本科毕业论文---gps-rtk和全站仪应用在土地测量上评价评估.doc

GPS RTK和全站仪应用在土地测量上评估
GPS RTK和全站仪应用在
土地测量上评估

Department of Technical Programs, Biga Vocational School, C¸ anakkale Onsekiz Mart University,Biga, C¸ anakkale, Turkey.
Dickinson Research Extension Center, North Dakota State University, Dickinson ND 58601,

实时动态全球定位系统（RTK-GPS）和全站仪（TS）在GIS环境调查的应用比较。在地质统计学的评价方法中包括，用球形的克立格法，指数法和高斯模型法。本次调查结果表明，面积为3.5公顷或更小的最好的解释是高斯模型，而较大的地区需要一个球形模型。可以观察到，高程误差大于60厘米和水平误差大于30厘米的测量点区域被淘汰。为了增加精确度每次测量点的最佳区域是20×20m2。研究表明，这种情况下可能会导致不准确的调查成本超过估计高达27％。
1 引言
许多农业工程实践要求大量的投资。因此，在设计和建设项目的时候准确的成本估计是至关重要的。在一个不准确的调查结果的地形图并不代表建筑面积。地形图中也已被用来描绘子场管理区（弗雷斯等2001年）和特定地点的土地管理（弗伦岑等1998）。
实时动态全球定位系统（RTK-GPS）是地形图测量的一个组成部分。 RTK是一个测量技术，保持一定的精度是必须的。在RTK，校正后的GPS信号被实时从一个已知的位置的基站接收机传送一个或更多的流动站接收机。随着近年来的发展基于RTK-GPS系统，通过补偿大气延迟，轨道错误和其他变量的GPS几何，水平精度可以实现1厘米，（Ehsaniet人，2004年）。
Satalich和Ricketson教授（1998）报道，依赖时间性的系统误差可能影响RTK-GPS系统的高程精度。克拉克和李（1998年）确定了地形的面积的使用RTK-GPS设备与海拔地区4-9厘米的误差。 Wilson等人（1998）即使是小的差异产生RTK-GPS在个别点的海拔可能会导致很大的差异等参数坡度和集水区。
虽然RTK-GPS系统的成功应用，可以提高地形测量的质量，有情况下，这些系统的准确性是值得怀疑的。自然的或人为的障碍，如树木和建筑物可以堵塞所有的RTK系统，使其有限或根本无法使用。在这种情况下，使用全站仪（TS中）。 Borgelt等人（1996年）比较RTK-GPS系统和全站仪的测量精度。他们得到的高程误差12厘米。精确的测量可以使用TS。然而，现场条件并不总是能达到仪器的精度。因此，这些系统通常被一起用来进行土地调查。
地质统计学方法，如克立格法是用来评估第二次独立调查的时间和成本的地形图的合规性。即使克里格法并没有提供第二次调查的明确答案，，它是一种有效而廉价的技术（巴顿等人，1999）。
为了计算的准确性，用地质统计学技术的克里格法得到的地形地表调查数据，用于创建预测的地图和空间统计。克里格法是一种不同的方式来思考是确定性的内插预测。克里格法，预测的值取决于两个因素：趋势和额外的元素的变异性。例如，有一座小山的顶部，海拔从低海拔上升的趋势。不过，也有可能是因为峡谷，溪流以及其他有限元分析的2006年（ESRI）的变化。可能是考虑到这些变化，克里格分配权重的一个子集附近的点被称为“最近的邻居（戴维斯和高利衡，1984）。为了找到权重，测量点之间的空间依赖性的信息应该是已知的（多德1984）。可以用一个半方差函数来表达这种空间依赖性（BRAS和罗德里格斯伊图尔韦1985）。
半方差函数可以表示为（巴顿等1999）：
（1）
其中γ（dij）是半方差函数的点Pi和Pj和高程Zi和Zj和距离d，基于“内在的假设”（德尔霍姆1983），等式（1）可以被写为：
（2）
其中N表示测量的点对数。
按照公式（2），半方差函数的一半就是距离为d的两个点之间的仰角平方差的预期。半方差函数的最大值，称为S，达到在距离称为范围R。范围是指需要观察Zi和Zj被认为是不相关的（Barton等1999）的平均值。最好的半方差函数模型，可以选择观察的根均方值与不同滞后大小和数字（ESRI2006）的试验和错误的方法。塞兰奇等人（2005）和Johnston等人（1996）报道，滞后大小×的滞后阶数的应该小于在数据库中的最长的距离的二分之一。后面数据的大小和数量是确定的，其他半方差函数模型参数，窗台，金块，范围和拟合模型的类型可以优化观测的最小均方根误差（RMSE）。一个典型的半方差函数和其在图1中所示的参数。

