR theory and principles. Guidelines are written to meet the general guidelines and practices in Nordic region but this document can be also used elsewhere in the world. The use and guidelines of GPR technology in other road and bridge applications is described in other documents published by MaraNord project. 2. Ground Penetrating Radar (GPR) Technology 6B2.1 General Ground penetrating radar method is based on the use of radiofrequency electromagnetic (EM-) waves. Used frequency range is MHz. Inside of this frequency range, it is said that EM- waves can propagate in a low electrical conductivity medium. Physical parameters affecting the wave are the medium s conductivity, dielectricity and magnetic susceptibility. In Nordic countries, magnetic susceptibility does not have a significant effect on ground penetrating radar signal s behavior. 7B2.2 Electrical properties affecting GPR wave propagation Dielectric value Dielectric value describes substance s ability to charge or polarize from influence of an electric field. After the electric field s effect ends, the substance returns to its initial state. If the material s structure is such that its initial state does not return completely, its polarization is partly lossed. In such cases, dielectric value can be considered as a complex quantity, where the real part describes reversible polarization and loss is in imaginary part. The most important element in molecular polarization in road materials and subgrade soil is the water molecule. The extent of dielectric value depends most on how much free water there is in the material because the dielectric value of water is more than 10 times higher than the dielectric value of other road materials. Therefore increasing water content increases the dielectric value of the material. 8B2.2.2 Electrical conductivity A medium s electrical conductivity describes the ability of free charges to move in the medium. External electric field s move charges from place to place. The more free charges, ions and electrons there are, the higher the conductivity of material and ground penetrating radar attenuation.

7 2.3 Principles of different GPR systems 9B GPR hardware systems used in road can be divided into two categories; impulse radars and stepped frequency radars. The principles of these system are described in the following Impulse radar Impulse radar is the most used ground penetrating radar type. The working principle is described in the following. A pulse, generated in a transmitter antenna, is sent into the medium. The length of the pulse is from under a nanosecond up to tens of nanoseconds, depending on the frequency. When propagating in road structures and subgrade soil a part of the pulse energy is reflected from surfaces of different electrical properties; some of them are propagating through the interface and are reflected back from the following interfaces. How the signal attenuates is a result of geometric attenuation, signal scattering, reflections and thermal losses. GPR system records the two way travel time and amplitudes of signal reflections as a function of travel time is presented. When measurements are made rapidly in sequential survey points, it can be compiled as a GPR profile (radar image). In the Nordic ground penetrating radar projects, the gray scale tone and a reflector polarity in the printouts should be like in Figure 1. Figure 1. Ground penetrating radar profile, which is measured with the air coupled antenna and its individual pulse. The profile has reflections from interfaces of two mediums with different dielectric properties (ε). Image structure s layer 1 describes the pavement, layer 2 describes base course, layer 3 sub-base and layer 4 filter course. The figure shows that dielectric value of materials (moisture) grows when going down from the road surface with an

8 exception that dielectric value of filter course (ε4), is less than in the sub-base and its reflection has inverted polarity (black line in the middle of two white lines). 01B2.3.2 Stepped frequency radar principles In a stepped-frequency radar the radar waveform consists of a series of sine waves with stepwise increasing frequency. The radar measures the phase and amplitude of the reflected signal on each frequency and used an inverse Fourier Transform of these data to build a time domain profile. Thus, the step-frequency radar collects data in the frequency domain and converts the data to time-domain data through computer processing. The resulting radargram is similar to impulse radar data, and they can be processed and interpreted the same way as impulse rada data. However, since stepped-frequency data are collected in the frequency domain, they allow advanced filtering and signal processing to be applied directly to the raw frequency domain data. 1B2.4 Principles of GPR image 1B2.4.1 Reflection and polarity When GPR signal propagates from medium 1 to medium 2 and medium dielectric values are E 1 and E 2, the reflection coefficient will be: R On the basis of the formula, polarity of the reflection changes if E 1 is smaller than E 2, which is usually the default situation in road and soil structures (moisture content is getting higher when getting deeper). If E 1 is bigger than E 2, then the polarity of the reflected wave remains the same as the progressive wave s polarity at the interface. In road radar measurements, however, it is common practice that the surface reflection is recorded as positive, even though the reflection coefficient is negative. Similarly, the other layers, where ε (upper) < ε (lower), are recorded as positive reflections. In the gray scale they should be presented in such a way that the white reflection is in the middle (see Figure 1. ε 1 < ε 2 ). Correspondingly, if the dielectric value of the lower layer is smaller than the upper, for instance in the structure of Figure 1 ε3 > ε4, the reflection is so-called negative and then the black reflection is in the middle. GPR signal polarity, leaving the antenna and progressing in the medium, can be changed 180 degrees easily by

