Field Manual 3-34.331 TOPOGRAPHIC SURVEYING 16 January 2001

TOC Chap1 2 3 4 5 6 7 8 9 10 11 AppA AppB AppC AppD Gl Bib

Appendix B

Control-Survey Standards

This appendix is designed as a quick reference for platoon leaders. It summarizes the standards for control surveys that were discussed in Chapters 6, 7, and 8.

DIFFERENTIAL LEVELING

B-1. Differential leveling is the conventional method of leveling for the propagation of orthometric heights. Table B-1 and Tables B-2 and B-3 show the overall standards and specifications for differential leveling.

HORIZONTAL-ANGLE MEASUREMENT

B-2. The observation requirements for horizontal-angle measurements are shown in Table B-4. Adherence to these requirements should ensure that the appropriate level of precision is achieved.

TRIGONOMETRIC OBSERVATIONS

B-3. Trigonometric observations are used to determine trigonometric elevations. To achieve a desired order of trigonometrical elevation, use the procedures and standards for the particular observation type (for example, vertical angle or distance) unless specified otherwise in Table B-5.

GPS TECHNIQUES

B-4. There are two fundamental GPS techniques relative and absolute-point positioning. The recommended practices for the GPS refer only to relative positioning. Relative positioning requires two or more GPS receivers. The two fundamental types of GPS receivers are navigational and survey (or geodetic). The receivers are distinguished by the accuracy level and type of measurements taken during surveys. Many receivers are capable of a number of measurement types. Pseudorange and carrier-phase measurements are the two fundamental types of made with GPS receivers.

B-5. Relative-positioning techniques can be divided into two main groups static and kinematic. The fundamental difference is that kinematic techniques require maintaining lock throughout the survey after ambiguity resolution. These static and kinematic techniques employ carrier-phase measurements. Since a carrier-beat-phase measurement is the only type that offers a sufficient precision in geodetic positioning at third order and higher, the use of receivers that measure the carrier phase is mandatory. Static and kinematic techniques can be grouped as follows:

• The static group can be divided into the following techniques:

• n Static (also referred to as classic static).
• n Pseudokinematic (for example, intermittent static, pseudostatic, or reoccupation kinematic).
• n Rapid static (also referred to as quick static or fast static).

• The kinematic group can be divided into the following techniques:

• n Stop-and-go kinematic (also referred to as intermittent kinematic or semikinematic).
• n Kinematic (also referred to as continuous kinematic).
• n OTF/ (also referred to as ambiguity-resolution OTF).

B-6. A third group of relative-positioning techniques is based on pseudorange measurements. These techniques, either in postprocessed or real-time modes, are referred to as DGPS and are generally not used for precise control surveys. DGPS is used for accuracies of 2 to 5 meters. Precise DGPS is used for accuracies of 1 meter or less.

B-7. By combining carrier-phase measurements with pseudorange measurements, it is possible to reach higher accuracies with DGPS techniques. While GPS measurements are receiver dependent, the selection of observation techniques is dependent on the precision required and the reduction process to be used.

B-8. When planning a GPS-S, the first step is to choose the appropriate technique for the precision required. Table B-6 provides a guide for what technique to use to achieve a particular order and class of survey. Table B-7 provides references to the order and class of >survey.

B-9. The location and distribution of points in a GPS-S do not depend significantly on factors such as network shape or intervisibility but rather on an optimum layout with sufficient redundancy for carrying out the intent of the survey. The intent of the network design should be to

• Locate new points so that the line of sight between them is clear (when possible).
• Provide error control in the minimum-constraint solution (to enable data validation) and analysis of the accuracy of the survey.
• Produce tie-offs for integrating the survey into previously established control.
• Locate ties to points with existing orthometric heights.

B-10. Redundancies play an important role in fulfilling this intention. All GPS-Ss must be connected to the existing control, the NGS, or the local project to ensure survey integration, legal tractability, and quality assurance. If established control stations are not available in the vicinity of the survey, bring control to the appropriate accuracy by using GPS or conventional techniques. When selecting established stations to connect to, give preference to the highest order of the nearest, established permanent marks (or geodetic stations) that are easily accessible. Connection should be made to a minimum of three points with suitable MSL heights, preferably enclosing the survey, and a minimum of two points with established (horizontal) coordinates. Additional points are to be connected to obtain quality control, with preference given to coordinated marks that enclose the surveyed area and height points spaced throughout the area. A least-squares adjustment of the control survey must be performed.

