Error Patterns of GPSDr R.K.Shukla Assistant Professor in Civil Engineering Dept of Civil Engineering M.M.M. Engineering College Gorakhpur, UP drshukla_gkp@rediffmail.com Mr. Naveen Kr. Sharma & Mr. Tarun Sinha Pursuing B.Tech in Civil Engineering M.M.M. Engineering College, Gorakhpur, UP. naveen.mmmec@gmail.com tarun.sinha@yahoo.com Abstract: The efficiency of working on any project solely depends upon three important dimensions viz. Time, Accuracy & Economy. The optimum achievement of all the three, hence, becomes vital. The use of code based hand held GPS is gaining momentum worldwide for its ease to achieve the above target. But there are certain issues which cause a deep impact on the results obtained from GPS technology. This paper throws light on such issues hence promulgating certain ways, supported by experimental results, to use the GPS technology for utmost accuracy in minimum time and economically soothing. The accuracy is often misunderstood in terms of the standard errors (S.E) as reported by the software. Studies were carried out to ascertain what is the relationship between the SE as reported by instruments and the real ground measurements. Comparison between the actual ground measurements and the SE do not bear any significant relationship between each other. The actual errors are often much more than the position errors as reported by the vendors. Introduction: The Global Positioning System (GPS) is a burgeoning technology, which provides unequalled accuracy and flexibility of positioning for navigation, surveying and GIS data capture. The GPS provides continuous three-dimensional positioning 24 hrs. a day throughout the world. Geographic Information System (GIS), also known as a geospatial information system, is utilised for capturing, storing, analyzing and managing data and associated attributes, which are spatially referenced to the Earth. The database of a GIS must include all the geographic features on the earth and hence the problem of location and dimension of a particular feature gets resolved easily when this sytem is used in combination with GPS. The use of GPS in order to create a GIS is aimed at to provide accurate, global scale, continuous, 3-D position information (Min et al 2005). on a suitably defined Earth-fixed coordinate system in minimum possible time. Thus accuracy and time become important constraint while defining the location of a particular feature on the earth (Kennedy, 2002) The desired accuracy is achievable through GPS but at the expense of data collection time at each point of interest. GPS also suffers from errors due to atmospheric conditions, multipath, ephemeris, and the receiver quality (Langley, 1997). These problems occur due to the electromagnetic behavior of signals being transmitted through the satellite and their interaction with the atmosphere. Therefore, it becomes necessary to study the effect of these errors on shorter distances commonly occurring in case of GIS databases. It has been a common practice to report the accuracy in terms of standard errors obtained through statistical techniques based on several epochs of signals accumulated over a considerable period of data collection. The errors, thus determined may or may not bear any significant relationship with the actual ground measurements. Objectives of study: The basic objective of the present study has been to relate the actual accuracy and data acquisition time with a GPS and thus to determine the optimal time of data collection for shorter distances involved in the GIS data collection. A commercially available, popular GPS receiver has been used to access, if Code based GPS data can be suitably used for GIS data collection. In addition, if the standard errors as reported by the commercially available GPS data processing software, bear any relationship with the actual ground measurements. It was also felt to study, if the error in position has any common direction so that it can be classified as a systematic error. Causes of errors in GPS: The origin errors in GPS can be traced as arising from any of the sources such as the satellite itself, the atmosphere in which propagation of signals take place and the receiver station and surrounding environment. Satellite based errors are mainly related to the inaccuracies in timing standards onboard each satellite to which all the measurements are referenced to, or the positional error associated with a satellite’s position along its orbital track. GPS signals contain information about ephemeris (orbital position) and about the rate of clock drift for the broadcasting satellite. The data concerning ephemeris errors may not exactly model the true satellite motion or the exact rate of clock drift. The errors, which occur during the transit of the GPS signal, from the satellite to the antenna, are termed as propagation based errors. These arise due to two main reasons: the reflectivity of the satellite and antenna environment and the properties of earth environment, which cause the signal to follow a different path, than that would it follow in a vacuum. The atmosphere, for GPS positioning purpose can be split into two main layers, the ionosphere and the troposphere, which are treated separately due to their different characteristics and effects on satellite signals. The ionosphere and troposphere both refract the GPS signals. GPS signals are subject to propagation and therefore travel on a curved path