9 changing the positions of the transmitter- and receiver antenna or when doing GPR data post processing by multiplying the signal with a factor B2.4.2 Depth penetration, resolution and interface depth Achievable depth penetration with ground penetrating radar depends on what antenna frequency is used and therefore the wave length of signal. The attenuation increases when GPR central frequency increases. A high conductivity of medium results in an increase in the amount of energy scattering objects, when the length of wave gets shorter. Similarly, the penetration depth gets smaller while the frequency gets higher. On the other hand the resolution gets better at the same time. The resolution gets better also when dielectric value increases. Resolution refers to how close interfaces can be from each other, when they can still be identified as separate interfaces. This applies to both directions, horizontally and vertically. The vertical dimension s resolution of the pulse can be calculated from the following formula: h c 2 r, where c = the speed of light in vacuum (0.3 m/ns) = the pulse length (ns) ε r = medium s relative dielectricity The depth to an observed interface can be calculated from this formula (for monostatic antennas): 0.5 twt c s v t r, where twt = two way travel time of the wave. 31B2.5 Ground Penetrating Radar in road surveys In Nordic countries there are at the moment tens of people employed or involved with GPR, working for universities, consulting companies and for contractors. The use of ground penetrating radar has spread from subgrade investigations and structural layer and thickness measurements to the structural layers

10 and subgrade quality investigations. One of the biggest advantages of 2D ground penetrating radar surveys is its ability to produce continuous line data from the investigated target. And when using 3D GPR system the changes in road structures and their properties can be measured also in road cross section. During last few years ground penetrating radar technology has been applied to other types of research as well. Pavement quality assurance and structure thickness quality assurance surveys have been applications that are coming increasingly popular. Bridge deck condition assessment and monitoring has developed rapidly in recent years. 2B3. Survey Equipment 41B3.1 General Most of GPR equipment used in road investigations are with pulse radar principles as previously described. Equipment for road investigations consists of several components (Figure 2). Antennas consist of a transmitter, which transmits the pulse to the medium and a receiver that receives the reflected signals. The antennas are controlled by a control unit, where the wavelength and strength of pulse are determined. In digitizers the received pulses are changed into digital format. Digitizer can be located in the antenna or in control or central processing unit. The data collection setup is controlled with a data collection software. Some setup parameters include scans per time or distance unit (for instance scan/sec, scan/m), measuring time window (ns), the number of the samples per scan ( for instance 512, 1024 samples/scan) and the data format (for instance 8, 16, 32 bit). Calibrated optical encoders (distance measuring instruments) are used to trigger the control unit as the systems moves over a distance. Nowadays GPS is also an essential part of GPR data collection and this also applies with digital video and still image capture process. A power supply is of course necessary for operation of the equipment.

11 UFigure 2.U Typical instrument configuration of the ground penetrating radar van, which is used in road surveys. Ground penetrating radar antennas can be divided roughly into two categories: air coupled antennas and ground coupled antennas (Figure 2). These in turn can be either monostatic, when the same antenna is a transmitter and a receiver, or bistatic, when transmitter and receiver are different antennas units. Most road survey antennas are bistatic, but the antenna elements are built in the same box. The frequency of ground coupled antennas in road rehabilitation surveys vary from 50 MHz to 2500 MHz. Their advantage compared to air launched antennas is better depth penetration, though ringing (caused mainly by strong coupling ), may cause interferences. The ground coupled antennas have better resolution of individual objects than the air launched antennas. A summary of different antenna qualities is presented in Table 1.