B-11. The planning of the observations should be such that the error budget is sufficiently minimized. Consider the error budget of a double difference, which consists of error sources affecting measurements; error sources that depend upon the site and the type of instrumentation used; and error sources resulting from reduction, adjustment, and transformation.

B-12. Error sources that affect measurements are tropospheric refraction, ionospheric refraction, and orbit errors. The main error sources affected by the site's location and the instrumentation are centering and antenna-height accuracy, antenna-phase center variation, 3D differential-antenna offset, multipath and imaging errors, differential tropospheric delay, and differential ionospheric delay when using single-frequency solutions. The main error sources resulting from reduction, adjustment, and transformation are the selection of the wrong ambiguities, insufficient redundancy for quality control of the transformation solution, and a geoid model that is too simple or based on too sparse data.

B-13. Redundancy in the observations is the best way of dealing with most of the error sources. Specific observing procedures and differencing techniques can eliminate other error sources that are more systematic. Error sources are reduced by careful site selection, averaging, and sufficient observation time to allow geometry change. Night observations or the use of dual-frequency receivers can minimize ionospheric errors. Antenna offset can be minimized by ensuring identical antenna orientations. Orbit errors are minimized by the use of precise ephemerides.

B-14. The concept of redundancy (when using a GPS) refers to such things as the following:

• Increasing the percentage of points with multiple occupations.
• Tying multiple baselines into one point.
• Observing common baselines between figures.
• Closing onto existing control.
• Computing the polygon closure using data derived from different sessions.
• Observing more than the minimum number of satellites.
• Averaging through observing a sufficient number of epochs.

B-15. Independent reoccupation of the same point (after a sufficient lapse of time) to observe a different baseline is the most common way of detecting gross error. An alternative to independent reoccupations is the inclusion of conventional observations of appropriate accuracy (for example, to create ties between unclosed GPS polygons in the same adjustment). In a control survey, all observations should be checked by the redundancies included in the network. The configuration of the network should involve the observation of closed figures, and closure polygons must combine data from different sessions.

B-16. An independent-baseline measurement in an observation session is achieved when the data used are not just different combinations of the same data used in computation of other baseline vectors observed in that session. In an observation session using five receivers, the total number of baselines can be computed as follows:

where
n = the number of receivers

B-17. However, only four (n - 1) of those baselines are independent. The remainders (10 - 4 = 6) are formed from combinations of phase data used to compute the independent baselines. The results from observations of the same baseline made in two different sessions are independent. Generally, independent-baseline processors assume that there is no correlation between independent vectors. Trivial baselines may be included in the adjustment to make up for such a deficient statistical model. If the mathematical correlation between two or more simultaneously observed vectors in a session is not carried in the variance-covariance matrix, the trivial baselines take on a bracing function that simulates the effect of the proper correlation statistics. And, at the same time, introduce a false redundancy in the count of the degrees of freedom. In this case, the number of trivial baselines in an adjustment should be subtracted from the number of redundancies before the variance factor (variance of unit weight) is calculated. If this approach is not followed, trivial baselines will be excluded from the network altogether.

B-18. To incorporate 3D PS-Ss into local horizontal and vertical data (WGS-84 and MSL), the number, type, and distribution of control points to which connections should be made must be considered. A determination of which technique to use to derive orthometric heights from ellipsoidal heights is necessary. The technique will influence the choice of well-placed strategic points with known orthometric heights that should be observed. Alternatively, orthometric heights can be brought to selected points in the GPS network.

B-19. For a small area (a few kilometers across) with a smooth geoid, solving for transformation parameters brings about a de facto surface fit (tilting the ellipsoid so that it is parallel with the geoid). When a single value for the geoid-spheroid separation is used at the orthometric-control points, it is assumed that the geoid is as smooth as the ellipsoid. For larger areas, choose between a geodetic-leveling, a geopotential model-based, a gravimetric, or a geometrically derived geoid.

B-20. The classification of GPS results (including height) is generally expressed using a linear propagation method, unless requirements specifically call for height classification using differential leveling. In both cases, the class and the order are assigned separately for horizontal and vertical control.

B-21. A system test is recommended to qualify equipment, techniques, and error modeling for a particular accuracy. Evidence of a test may be required after acquisition of new equipment or software, when trying new techniques, or as justification of a chosen method of error modeling. This evidence serves 

• As a justification of observing and processing techniques.
• To validate (under similar conditions) the same equipment, the software, and the observation method.
• To justify the error modeling.
• As a justification of a multiplier used to increase the baseline-vector variance-covariance matrix elements when these are unrealistic.
• To validate data when combining results from different equipment and software.