图1一个典型的半方差函数和它的参数。

图2数据分布的表示
常见的半方差函数模型包括球形的，指数的和高斯的（Christakos1984）。一旦这些半方差函数模型的最佳参数被确定下来，这种交叉验证的技术可用于选择完美的半方差函数（戴维斯，1987年，1999年Bartonet人）。

图3 曲面拟合办法

本文的重点是评估RTK-GPS和TS系统在农业工程设计实践中动态的精度的测量土地，。实现这一目标的实现以下具体指标：
l 通过调查RTK-GPS和ITS系统获得数据的准确性评价。
l 利用地理信息系统和统计方法来执行上述数据质量分析。
l 探究调查数据的质量对工程设计的影响。
2、材料和方法
2．1测试站点定位和数据收集
在美国北达科他州的距离Lefor附近一块儿50公顷的面积土地被选为试验现场，因为那里没有自然的或人为的障碍遮挡卫星信号。使用Trimble5700 RTK-GPS（Trimble导航有限公司，美国加州桑尼维尔市）的调查系统对该地区进行了调查。一个基站和4个参考点，建立在调查区的最高点。调查点之间的距离是15米。RTK-GPS的连续测量模式下，一个人使用GARMIN GPSMAP76C（Garmin国际公司，Olathe的堪萨斯州）的手持GPS装置，保持5-10公里/小时的速度运动（Renschler等人，2002）。为了保证的测量点之间的距离相等。引用4个参考点，尼康NPL-350全站仪（尼康Geotecs有限公司，东京，日本）对调查的区域进行现场校准，该区域的西北部分占地约3.5公顷。GPS调查车辆跟踪和手持式GPS装置用来收集TS读数尽可能接近的RTK-GPS测量点。
2．2数据处理
最初可用的地形数据集存储点的测量。每个点北向，东向和高程值。做数据处理的目的是创建等高线图并用于设计引水渠道，2006年的Autodesk土地桌面系统。创建三层为50公顷面积RTK-GPS系统，在被调查的TS和RTK-GPS系统为3.5公顷。现在起，这些层将被称为“完整的GPS，TS，部分GPS”，分别为（图2）。部分GPS数据，使我们能够观察到的调查的那些调查点质量。
所有三个层次的北向，东向和海拔每一个点的调查数据被保存为数据库文件（DBF）在Autodesk。然后，ArcGIS的Geostatistical Analyst扩展模块被用来进行地质统计学分析。在分析中，结构化的过程随后asshown在图3中。

图4 全部和部分RTK-GPS等高线地图

图5 清洁引水渠道布局后
最后，调查质量的影响，在工程设计中，创建一个189米干净的水导流明渠以4:1的边坡比在所有这三个数据集（层）。挖/填在Autodesk软件进行计算。
3 结果与讨论

图6 净水分水渠剖面

图7通道切体积部分，完整的RTK-GPS的地图。
图表一显示了半方差函数由于球形、指数、高斯模型、交叉验证统计产生的很小的均方根误差。交叉验证提供了一个想法“多好”模型预测创建表面时未知值。在交叉验证模型省略了一个点顺序,预测其值使用其余的数据,然后比较了预测值和观测值。在一个最好的模型中，平均错误应该接近于零；均方根和平均标准差应该尽可能小。最小的均方根误差值由完整的RTK-GPS测量用球状模型（格瓦特 2000）获取。交叉验证的表格还显示这个区域的大小和数量的读数可以导致不同的半变差函数模型。尽管三次测量读取时间间隔是相似的，完整的RTK-GPS数据是由一个球面模型解释而其他的是用高斯模型。
数表4显示了部分和完整RTK-GPS测量的不同。局部等值线图使用相同的测量点在一个较小的区域可引起高达30厘米的水平误差，和0.7904的均方根误差，而相同数量的点对完整轮廓地图是0.4144。 分析表明，有更多的读数的较大地区，地图质量更好的。