12 UTable 1U.Different ground penetrating radar antenna qualities in road rehabilitation surveys. The table depicts general characteristics and it must be remembered that depth penetration and resolution depends on electrical properties of a road structure and sugrade and it should also be noted that the resolution of wide band antennas close to the surface is significantly higher than the antenna center frequency would suggest. A n t e n n a C e n t r a l D e p t h - Resolution A p p l i c a t i o n s Frequency penetration (m) (mm) (MHz) (ε r 6) Ground coupled antennas 1 high frequency ,4 1, pavement, base course, sub base, steel net 2 medium frequency ,5 4, base course, sub base, overall thickness of structure layers, low embankments, subgrade soil < 3m 3 Low frequency overall structure thickness, embankment, subgrade soil < 20 m (excl. clay and silt) Air coupled antennas 1 high frequency ,4 0, gravel road wearing course., pavement, steel net, base course, subbase 2 medium frequency ,5 1, pavement, base course, sub base, steel net 3 low frequency ,0 3, base course, sub base, overall (not yet in use in structure thickness, subgrade soil Finland) < 3m

13 Broadband aircoupled antennas (for steppedfrequency radar) ,0 3, gravel road wearing course., pavement, steel net, base course, subbase, overall structure thickness, subgrade soil < 3m Air coupled antennas are mainly horn antennas, but lately also other antenna types have been developed. Frequencies vary from 400 MHz to 2.5 GHz, but the most commonly used central frequency is 1.0 GHz. Also GHz air coupled antennas are also available in the market. These antennas have proven to be effective, for instance, in detecting individual layers in bitumen bound layers, in gravel road wearing course thickness surveys and in locating steel nets. Normally the depth penetration of air coupled antennas is m and that is why they are mainly used in pavement structure and in bridge surveys. During the measurements, air coupled antennas are typically m above the pavement surface. Since pulse radar air coupled antennas are suspended in the air, the antenna s coupling does not change when electrical properties of the road surface change. That is why these antennas can also be used in repetitive measurements, where the ground penetrating radar data quality should not change if and when the properties of the road surface are changing. This makes possible to calculate quality parameters on the basis of reflection amplitude and frequency response. In addition, air coupled antenna data collection can be done using survey speed up to km/h and thus without distracting the other traffic. At the moment there are several device manufacturers, which produce GPR equipment suitable for road surveys. The biggest ones are Geophysical Survey Systems, Inc. (GSSI), IDS (Italy), Mala Geoscience (Sweden), 3d-Radar A/S (Norway), Sensors&Software (Canada), Utsi Electronics (U.K), Penetradar (USA), Radarteam Sweden Ab (Sweden), Geoscanners AB (Sweden). In addition, there are also several smaller companies involved in the business. 1 Frequency range for stepped frequency radar is specified as minimum and maximum edge frequencies instead of centre frequencies. This means that a MHz antenna spans approximately the same frequency range as a combination of a 400 MHz, 900 MHz, and 1500 MHz antenna.

14 51B3.2 Ground penetrating radar quality requirements In Europe all Ground Penetrating Radar hardware systems sold have to be approved by ETSI and their use requires further ECC approval. At the moment in Finland all GPR units have to be registered and licensed in Ficora before they are allowed to be used. Ficora licensing provides also some extra instructions and restrictions for the use of GPR in Finland. For instance GPR survey log book has to be kept where each survey time, place and survey equipment info is documented. Similar GPR registration and licensing system will be most likely applied also in Sweden and in Norway within the next few years. In addition, when doing GPR measurements for the Finnish Transport Agency pavement quality control surveys, only tested and approved equipment can be used. If the equipment is not annually approved, it cannot be used. Measurements performed with improperly working equipment must be repeated with the properly approved equipment. The annual quality tests for air coupled antennas (1.0 GHz) are described by Scullion et al. (1996). These tests must be passed every year. Tests are done in approved laboratories, which are impartial to parties involved. The Finnish Road Agency publishes a list of the approved laboratories every year. Measurement licenses are issued to companies with approved equipment. After ground penetrating radar tests, the antennas are classified in categories A and B and the requirements are presented in Table 2. If antennas are in category A, they can be used as is. If it is in category B, they can only be used in specified surveys and with specified settings mentioned in the license. When surveying category B equipment, the complete straight pulse amplitude should be recorded, making it possible to track interferences and noise level. In these tests, time linearity needs to be done only once for each antenna, all other tests must be performed annually.