B-22. The total GPS process is comprised of the following four distinct components:

• Satellites.
• Receiver hardware.
• Field procedures.
• Software.

B-23. The following procedures describe a system test that considers all of the components of the system and are designed to evaluate the performance of multiple receivers used in a differential mode. The field practices and system test have to reflect the particular observing strategies (for example, static, rapid static, or stop and go) that are employed on a project. The equipment should be operated according to the manufacturer's specifications. The test consists of a measurement of a small test network and the ongoing analysis of production results.

B-24. Control should be established on at least one baseline of the small test network. This control consists of a measurement of 

(n + 1) stations and (n + 1) independent baselines

where-
n = the number of receivers

B-25. The test network observed should be a polygon with station spacing not less than 50 meters and not more than 10 kilometers. The independently observed baselines should be processed, baseline by baseline, to produce differences in Cartesian coordinates in the satellite datum (X, Y, and Z) for each baseline. The summation of these differences, for any closed figure, will give a preliminary indication of the performance of the total GPS and is an initial, minimum field analysis. At the first opportunity, performance of a more rigorous approach is essential. The vectors and their associated variance-covariance matrices should be adjusted by the least-squares method to obtain a more complete and comprehensive report on the equipment test. If the results meet the manufacturer's specification, then the manufacturer's specification can be adopted by the user as the measure of the precision attainable with the system. If not, the user's measurement system must be modified to meet the manufacturer's specification or the lower precision must be accepted.

B-26. The measurement of a GPS-S network involves the observation of closed figures. An analysis of the closure of all figures should be carried out to ensure that each figure closes within the expected precision. Closure polygons must combine data from different sessions. A network adjustment is the most efficient way to confirm agreement with established control at the required accuracy.

>B-27. The observation duration has to be long enough to resolve ambiguities and, depending upon the required accuracy, it also has to be long enough to average out multipath effects. This is especially true for second-order and higher surveys. It is preferred to observe five or more satellites, although most techniques will work with a minimum of four satellites. The extra satellites give protection against loss of lock from one of the satellites and speeds up the ambiguity-resolution process.

B-28. Equipment users should refer to the manufacturer's specifications for DOP. DOP is an indicator of the geometrical strength of a four-or-more satellite constellation as it applies to instantaneous point-position fixing. PDOP refers to the three position coordinates (while GDOP includes a term for the clock offset). The lower the number, the better the geometry for achieving an accurate point position. Use caution in applying this parameter as an absolute acceptance or rejection criterion, particularly in relative GPS positioning where longer observation periods remove most common biases. However, sufficiently changing geometry during a recording session assists in the determination of ambiguities. Once the ambiguities are resolved, PDOP should be kept low.

>B-29. The minimum satellite elevation is 15". This requirement can be reduced to 10" for third-order and lower surveys. It is necessary to ensure that the receiver-data rates are the same or a common integer factor of 60", which results in sufficient common data to resolve ambiguities. Give special attention when processing data collected from different types of receivers (5", 10", 20", or 30" are typical). The time intervals must also be simultaneous.

B-30. When a reflective environment (horizontal, vertical, or skew) cannot be avoided, refrain from using both low satellites and satellites within half an hour either side of culmination for that site. Occupy the position (and the others in the same session) for a minimum number of minutes that is equal to 40 divided by the perpendicular distance to the suspected reflecting surface. This will generally increase the chance of capturing at least one full swing of the interference. When third-order and higher accuracies are required and a site with a reflective environment cannot be avoided, it is worthwhile to average longer observation times of the interference (virtually the equivalent of a static survey).

B-31. Manufacturers generally give a guide for the average time required to resolve ambiguities. When these times are shorter than the minimum observation duration recommended above for reducing multipath, the observation duration should be lengthened (following the above guide) if in a reflective environment.

B-32. Field procedures are substantially the same as recommended above when using static and kinematic techniques for requirements less stringent than second order. This is because relative GPS can routinely deliver second-order accuracy. Refer to the manufacturer's manual for any additional requirements. RTK carrier-phase techniques already impact on the first 10 kilometers of second- and lower-order surveys. At this stage of the analysis, different criteria should be applied depending on the project requirements. A summary of the various observational techniques follow.

B-33. Static surveying uses two or more receivers that remain stationary for 30 minutes or more, depending on the line length and the required accuracy. Carrier-phase observations are made, and to enhance the carrier-phase ambiguity resolution, the satellite geometry should be given time to change. Observations are made (with two or more receivers that have a common data rate) to four or more satellites with elevations above 15". An accuracy of 0.1 to 10 ppm is possible, depending upon the quality of the data, the processing, and the length of the baseline vectors.