图8调查点的数量和削减量的关系
同时还发现，一个小数量的测量点和一个较小的区域对垂直精度有明显的效果。清洁引水渠道剖面代表三次测量数据的垂直精度（数表5和6）。正如预测的一样，全站仪和部分RTK GPS地图因产生了类似的地面剖面。
为了进一步观察分割成不同测量区域对计算体积的影响（如图8），分别选择33，62，80，100，和162个部分勘察区。完整的调查区域有749个调查点，分割后的体积为1460m3。随着测量点的数目减小，切割的体积增大。3.5公顷的面积分为100个测量点和50公顷分为749个调查点的区别是可以忽略不计。因此，每3.5公顷区100个调查点或每20米的距离一个测量点得到的结果是相识的。
4 总结
使用调查的数据建立精确的地形图必须特别小心，因为这是所有的农业设计的关键。地理分析包括测量区域的面积和任何对半方差模型的影响数据。此外，它的目的是观察不同的测量精度对工程设计的质量和成本估算的影响。不同的区域的大小和读数的数量表现出完全不同的结果。
l 包括RTK-GPS和TS两种技术的比较表明，基于RTK-GPS的调查不仅实用，速度快，而且得到更精确的地形图设计的目的。因此，RTK-GPS系统的投资是值得的。
l 即使在同一地区进行了调查，50公顷的RTK-GPS测量不仅用一个球形的半方差函数模型描述，同时采用高斯模型描述。因此，对于一个成功的调查，不应该只是限制施工现场的区域。提高面积较大的区域的调查质量。
l 调查技术，区域和测量点之间的距离的不同，导致在高程误差可能会达到60厘米，这样会产生成本估算和实际成本之间的巨大差异。

参考文献

Barton J M H, Buchberger S G and Lange M J 1999 Esti-mation of error and compliance in surveys by Kriging;J. Surv. Eng. 125(2) 87–108.
Borgelt S C, Harrison J D, Sudduth K A and Birrell S J 1996 Evaluation of GPS for applications in precision agriculture; Appl. Eng. Agric. 12(6) 633–638.
Bras R L and Rodriguez-Iturbe I 1985 Random Functions and Hydrology (Reading, Mass.: Addison Wesley),pp. 359–425.
Christakos G 1984 On the problem of permissible covariance and vario-gram models; Water Resour. Res.20(2)251–265.
Clark R L and Lee R 1998 Development of topographical maps for precision farming with kinematic GPS ; Trans .ASAE 41(4) 909–916.
Davis B 1987 Uses and abuses of cross-validation in geo-statistics; Math. Geol.19(3) 241–248.
Davis M W and Culhane P G 1984 Contouring very large data sets using Kriging; Geo-statistics for Natural Resources Characterizations, Part 2. (Dordrecht, The Netherlands: Reidel), pp. 599–619.
Delhomme J P 1983 Kriging in the hydro sciences; In:Flow through Porous Media (ed.) G P Findler, pp. 99–113.
Dowd P A 1984 The variogram and Kriging: Robust and resistant estimators; Geo-statistics for Natural Resources Character zations , Part 2. (Dordrecht, The Nether lands: Reidel), pp. 91–106.
Ehsani M R, Upadhyaya S K and Mattson M L 2004 Seed location mapping using RTK GPS; Trans. ASAE47(3)909–914.
ESRI 2006 Introduction to ArcGIS geospatial analyst extension; Environmental Systems Research Institute,C.
Fraisse C W, Sudduth K A and Kitchen N R 2001 Delineation of site-specific management zones by use of unsupervised classification of topographic attributes and soil electrical conductivity; Trans. ASAE 44(1)155–166
Franzen D W, Cihacek L J, Hofman V L and Swenson L J 1998 Topography-based sampling compared with grid sampling in the northern Great Plains; J. Prod. Agric.11(3) 364–370.
Goovaerts P 2000 Geo-statistical approaches for incorporating elevation into the spatial interpolation of rainfall ;J. Hydrol. 228(1–2) 113–129.
Johnston K, Hoef J M V, Krivoruchko K and Lucas N 1996Using ArcGIS Geo-statistical Analysis(New York, N.Y:ESRI)
ND-NSPMP 2005 North Dakota nonpoint source pollution management program; Bismarck, North Dakota: North Dakota Department of Health.
Renschler C S, Flanagan D C, Engel B A, Kramer L A and Sudduth K A 2002 Site-specific decision-making based on RTK GPS survey and six alternative elevation data sources: Watershed topography and delineation; Trans.ASAE 45(6) 1883–1895.
Sarangi A, Cox C A and Madramootoo C A 2005 Geo-statistical methods for prediction of spatial variability ofrainfall in a mountainous region; Trans. ASAE 48(3) 943–954.
Satalich J and Ricketson R 1998 Field test of trimble 4000 real-time kinematic GPS survey system; J. Surv. Eng.124(1) 40–48.
Wilson J P, Spangrud D J, Nielsen G A, Jacobsen J Sand Tyler D A 1998 Global positioning system sam-pling intensity and pattern effects on computing topographic attributes; Soil Sci. Soc. Amer. J. 621410–1417.