15 UTable 2.U Mandatory Ground Penetrating Radar system quality requirements in the Finnish Road Agency asphalt quality control surveys. Tests Classification A. Classification B. No restrictions in FINRA projects Can be used following specific rules and conditions signal/noise -ratio 5 % 10 % short time amplitude stability 1 % 3 % long time amplitude stability 3 % 6 % long time stability 5 % 10 % linearity of time 5 % 7,5 % Ground coupled antennas will also have their own quality standards in the future. 3B4. Survey Planning and Performance 61B4.1 Measurement timing, planning and information Surveys with ground coupled antenna systems can typically be made any time of the year, although there are some restrictions. In areas where there is frost, measurements should be made either before or after the frost has melted from the subgrade soil, but preferably not during thawing periods. It is not recommended that gravel surface roads are measured in the summer without client s permission, because salt used to reduce dusting attenuate the signal. Winter surveys provide additional information about the frost depth. If the client demands, the surveys can be done in the summer, but then the client is also responsible if data quality is poor. When using air coupled antennas in the winter, measurements should be when the frost front is deeper than 0, 8 m from the road surface. On paved roads caution should be taken with winter measurements as when deicing salt is causing signal attenuation.

16 In pavement and pavement structure surveys it is recommended to use air coupled antennas in the summer time especially when the interest is in bituminous pavement thickness and its quality, unbound base course and its quality, wearing course of the gravel roads and its quality. In this case the dielectric value can be used to determine the water sensitivity of the wearing course. Before the surveys, a revised plan should be sent to the client s contact person, describing when and how the measurements are intended to be done (longitudinal sections and cross sections), measurement equipment and personnel. The note can be sent via mail, , or fax or if this is not possible, then taken care of over the phone. Safety considerations for the ground penetrating radar measurements are described in a road works safety plan required by each Nordic road agency. But in general survey crews have to have accepted national road work safety course documents and also each country require traffic safety plan and advance notice for GPR works especially on high volume roads. Nordic government authorities may require also radio license for the use of each GPR equipment in the country. These national guidelines may define also special geographical areas, such as airports, astronomical survey stations, hospitals, prisons or defense force area, where the use of GPR is prohibited or a special license is required. GPR equipment, if working properly, will not cause interferences but broken equipment can possibly cause false alarms. 71B4.2 Implementation of measurements The ground penetrating radar hardware must be safely and securely mounted to the survey vehicle and the measurements should be conducted in a way not to endanger the workers, road users or anyone in close vicinity of the road. The survey vehicle must be equipped with nationally accepted warning lights and safety signs. To ensure GPR data quality and for traffic safety reasons there should be always two persons in the survey car, one focusing on driving and the other running and controlling the measurement devices. The driver can make comments from the road s condition, if necessary. The measurement staff should have completed necessary local road safety courses (Finland: Tieturva I, Sweden: Arbete på Väg, Norway: Kurs i arbeidsvarsling).

17 2D GPR surveys can be made with a multi-channel system where it is possible to measure with air coupled and ground coupled antennas at the same time. In this case, the data can be collected for example with a GHz air coupled antenna and a MHz ground coupled antenna. Ground penetrating radar antennas can be sequentially or side by side. The basic principle is that the measurement is done in outer wheelpath in the main direction of the road. If the antennas are side by side, the air coupled antenna should travel on the right wheel path and ground coupled antenna in the middle of the lane. If the measurements are done with two separate runs but with the exact same start and end point, it must be ensured that the data later is scaled so that the lines are exactly the same length in case there are differences in the driving path. Different measurement profiles can be matched with each other using culverts and/or bridges as a marker. The maximum longitudinal difference is 2.5 m presented in the data with individual objects. On special occasions, such as roads with geotechnical problems, the client can order separate GPR surveys with MHz antennas. Usually there is need for better depth penetration, for example to map the bottom of a peat layer. Pavement surveys made with the air coupled antennas gather information from the pavement s and base course s dielectric values. When using air coupled antenna, antenna bouncing (distance from pavement surface) can provide some information about the roughness of the road. Ground coupled antennas should be used to for example, in road cross section measurements. To minimize the changes in coupling of the ground coupled antenna, the distance from the road surface should not change. With MHz antennas, distance from the road surface may not exceed 8 cm and with 1.0 GHz or higher frequency antenna the distance may not exceed 5 mm. In practice this means that the antenna should dragged along the pavement surface. For scan interval, a very accurate and calibrated pulse encoder should be used. Time-based sampling should not be accepted, except in cross section measurements, where lines have to be painted on the road and data has to be matched to them using markers. In final result delivery, no time-based data is allowed; all data has to be distance based.