B-34. Rapid-static surveying uses various combinations of observations (for example, C/A-, P- or Y-code range data and L1 and L2 carrier-range data). If the view of the sky is limited, rapid-static surveying depends on least-squares ambiguity estimation for a determination of the correct ambiguities. The reliability is enhanced when data from six or more satellites are used and multiple occupations are made at different sidereal times. Dual-frequency receivers are advantageous because they allow various data combinations (for example, widelaning) in estimating a solution. Occupation times of 2' to 10' is required to obtain centimeter-level accuracy for vector lengths up to 10 kilometers.

B-35. Stop-and-go-kinematic surveying involves alternately stopping and moving of one receiver, with the main interest being in the stopped positions. This technique relies upon determining baselines, with a minimum amount of data, by resolving the carrier-phase ambiguities at the beginning and maintaining lock throughout the survey.

B-36. In stop-and-go surveying, two receivers observe a predetermined baseline and perform an antenna swap. The antenna swap is used to obtain the baseline in a matter of a few minutes. The process where carrier-phase ambiguities between satellites and receivers are resolved before the other receiver starts roving is called initialization.

B-37. The second receiver then starts roving, staying stationary over points for a few seconds to a few minutes. Constant satellite lock should be maintained on at least four satellites and is the major factor with this technique, which makes it suitable for open terrain only. An accuracy of 20 to 30 millimeters is possible, and accuracies of 1 to 10 ppm have been quoted. Good geometry and the observation of a minimum of a dozen epochs at each survey point are important for this technique. The short occupation times give a rapid drop-off in height accuracy. Good planning is advantageous, and the occasional occupation of a known point is necessary in case the geometry deteriorates or a cycle slip occurs before the survey can be closed. RTK methods fit this category, because RTK presupposes access to actual phase observations at a site with known coordinates, to produce a double-difference, ambiguity-fixed solution in real time.

B-38. Kinematic surveying proceeds as in stop-and-go-kinematic surveying but without stopping. Vectors are created that are associated with single epochs in time.

B-39. Pseudokinematic surveying does not depend on continuous lock of the rover(s) while traveling but requires continuous lock while stationary. The same point is reoccupied after 1 to 2 hours by the same receiver and again for about 3' to 10'. This creates a situation of having one deliberate cycle slip dividing the data. This paired observation is defined as a single station observation. Obtaining the change in satellite geometry enhances the ambiguity resolution. A constant antenna height allows the two data sets to represent measurements to the same physical point in space.

B-40. Accuracies can reach 20 to 30 millimeters depending upon satellite availability and PDOP. Accuracies of 2 to 20 ppm have been quoted. Single- or dual-frequency carrier-phase receivers can be used. Dual-frequency observations, although not necessary, enhance the determination of the ambiguities. For practical purposes, maximum vector lengths are about 15 kilometers.

B-41. While not as productive as the stop-and-go-kinematic technique, the pseudokinematic technique does not rely on maintaining satellite lock. The pseudokinematic technique is much more practical in areas where trees, buildings, tunnels, overpasses, or other obstructions are likely to interrupt the signal or where interstation access is slow.

B-42. OTF/RTK surveying uses a continuous kinematic technique, which is ideal when the roving receiver cannot stop for an initialization. OTF/RTK does not need initialization; it performs auto-reverse processing as soon as the ambiguities are resolved. Contrary to the definition of kinematic techniques, OTF/RTK does not need initialization at the start. A sufficient number of dual-frequency observations to, preferably, five satellites with good PDOP are required. After the dual-frequency observations, only four satellites are required. Vectors are created that are associated with single epochs in time. For distances up to 20 kilometers, a conventional static or rapid-static setup is required as initialization. Single-frequency techniques are also used with OTF/RTK.

B-43. The term differential is generally used with pseudorange techniques that resolve the errors in a single position. One of these techniques is real-time DGPS, which resolves the errors in real time. This is in contrast to the vector approach of relative GPS, which is achieved by observing C/A-code-phase (pseudorange) error measurements at one or more known stations and then transmitting the data to the remote station(s).

B-44. Table B-8 shows procedures for static- and kinematic-GPS techniques. Occupation time at a point is equipment and distance dependent and is sometimes indicated by the receiver. The longer the occupation time the greater the chance that ambiguities are resolved and that instrument noise and multipath interference is averaged out, which gives more reliability.

B-45. The quality of the results of a GPS-S is determined by both the method of observation (including choice of equipment) and the quality of the reduction, adjustment, and transformation procedures. The initial station position of the datum for any baseline calculation should not exceed 10 meters for each ppm accuracy required and is best obtained by transformation or by connection to another point with known coordinates in the satellite datum.