instead of a straight line from the satellite to the receiver (Rizos, 1999). This causes the speed of the GPS signal in the ionosphere and troposphere to be different from the speed of the GPS signal in space. Use of the dual frequency observations easily provides the solution free from ionospheric delay. While on the other hand the troposphere is a non-dispersing medium. The refraction is independent of frequency of the signals and therefore, will affect both of the code modulation and the carrier phases. Troposphere delay can not thus be removed by the dual frequency observations. Error sources that originate at the GPS station are related to either the receiver, the antenna type or in the positional accuracy of the location of GPS station. The receiver clocks are inferior to the cesium-rubidium clocks onboard the satellite. The receiver clock error which biases the measurement with respect to GPS time is, however removed when the observations are differenced with respect to one extra satellite than the minimum requirement Satellite geometry also affects the accuracy of GPS positioning. This effect is called Geometric Dilution of Precision (GDOP). GDOP refers to where the satellites are in relation to one another, and is a measure of the quality of the satellite configuration. It can magnify or lessen other GPS errors. All of these sources of error do not remain constant over periods of time, but they might vary at every epoch of GPS signal. At certain times in an area, a good satellite constellation may be available, while at others the constellation may be very poor (Langley, 1997). At the time of data collection, the GIS user might not be serious about these sources of errors. However, GPS users have a common practice of relying on the statistical indicators of errors, obtained during the processing of several positions computed through many epochs of GPS signals collected at a single station. Description of the Instrument used: In the present study, a hand held GPS receiver Leica GS20 was used. GS20 Professional Data Mapper (PDM) from Leica Geosystems is amongst one of the most advanced handheld GPS for GIS and mapping. Data Collection is used for the initial recording and attribution of points, lines, and areas. Data Management is used for the update of attribution and geometry of an object; including relocation and continuation of existing geometry. It is a rugged and highly portable device for any field personnel engaged in GIS data collection. This instrument has been designed keeping in mind the ease in GIS data collection. Thus it provides the all-in-one simplicity of a recreational GPS handheld with the power and flexibility of a professional grade mapping system providing the user with a true turn-key GPS mapping solution. It is a 12-channel; single frequency (L1) GPS receiver offering DGPS capabilities the dual frequency GPS receivers being very costly are seldom used for GIS data collection. However, for geodetic measurements the use of dual frequency receivers is unavoidable in order to achieve the millimeter level accuracies. Description of Software used: The Leica GIS DataPRO software as supplied with the hand held GPS was used for the processing of the GPS data. The various optional processing parameters are available in this software. In general these parameters include selection of elevation cut-off angle to eliminate the noisy signals because of grazing and thus affecting the location. Several correction models are also available for tropospheric and ionospheric delay (Rizos, 1999) and thus the corrected positions can be obtained. Experimental Setup: 
First of all a grid of squares of 20 m side was prepared. This 20 m dimension of grid was assumed because generally the dimension of most of the common objects of interest in a GIS is of the order of 20 to 30 m range. Thereafter, it was decided to use two connected squares of 20 m sides. A theodolite was used for making right angles accurately and marking the ground stations. Special attention has been taken for the measurement of angles and lengths. The geodetic or survey control points for such a short distances are not available in the study area of Gorakhpur. This simple network has distances available between 20 to 45 m. Data collection: After attaining the experimental setup, the observations were taken using Leica GS20 GPS receiver, by differential correction technique. The recording time of the data from GPS was again to visualize real life situations. Hence observations were taken for 10 min, 5 min, 3 min, 2 min, and 1 min. Maximum time for data recording was taken 10 min because time greater than this cannot be feasibly spent at particular locations keeping in mind the duration and economy of the project undertaken. It is well known that due to ambiguity resolutions involved in carrier phase observations, it requires a longer duration of data acquisition at each of stations, and therefore, it was not considered in this study. Data processing strategies and Analysis of results There are different combinations of strategies, due to several optional parameters for post processing of the recorded data available in GIS DataPro. All the combinations were tried out, but some of them had the similar output, (without significant differences) so different sets of observations had been short listed for further investigations in this study. The variance comparisons showed that simplified Hopfield