18 The amount of survey longitudinal survey lines depends on the information needed from the road. Table 3 presents different types of GPR surveys on road that can be used in different kinds of projects. Table 3. Different kinds of 2D and 3D surveys on roads and their characteristics. *) depends on the frequency The sampling interval in GPR surveys should be at least 10 scan / m in longitudinal survey profiles and at least 40 scans / m in cross section profiles. All the measurement data should be recorded in minimum 16 bit format and the smallest sampling density is 512 samples / scan. One-point gain (so called flat gain) must be used with the air coupled antenna. Ground coupled antennas allow smooth multipoint gaining, avoiding any visible differences in time window. Signal clipping is not allowed in any data. Filtering can be done according to the manufacturer s recommendation. These values should be the same as used in antenna tests. Horizontal filtering should be avoided because it has a tendency to remove also reflection information of horizontal structures from the road. When defining time windows for ground penetrating radar surveys, the following principles should be used: Antennas ( GHz): The time window should be at least 20 ns and the time window below surface at least 12ns. Correspondingly antennas greater than 2.0 GHz, the measuring time window should be at least 10 ns. Antennas ( MHz): time window below the surface must be at least 60ns.

19 Antennas ( MHz): the measuring time should be agreed with the client case by case. Here the expertise of GPR provider must be used to estimate the depth penetration of the antenna and the depth of the wanted information. Table 3 shows the required time windows (two way travel time) for different target depths and different soil types (different E r). UTable 3.U The required time windows for different soil types and different target depths. target depth twt (ns) sand, gravel; E r = 5 twt (ns) moraine; E r = 9 twt (ns) peat; E r = 40 3 m m m m Road number, road section or distance, direction and the survey personnel should be marked into the header file. In addition, there should be a logbook in the survey vehicle, where time of the survey, survey equipment, weather, possible problems and other possible notes are documented. In addition in Finland Ficora requires that the following survey information is documented: the location, date, time, unit serial number, antenna types and serial numbers used in each survey. In practice each central unit has a logbook, where all the surveys are marked with required information. Before starting a survey, the correct location must be ensured. Usually the surveys start and end in road registry s start and end points or nodes, even though the interpretations can be made from a shorter section. The GPR files should be cut from the key points such as road data base knot points or road section breaks. If the survey does not follow the road registry, the start and end point should be marked with paint on the road, unless they are simple and unmistakable locations such as culverts or bridge joints. When surveying new roads and pavements it is sometimes easier to follow project distance and chainage. The GPR operator can add markers in locations of intersections, bridges, culverts, etc. These markers can be used to link the GPR file and video, unless linking has been done already during the survey.

20 Markers are also used, when two or more lines or lanes are surveyed and there is no GPS from all survey lines. Surveys should not be done when it is raining or when the road is wet. Survey speed should be constant and stopping minimized. When using air coupled antennas, after every survey session and before the GPR unit is switched off, a metal pulse should be taken. It is recommended that metal reflection is done also before the start of the measurement. In addition, if a height correction file is not available, a bouncing file should be recorded. Figure 3 presents basic calibration tests that should be done when GPR data is collected. When measuring with air coupled antenna, so-called direct pulse (from transmitter antenna to receiver antenna) must be completely visible all the time. Correspondingly when measuring with the ground coupled antenna, the surface pulse must always be visible. With ground coupled antenna a lift test files or similar other test can be recorded to locate the pavement surface in the scan. Figure 3. GPR antenna calibration tests before or after data collection: photo A. presents the lifting test for a ground coupled antenna, photo B. the metal plate test for a horn antenna and photo C. the antenna bouncing test for a horn antenna.

