Issues with Extending a UTM Zone Page 1 Issues with Extending a UTM Zone Ryan Brazeal, M.Sc., P.Eng., P.Surv., PMP Geomatics Technology Manager, Caltech Surveys Ltd. July 2016 Introduction The Universal Transverse Mercator (UTM) system is designed to provide a means of representing each point on the Earth (except for areas surrounding the Poles) using a set of Cartesian (X,Y or E,N) coordinates. The benefits of representing points on the Earth with Cartesian coordinates are for easy planar mapping, easily deriving spatial information (e.g., distances, angles, areas, etc.) from points’ coordinates as well as the easy ability to calculate the coordinates of a point of interest based on spatial information. However, the trade-off for the easiness of these operations is often at the expense of the accuracy of the solution. The UTM system was adopted by the U.S. Army in 1947 for large-scale military maps of the entire world (Synder, 1987). Here in Canada, the UTM system has also been utilized for many national and provincial mapping purposes. The Canada Centre for Mapping and Earth Observation (CCMEO) uses the Universal Transverse Mercator Projection for its popular National Topographic System (NTS) series at 1:50,000 and 1:250,000 scales (NRCan, 2016). Information Services Corporation of Saskatchewan (ISC) requires that surveys submitted must have two points georeferenced to the NAD83(CSRS) datum UTM Extended Zone 13 in order to be approved for filing in the Land Surveys Directory (ISC, 2014). Before we begin discussing the issues with extending a UTM zone, let us first clearly understand the ideas and concepts behind a regular UTM zone. The Regular UTM Zone The regular northern UTM zone coordinate system definition (for all but a select few geographical areas on the Earth) only allows geodetic points within a 6 degree longitudinal area, between 84 degrees north latitude and the Equator, to be converted into projected points. Our discussion will be constrained to these northern UTM zones and as such southern UTM zones will not be discussed here; an interested reader is referred to Snyder, 1987. These 6 degree longitudinal areas are predefined, and hence cannot be any arbitrary 6 degree longitudinal area. The 1st predefined area (called a zone) is between 180 degrees longitude and 174 degrees west longitude with each successive zone progressing easterly. Figure 1 illustrates the extents of each UTM zone (zone numbers are shown along the bottom axis in Figure 1). Issues with Extending a UTM Zone Page 2 Figure 1: UTM zones (source: Synder, 1987) There are 6 defining parameters underpinning a regular northern UTM zone coordinate system, they are: 1) A transverse Mercator projection (i.e., a set of mathematical algorithms) is used to convert a longitude, latitude coordinate pair (i.e., geodetic point) to an X, Y coordinate pair (i.e., projected/ Cartesian point). Metric units are traditionally used. 2) The Y axis of the Cartesian UTM coordinate system is aligned with the meridian at the midpoint of the zone limits (e.g., for the 1st Zone it’s Y axis would be aligned with the meridian at 177 degrees west longitude). The meridian that is used for this purpose is called the central meridian of the UTM zone. Any geodetic point that has a longitude equal to the central meridian’s longitude would have a projected X coordinate of 0m. 3) The X axis of the Cartesian UTM coordinate system is aligned with the equator. This is common amongst all UTM zones. This parallel is called the latitude of origin of the UTM zone. Any geodetic point that has a latitude equal to the latitude of origin will have a projected Y coordinate equal to 0m. 4) In order to avoid experiencing negative X coordinate values within a UTM zone, an offset is applied to the X Cartesian coordinates along the central meridian. An offset of +500,000m is applied to all X coordinates. As a result, any geodetic point that has a longitude equal to the central meridian’s longitude will now have a projected X coordinate equal to 500,000m. This concept is commonly called the false easting of the UTM zone and is common amongst all UTM zones. Issues with Extending a UTM Zone Page 3 5) For northern UTM zones there is no chance of experiencing negative Y Cartesian coordinates and as such no Y coordinate offset needs to be applied (i.e., offset of 0m). This concept is commonly called the false northing of the UTM zone and is common amongst all northern UTM zones. 