Evaluation of RTK-GPS and Total Station for
applications in land surveying
Department of Technical Programs, Biga Vocational School, C¸ anakkale Onsekiz Mart University,Biga, C¸ anakkale, Turkey.
Dickinson Research Extension Center, North Dakota State University, Dickinson ND 58601,

Accuracies of Real-Time Kinematic Global Positioning (RTK-GPS) system and Total Station (TS) were investigated in GIS environment. In geostatistical evaluations, Kriging method was used with spherical,exponential, and Gaussian models. The survey results demonstrated that an area of 3.5 ha or smaller can be best explained with Gaussian model, while the larger areas require a spherical model. A vertical error of 60 cm and a horizontal error of 30 cm can be observed when the survey points outside the construction area are eliminated. The optimum area per survey point was calculated to be 20×20 m2to increase the accuracy. This case study showed that an inaccurate survey can result cost over estimations up to27%.
1. Introduction
Many agricultural engineering practices require a substantial investment. Therefore, an accurate cost estimate is vital for design and construction of the projects. An inaccurate survey results in a topo-graphic map that does not represent the construction area. Topographic maps have also been used to delineate sub-field management zones (Fraisse et al2001) and site-specific soil management (Franzenet al 1998).
The Real-Time Kinematic Global Positioning System (RTK-GPS) is an integral part of topo-graphic surveys. RTK is a technique employed in practices where precision is a must. In RTK, corrected GPS signals are transmitted in real time from a base receiver at a known location to one or more rover receivers. With the recent developments in RTK-based GPS systems, a horizontal accuracy of 1 cm can be achieved by compensating for atmospheric delay, orbital errors and other variables in GPS geometry (Ehsaniet al 2004).
Satalich and Ricketson (1998) reported that time-dependent systematic errors may affect the vertical accuracy of RTK-GPS systems. Clark and Lee (1998) determined the topography of field size areas using RTK-GPS equipment with elevation errors of 4–9 cm. Wilson et al (1998) reported that even small differences in RTK-GPS derived elevation at individual points can result big differences in such parameters as slope gradient and catchment area.
Although successful application of RTK-GPS systems can increase the quality of a topographic survey, there are cases where the accuracy of these systems is questionable. Any blockage from natural or man-made obstacles such as trees and buildings can make use of RTK system limited or impossible. In such cases, Total Stations (TSs) are used. Borgelt et al (1996) compared the accuracy of RTK-GPS systems to TS. They reported elevation errors of 12 cm. Precise measurements can be made using TS. However, field conditions do not always allow the instrument’s accuracy.Therefore, the systems are being used together in land surveys.
Geostatistical methods such as Kriging are used to evaluate the compliance of the topographic maps with second independent survey in terms of time and cost. Even though Kriging does not provide as definitive an answer as a second survey, it is an effective and inexpensive technique (Barton et al1999).
To account for the accuracy of topographic survey data, Kriging, a geostatistical technique, is
used to create prediction maps and spatial statistics. Kriging is a different way of thinking about the