21 81B4.3 Digital image capture during GPR road surveys 91B4.3.1 In general When the road authorities make a purchase order for GPR surveys, simultaneous digital video or still images is strongly recommended to be required because it has proven to improve the quality of GPR interpretations and avoid mistakes. If digital video is collected it should have an audio track, where oral commentary of the road condition done by surveyor is recorded. Digital video or digital images help road engineers, designers and contractors to get a better understanding of the road condition and its surroundings and based on this data road surface conditions and the condition of the road assets can be documented reliably. 43B4.3.2 Digital video or still image capture equipment and data collection The digital image(s) from the road can either be performed with a video or high class digital camera. The mionimum requirements when using video camera should be at least 140*480 resolution, 12 fps and progressive scan must be used. When collecting still images the maximum distance between images should be 10 meters and minimum resolution 1024 *768 pixels. The camera optics should be of high quality and the lens should be wide-angle. The lens should be clean. Focusing should be made so that the area of interest is as precise as possible. Autofocus systems should not be used. The camera should be mounted horizontally and firmly to minimize shaking. It should be on the roof of the survey vehicle, or as high as possible. Recording through car window should be avoided. The camera angle should be such that on flat, open road section, the image is mostly road surface the sky can be seen only as a thin stripe in the top of the image. It is preferable that part of the antennas, would be visible in the bottom of the picture, making exact location analysis possible. The camera angle depends whether the measurement is ordered from one or both lanes. If images are recorded from only one lane, the camera should be mounted in the left (centerline) side of the car and the camera angle adjusted to show the whole road and surroundings widely. If the measurements are made from both lanes, the camera should be in the outer side, so that the surveyed lane is visible, but also the grass and soil verges and ditches.

22 UFigure 4.U Example of the two video camera setup. Camera A is describing a situation, when there is only one video from the road. B and C are describing situations where there are videos taken in both lanes. The image capture should always be done in good light, so that details can be seen as well as possible. If the ground penetrating radar measurements are made at night, for example, because of the traffic, the video can be taken separately in the daytime. The survey speed should be such that the picture quality is accurate, the recommended speed on an average is km/h, where there is also enough time to make reliable oral observations. In good conditions slightly higher speeds can be used remembering the limitations of observation times. During the recording, survey personnel should make comments about the roads condition and the surroundings. The comments should be focused on the following subjects: Is the road in a cut, embankment, side sloping ground or 0 level Drainage conditions Observed bedrock Visible damages in the road surface, above all shoulder deformation, which is sometimes hard to see from the video. Estimates of the soil type

23 Other road design factors The principle is that the commentary is continuous, but observations should be done at least every 200 meters. In the beginning of each recording, a short summary of the road, road section and for what project must be announced. The image capture should always be attached to a location (GPS coordinates, chainage), which allows the data to be linked to software projects and other data. 02B4.3.3 Video data processing and linking to software projects The collected video will be linked to software project. In addition, if required, a subtitle including road number, section and corrected road register address can be added. Video should be compatible with Microsoft Windows and the data compression system must be such that the video compatible with common media player software (for example Media Player 10). Linking the video should be in a way that the clear GPR reflections match the video view and the difference must not be more than 1 meter. Linking should match also with GPS coordinates. Preferred linking method is that the video s frame number is linked into radar data s scan number, but also other sufficiently precise linking methods are allowed. The principles for video linking should be presented in the quote. 12B4.4 GPR surveys and reference coring In most Nordic GPR survey projects on roads, it is required that reference samples are taken to support and confirm the interpretations. Sampling can be used also to define properties of pavement structure layers in laboratory analysis. There are three basic types of reference coring: A. Pavement and base course B. Pavement, bound and unbound base course C. Pavement, bound and unbound base course, other structure layers and their total thickness and the sub grade soil quality.

24 Normally reference coring should be made at 2-3 km intervals, with a minimum 1 / 10 km road section. More samples can be taken if accepted by the client. Representative sampling points are decided after preliminary analysis of GPR data. The preferred sampling point is the outer wheelpath, but exceptions can be made, if accepted by the client, especially on high traffic roads. The exact reference core location must be described in sampling documents. From the reference cores, the thickness of pavement layers and bound layers are determined with an accuracy of 4-5 mm, thickness of the unbound layers with 2 cm accuracy and thickness of other pavement structure layers with 5 cm accuracy. If the sample is needed for laboratory analysis, it is written in the quote. However, it is recommended, that at least the drill core samples from bound layers are photographed and the photographs delivered to the client. 4B5. GPR Data Processing and Interpretation 2B5.1 In general The major goal for the GPR surveys is to provide information for rehabilitation and pavement design. In this case the most important thing is to produce precise information about a) the total thickness of the pavement b) the total thickness of the base course, c) total thickness of the unbound layers, d) thickness of the embankment or depth to the bottom of the embankment, e) the quality of subgrade soil and f) road assets, such as culverts. In addition, the interpreter can produce other information needed in the rehabilitation design as much as possible, such as reasons for damages and distress, other road furniture, etc. If the data collection has been carried out properly with appropriate equipment, the information described above can be produced with careful and professional processing and interpretation. The following chapters describe the general guidelines for data processing and interpretation. 32B5.2 Preprocessing In this case, preprocessing means GPR data editing and combing other road testing data, acquired in different ways. Data editing includes operations which do not change the original information content of