6) The final defining parameter is a bit harder to describe. One needs to have a firm understanding of the geometric perspectives of map projections in order to clearly visualize this point (this is outside the scope of this article). In short, the derived X and Y coordinates are scaled by a numeric factor (commonly called the scale factor or grid scale factor) in order to minimize the scale variations within a UTM zone (Snyder, 1987). For all UTM zones this scale factor is defined to be 0.9996 (a 1:2,500 scale reduction) and is applicable along the central meridian. This has the implication that spatial quantities (e.g., distances) within a UTM zone will be either too small, too large or truly correct, with respect to quantities measured on the reference ellipsoid. This depends upon where within the UTM zone the quantities are being observed. One important point to recognize is that this scale factor relates how points on the surface of a reference ellipsoid (i.e., geodetic points) relate to their derived projected points on the mapping/planar surface. Neither of these surfaces are guaranteed to align with the topographical surface of the Earth where we actually make measurements (commonly referred to as the ground surface). Extra considerations must be taken into account in order to use UTM coordinates in deriving so-called ground information. In summary, there are 6 defining map projection parameters for a UTM zone. For northern UTM zones only the central meridian parameter varies while the other 5 parameters remain constant. If the defining rules for UTM zones are rigidly applied, the derived spatial information from a UTM defined map/dataset can be guaranteed to have an accuracy of no worse than 1:1,000 with respect to the reference ellipsoid (Synder, 1987). However, in Saskatchewan if the UTM extent rules are rigidly applied, the derived spatial information from a UTM dataset is guaranteed to have an accuracy of no worse than 1:2,500 with respect to the reference ellipsoid (we will call this term the UTM accuracy). These rigid rules dictate that any point on the Earth only belongs within a single UTM zone. Even two points positioned geographically close to one another can each be within a different UTM zone. These two points are then not directly relatable to one another as their coordinates are each in a different coordinate system. This by-product of the rules can be undesirable and can present challenges in mapping, measurement, calculation and design tasks. A commonly used ‘solution’ to this problem is to disregard the rule that states that only points within +/- 3 degrees of longitude with respect to the central meridian are allowed to be projected into the applicable Issues with Extending a UTM Zone Page 4 UTM zone. This ‘solution’ mathematically extends the UTM zone but in doing so it introduces greater scale (and angular) variations as well as the potential to introduce more mathematic fragility in the derived UTM X, Y coordinates. These aspects need to be closely examined in order to better understand the issues surrounding extending a UTM zone. Transverse Mercator Algorithms The transverse Mercator projection mathematics require the use of calculus in order to convert a geodetic point to a projected point, and vice versa. However, in order to create practical algorithms, certain calculus operators (e.g., an integral) are numerically approximated using one of many different techniques. These numerical approximations are a necessity but can introduce rounding type errors within a solution. Commonly used transverse Mercator algorithms are based on Taylor series expansions and begin to grade beyond the standard 6-degree zone-width of UTM (Stuifbergen, 2009). The transverse Mercator algorithms shown in Figure 2 are somewhat cryptic in appearance but nicely illustrate - besides the typewriting skills of someone in the early 70’s - how a series expansion (which technically contains an infinite number of terms) is cut off after a certain number of terms. However, when a UTM zone is extended, the probability increases that more terms should be included within these algorithms in order to achieve suitably accurate results. This is a difficult concept to test, as it requires one to be aware of the exact algorithms