prediction than is done with deterministic interpolators. In Kriging, a predicted value depends on two factors: a trend and an additional element of variability. For example, there is an upward trend in elevation from a lower elevation to the top of a hill. However, there are likely to be variations because of valleys, streams, knobs and other features (ESRI 2006). May be to account for these variations, Kriging assigns weights to all or a sub-set of nearby points called‘nearest neigh ours ’(Davis and Culhane 1984). In order to find the weights, information on spatial dependence among the survey points should be known (Dowd 1984). A semi variogram can be used to express this spatial dependence (Bras and Rodriguez-Iturbe 1985).where N is the number of pairs of survey points.
As per equation (2), semi-variogram is half the expected value of the squared difference in elevation between two points with lag distance of d .The maximum value of the semi-variogram, called the sill S , is attained at a distance called the range R . The range indicates the average separation needed for the observations z I and z J to be considered uncorrelated (Barton et al1999). The best semi-variogram model can be selected observing the root-mean-square values with a trial and error approach for different lag sizes and numbers (ESRI 2006). Sarangi et al (2005) and Johnston et al (1996) reported that lag size × the number of lags should be less than one-half of the longest distance in the database. After the lag size and number of lags are determined, other semi-variogram model parameters, sill, nugget, range and fitted model type can be optimized observing the smallest root mean square error (RMSE).A typical semi-variogram and its parameters are shown in figure 1.
Common choices for a semi-variogram model include spherical, exponential, and Gaussian (Christakos 1984). Once the optimum parameters for these semi-variogram models are determined, a technique called cross-validation can be used to select preferred semi-variogram (Davis 1987;Bartonet al 1999).
The focus of this paper is on evaluating the dynamic accuracy of RTK-GPS and TS system in surveying land for agricultural engineering design practices. This goal was achieved through following specific objectives:
l Evaluating the accuracy of data obtained through surveys where RTK-GPS and TS systems were used.
l Using GIS and geostatistical methods to perform above-mentioned data’s quality analysis.
l Investigating the effects of survey data quality on engineering designs.
2. Materials and method
2.1 Test site location and data collection
The 50-ha area, near Lefor, North Dakota, USA was chosen as the test site because there is no natural or man-made obstacles that can block satellite signals. The area was surveyed with a Trimble 5700 RTK-GPS (Trimble Navigation Ltd., Sunnyvale, California) survey system. The base station and four reference points were established over the highest point in the survey area. The distance between the survey points was 15 m. An operation speed of 5–10 km/h was maintained (Renschler et al 2002) in RTK-GPS’s continuous topo mode. In order to maintain equal distances between the lines of survey points, a hand-held Garmin GPS map 76C (Garmin International Inc.,Olathe Kansas) GPS unit was used. Referencing the four reference points, a Nikon NPL-350 (Nikon Geotecs Co. Ltd., Tokyo, Japan) Total Station was site-calibrated to survey NW section of the area that covers about 3.5 ha. GPS survey vehicle tracks and hand-held GPS unit were used to collect TS readings as close as possible to the RTK-GPS survey points.

2.2 Data processing
The available topographic datasets were originally stored as point measurement. Each point had nor-thing, easting, and elevation values. Data processing was done to create contour maps that would be used to design diversion channels in Autodesk Land Desktop 2006 (Autodesk, Fremont California) soft-ware. Three layers were created for 50 ha area that was surveyed with RTK-GPS system, and 3.5 ha for TS and RTK-GPS systems. Now on wards, these layers will be referred as ‘full GP