25 the data. This means distance scaling, joining and splitting of different lines and reversing directions (if needed). If the surveys are ordered from the both lanes, they must be arranged in same direction with the lines matching each other (for instance the culvert is in coinciding locations). If the lines do not match, they must be scaled to match with markers and culverts. Linking the distance coordinates means that the survey has a starting point and all the data is scaled to this starting point with equal scale. This is suitable for situations, where surface coordinates are not important. Even in these cases, it is useful if the altitude information is available. It is strongly recommended to scale the GPR surveys to the road register length. If there is a big difference (>15m / 5km) between the announced registry length and the length of ground penetrating radar survey, it should be discussed what is the possible reason for this and if the survey lines must be scaled to an fictive length. If the coordinates at certain points on a line are known, these points can be linked to the data. Commonly the coordinates are collected simultaneously during the surveys, using real time GPS systems. Data processing software combines all the information. 42B5.3 Air coupled antenna data processing and interpretation Data processing of air coupled antenna is described in Figure 4. More detailed process is described for instance in GPR text books and software manuals made by different GPR manufacturers. In the first stage, air pulse (scan 2) should be removed from raw data (scan 1), if there abnormal ringing below the zero level (P-level). (Important notice: According to ECC/ETSI guidelines, when measuring air pulse, antenna must not be pointed up to the sky) If the system and the antenna is high quality, the pulse is practically straight after the surface level, it is not necessary to remove the air pulse and the processing is made only using metal reflection data (scan 3). After the metal reflection removal, so-called surface leveling (elimination of antenna bouncing) should be done and at the same time the dielectric value of the first layer is calculated. At this stage, in many GPR software, the surface resolution is improved by using different techniques. The use of these techniques is allowed, even desirable, because the pavements in Nordic countries are mainly thin.

26 In the following, the basic principles for air coupled antenna data interpretation are shown. Instructions apply to all other surveys except pavement quality (void) measurements. UFigure 5.U The pulses used in the processing of air coupled antenna. S is so-called direct pulse and it is used as a reference level. P is the reflection from the surface. Pulse 2 is so-called air pulse, which describes the reflections caused by the antenna elements themselves. Pulse 3 is a metal reflection pulse, where the air pulse (background, pulse 2) has been already removed. Pulse 4 is the resulting pulse used in the final interpretation. Pulse 4 is obtained by reducing the pulse 1 and part of the pulse 2 from the pulse 3. Removing the pulse 3 will improve the surface resolution. The maximum amplitude technique should be used in all interpretations, observing also the polarization variations. In the pavement antenna interpretation, the minimum interpretation (basic package) is as follows the total thickness of bound layers (pavement or pavement + bound base) (code 21) total thickness of unbound base course (code 41) If the layer interfaces are weak and the interpretation is uncertain, the interpreter must use interface quality codes published in Ground Penetrating Radar (1992), a booklet by Geotechnical Society of Finland. These interface codes are:

27 a) a clear, strong and reliable interface (solid line) b) unclear, but probable interface (a line longer than the interval) c) unclear interface (interval longer than line) If the interpreter cannot interpret the required interfaces, a separate report should be written to the client clarifying the reasons of not succeeding in the task. For instance frost fatique is not a reason to leave layer interfaces to be left uninterpreted. In unclear situations, the client can commission a comparative survey of the work. Besides the minimum interpretation, there many other interface and objects that can be interpreted and /or calculated from the GPR data. Their identification and reporting requires extra work and have to be ordered separately. Some of the extra packages are listed in the following: 1) Extra package A: Dielectric value of asphalt and base course (if reliable), antenna elevation info 2) Extra package B: Identification of other unbound layers such as sub base and filter course 3) Extra package C: Identification and detecting of steel reinforcements, geotextiles and frost insulation boards from the data 4) Extra package D: Identification of damages inside pavement, such as debonded bound layers or stripping inside the pavement 5) Extra package E: Identification of bitumen bound layers located in the middle of unbound layers (sandwich structure) less than 1.0 m from the surface. At least the upper surface should be interpreted from these. Horn antenna interpretations should always be recorded at least in one meter intervals. So-called few points interpretation (the thickness is interpreted only from changing points) is not allowed, because it substantially impairs the accuracy. The most common errors can be avoided with the following checklist: Preprocessing must be carried out correctly