B-46. The reduction procedures outlined in Table B-9 give a broad overview of the essential components to consider when undertaking the reduction of GPS data. Adhering to the procedures in this table does not remove the necessity for statistical analysis of the results. The table format gives a clear picture of the specific reduction requirements for achieving a given geometric standard of survey. These reduction procedures indicate the minimal requirements.

B-47. Because of the effect of the ionosphere, dual-frequency receivers are used on lines over a certain length. L1-only solutions often show less noise for vector lengths below 10 kilometers. Single-frequency receivers can still satisfy high-order survey requirements up to 20 kilometers but need an increasing number of hours of observation if a higher order of survey is required or if longer baselines are observed. Dual-frequency ambiguity-fixed L1/L2 solutions in their ion-free linear combination are usually obtained for vector lengths from 10 to 50 kilometers. An ambiguity-fixed solution is preferred, but as the distance increases, it becomes harder to achieve. Ion-free, ambiguity-float L1/L2 solutions have become more common for vectors of 40 to 90 kilometers. For longer baselines, triple-difference solutions can be used if the observation time is long enough to enable a sufficient change in the satellite geometry during the recording session. As a guide, use 30 minutes as a minimum plus an additional 20 minutes per each 10 kilometers of baseline length.

B-48. When processing minimally constrained adjustments, the processing software must be able to produce the variance/covariance statistics of the observed Cartesian vectors so that the adjustments can be input into a 3D adjustment program. A least-squares adjustment must be performed when deriving values for control surveys. The software must be capable of determining transformation parameters between the observed Cartesian vectors and the local geodetic system.

B-49. Error ellipses should be calculated after a minimally constrained least-squares adjustment. These calculations prove the quality of the network design rather than the quality of the observations. The error ellipses should be scaled by the a priori variance of unit weight (generally equal to one), unless the a posteriori estimate of variance does not pass the chi-square test. In the latter case, the observations, the statistical model, or even the mathematical model should be examined and the problem remedied and the adjustment rerun. In the case of not being able to remedy the situation, the error ellipses should be scaled by the a posteriori variance factor.

B-50. To confirm the quality of the observations, the standardized residuals should be checked for outliers. The checking of the statistics often involves critical evaluation of the a priori standard deviations of the observations. If the baseline variance/covariance matrix is routinely modified by a multiplier, documentation of a measurement over a test network may be required as confirmation of the multiplier used.

B-51. To conform to the internal consistency requirements for a particular geometric accuracy, the error ellipses should confirm the capability of the network design to meet the specifications. The standardized residuals and the estimate of variance should confirm that the observations have actually met the required standard.

B-52. All points in a survey should conform to the specifications belonging to the relevant classification. This applies whether the points are connected by baseline observations or not. This is also valid when relative accuracy values are calculated to points with previously established coordinate values. Geoid-separation values are applied to orthometric heights of points that will be constrained in the transformation and adjustment.

B-53. The following four methods are used for determining geoid heights:

• Geodetic-leveling-geoid method. The use of a geodetic-leveling geoid.

• Global-geopotential-model method. The use of different types of global-geopotential models derived from gravity and active or passive satellite information.

• Gravimetric method. The use of an intensive grid of local gravity, together with a high-degree global-geopotential mode.

• Geometrically modeled geoid method. The use of a geometrically modeled geoid by contouring geoid-height values derived from orthometric and ellipsoidal heights.

B-54. The relative accuracy of height values resulting from the global-geopotential-model method are dependent on the grid spacing of the geopotential model used. The spacing of points with observed local gravity in the gravimetric method and the spacing of leveled points in the geometrically modeled geoid method determine the relative accuracy.

B-55. The geodetic-leveling-geoid method is generally not accurate enough to convert GPS-ellipsoidal heights into orthometric but works well with height differences. The global-geopotential-model method is useful in case of long baselines an area a smooth geoid and scarce orthometric-height points. gravimetric most accurate when sufficient dense grid gravity information available. geometrically modeled notsurface fitting contouring are recommended for short distances (10 kilometers or less).

B-56. The next step is derivation of transformation parameters between minimally constrained adjusted vectors and selected points in local geodetic system. This usually carried out together with a least-squares adjustment. adjustment subjected to same analysis as Error ellipses are calculated again network allocated an accuracy order that enables its orderly integration database contains existing data set established coordinates.

NOTE: Refer to EM 1110-1-1003 for a complete sample of an adjustment statistics summary.