tropospheric model and Standard Ionospheric models showed best results. Further in the processing stage, the GPS co ordinates obtained from GIS Data Pro were converted to UTM system (Maling, 1992 and Ordnance Survey, 1995) using a computer program developed for this study. Based on the UTM northing and easting of the first line, the coordinates for the remaining points of the grid were calculated based on a square traverse and considered to be the correct ground coordinates The GPS derived coordinates were compared with the traverse computed coordinates. The difference in each coordinate is considered to be the actual error present in the GPS measurement system. The similar approach of error determination was adopted for all sets of data corresponding to 1, 3, 5 and 10 minute data collection sets. The error in terms of distance measurements are shown in Table 1. The elevation cut off angle was set up as 10º while processing in GIS DataPRO. Table 1: Error in distances based on 1, 3, 5 and 10 minute stand alone and Differential GPS Data
It can clearly be observed from Table 1, that the data collection sets of at least 5 minutes must be used to achieve the errors in relative positions or distances of the order of 20 to 50 meters. These distances are common in all urban GIS databases. From cartographic considerations, in a map of scale 1: 5000, the errors of the order of 1 m are permissible (NMA, 1947). However, for the purposes which necessitate larger scale maps, use of single frequency code based GPS data collection for 5 to 10 minute duration should be avoided, even if the data is differentially processed. The GPS data acquisition time must be increased. In case of stand alone GPS data collected for duration of 10 minutes, the average error has been observed to be of the order of 3 m (Table: 2) with a maximum error of 5.167 m. Therefore, stand alone GPS data collection should not be carried out for GIS database generation and updation or in other words it should always be differentially processed, if data has been acquired for a short duration of 5 – 10 minutes.. UTM coordinates were obtained for GPS positions. Based on these coordinates for line AB (Figure 1), the positions of remaining points were computed as a theodolite traverse. The results in the form of comparison between computed UTM northing and easting based on GPS positions, and traverse coordinates in form of the first line of the grid to be taken from GPS coordinates and remaining are computed based on the connected squares of the traverse. The results in Table 2 show the positional errors unto 2.5 meters in either direction, while these error vectors do not demonstrate any common directionality. It can thus be concluded that the errors are mostly in random directions in the case of differentially processed GPS coordinates with a data acquisition time of 10 minutes at each point. Table 2: Comparison of GPS derived and Ground coordinates (in UTM) for10 minutes.
Note: A constant quantity has been subtracted from each Easting and Northing ( Easting: 741000 and Northing 2958000)  Figure: 2: Actual and GPS grid Δ-Ground grid point ◊-Grid point from GPS An overlay analysis of the GPS derived grid coordinates over the ground coordinates demonstrate similar patterns as already described in Table 1 above. It can clearly be observed in figure: 2, that 5 and 10 minute observations, if differentially processed do not show any significant improvements of errors in results. However data captured for time less than 5 minutes show an erratic behavior of shifts in error, such that the results are unsuitable for a GIS database which might be used for applications demanding scales larger than 1:10000.  Figure 3: Average Error patterns over distances between 20 to 45 m An analysis was also carried out for the determination of relationship between average errors with respect to distances. In our study grid distances between 20 to 45 m were available. These distances were occurring more than once, so it was considered to observe the relationship between distance and average error with respect to time of data capture with GPS. The error plots for different time intervals are shown in Figure 3. It again demonstrates the same conclusion that data capture for a duration less than 5 minutes is not suitable for GIS data capture if it has to store the information for applications involving details with permissible errors of 1 m. Conclusions: Single frequency, code based, differentially processed GPS data can suitably be used for GIS database generation and updating. The GPS observations should be used for data collection for at least 5 to 10 minutes at each point of interest. Increase in time of data acquisition will improve the quality of results. Standard error as reported by the GPS processing software is merely a statistical indicator which does not have any relationship with the distance on ground. or in other words it can be said that the position coordinates do not shift in the same direction. The residual errors do not show any specific direction therefore it can also be understood that they do not contain any systematic error. Single frequency hand held GPS can suitably be used for GIS data capture or cartographic applications involving scales up to 1:5000 or smaller. GPS technology can help GIS data capture with utmost accuracy with economy of time and cost. References:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||