28 The data must be scaled correctly, and it has to match with the other data such as videos, maps and profilometer data Different antennas must match the depths of reflection, taking into account the fact that antennas might be side by side. Ground coupled antenna s 0-level can be roughly matched to the data by comparing the response of the air antenna data. Reflection travel times differ only in strongly dispersive dielectric materials. There should be always polarity control in the interpretation. Especially the interface between sub base aggregate (higher dielectric value) and the filtering sand below (lower dielectric value) has inverted polarity and the interface should be interpreted in the middle of the black line. Dielectric values have been calculated and updated. Especially the dielectric value of the pavement should always be visible. When presenting the results, a line with max 1 m running average is used. Also deviation of dielectric value of bituminous pavement is a good indicator of the quality of pavement. If fixed dielectric value is used it should be reported. If there are sections with so-called sandwich-structure unbound layer between bound layers the interpretation line should be drawn directly up to the bottom of the upper bound layer (Figure 6).

29 UFigure 6.U Example how to interprete bound and unbound base layer thicknesses in the case of sandwich structure. The bound layer interface picking should be forced to be lifted up in the point where sandwich structure starts and back to old bound base bottom where new pavement ends. 52B5.4 The ground coupled antenna data processing and interpretation 53B MHz ground coupled antenna data The MHz center frequency antennas basic pre-processing usually includes only the definition of the surface level and background removal filtering operation. If needed, vertical high pass and low pass filtering and gaining can be used. Ground coupled antenna interpretations must be recorded at least in one meter intervals. Interpretation should include at least the following interfaces and reflections (included in the same basic package described in chapter 5.3): Bottom of the unbound structure layers / overall structure thickness (code 71) Bottom of the embankment (code 81) (when the road is on embankment). If the new road is built on an old and poor quality road (gravel road), it should be marked as embankment. Culverts. Found culverts should always have road register distance marked in them. Also culverts transition wedges should be interpreted if they exist and they are detectable. As with air coupled antennas, there many other interface and objects that can be detected and interpreted from the ground coupled GPR data. Examples of the ground coupled extra packages are listed in the following: 6) Extra package F: The quality and type of subgrade soil, and clear interfaces in it (soil type changes 7) Extra package G: Bedrock surface (code 201), if possible to be seen The information from the falling weight deflectometer (FWD) and depth to stiff layer calculation-algorithm information can and should be used to support the interpretation. 8) Extra package H: The old road structures inside the unbound structures (code 35). This layer is usually also the bottom of old base course or sub-base layer and it should be marked as text notification

30 9) Extra package I: big boulders in road structure and subgrade soil 10) Extra package J: Soil replacement structures and other special structures in the road Reliability of the interpretation is described in chapter 5.2 using evaluation method. Ground Penetrating Radar book The most common errors can be avoided with the following checklist: Data is scaled to the length mentioned in Section 5.2 so that the culverts chainage match the road register chainage. Verification of the MHz data to match other antennas data (culverts are a good control and they must be precisely at the same point in all data) Interpreted interfaces must not cross with any air coupled antenna interpretations Interpretation should always include: o Total thickness of structure (code 71), and it can be: a. subgrade soil interface b. top of the embankment or c. old road structure, if it can be classified only as embankment (for instance old gravel road) All embankments should be interpreted, if the interface is inside the time window. To help the interpretation, video control should be used to confirm embankment starting and ending points. Old pavement or other bound layers inside the unbound layer and the other significant structures (for example bound base) must be interpreted and marked as a text annotation at least every 250 meters. The quality/type of the subgrade soil must be marked often enough to help the design. It is recommended to use the falling weight deflectometer information in the interpretation if it is available. Geological control must acknowledged (for instance silt and peat does not appear under the moraine in Finland) When interpreting subgrade soil structures and road structures the following codes should be used: transition wedges 55 (are often interpreted with code 71), bottom of a soil