that his/her software is utilizing. Nonetheless, it is important to understand that different software may produce different results for the projected coordinates of a point within an extended UTM zone. Figure 2: Transverse Mercator Forward Mapping Algorithms (source: Krakiwsky, 1973) Testing Points In order to understand the issues of scale variations within an extended northern UTM Zone 13, our discussion will now focus on some selected testing points within Saskatchewan. In order to conserve space, angular variations are not quantitatively discussed in this article. Figure 3 illustrates the three different Issues with Extending a UTM Zone Page 5 northern UTM zones (i.e., 12, 13 and 14) that are applicable in Saskatchewan. The testing points are also shown in Figure 3 and labelled as A, B and C. These three points were selected because they reside within Saskatchewan but are outside of the regular northern UTM Zone 13. These three points will represent the extreme cases of the scale (and angular) variations experienced as a result of using an extended UTM Zone 13 within Saskatchewan. Figure 3: Saskatchewan’s UTM zones (source: Ryan Brazeal, 2016) Examining testing point A (49o 00’ 00” N, 101o 21’ 43” W), we notice that it is just slightly outside the extents of the regular UTM Zone 13 (approximately 47 km); but the scale factor has already worsened by over 2 times what it was at the regular extent of UTM Zone 13. The scale factor at testing point A is 1.000470 (1:2,127 UTM accuracy) while the scale factor at the regular extent of UTM Zone 13, nearest to testing point A, is 1.000191 (1:5,235 UTM accuracy). Testing point C (60o 00’ 00” N, 110o 00’ 00” W) is located approximately 111 km outside the extents of the regular UTM Zone 13. The scale factor at this point is 1.000552 (1:1,811 UTM accuracy) while the scale factor at the regular extent of UTM Zone 13, nearest to testing point C, is 0.999943 (1:17,543 UTM accuracy). Testing point B (49o 00’ 00” N, 110o 00’ 00” W) is located approximately 146 km outside the extents of the regular UTM Zone 13. The scale factor at this point is 1.001243 (1:804 UTM accuracy). As noticed, testing point B experiences the worst UTM accuracy of any point within Saskatchewan when using an extended UTM Zone 13. This means points located within the southwest corner of the province will be effected the most when calculated/mapped within an extended Issues with Extending a UTM Zone Page 6 UTM Zone 13. Previously it was stated, that in Saskatchewan if the regular UTM zone extent rules are rigidly applied, the derived spatial information from a UTM dataset is guaranteed to have a UTM accuracy of no worse than 1:2,500. By now applying an extended UTM Zone 13 across the province, this statement must be changed to state that the guaranteed UTM accuracy will be no worse than approximately 1:800. Practical Numerical Problem A practical question concerning the use of extended UTM Zone 13 coordinates was submitted to the Practice Committee by SLSA members and was the main reason for the creation of this article. The question asked was (and I am paraphrasing here), when calculating the position of the midpoint between the NE-36-41-28-W3 and the NE-36-42-28-W3 when using the extended UTM Zone 13 versus the regular UTM Zone 12, the position varies by ~12cm. Why is this? Let us begin with a geographical analysis of the problem using mapping software. First, the NAD83 geodetic positions (i.e., latitude and longitude) of the two NE township corners in question were calculated using NRCan’s GeoMath software (also known as MathTool), available here http://geoscan.nrcan.gc.ca/ starweb/geoscan/servlet.starweb?path=geoscan/fulle.web&search1=R=296440). Next, these geodetic points were then converted into UTM Zone 12 and extended UTM Zone 13 projected coordinates using NRCan’s TRX online tool, found here http://webapp.geod.nrcan.gc.ca/geod/tools-outils/trx.php. Table 1 displays the respective coordinates. Point Latitude Longitude UTM Zone 12 Northing UTM Zone 12 Easting Ext. UTM Zone 13 Northing Ext. UTM Zone 13 Easting NE36-42-28-W3 52 40’ 03.0700” o -109 53’ 17.2000” o 5835865.985 575189.980 5846505.849 169548.777 NE36-41-28-W3 52 34’ 49.2300” o -109 53’ 17.2000” o 5826169.273 575339.536 5836817.974 168890.375 Table 1: Township corners’ NAD83 coordinates Next, the coordinates of the midpoint between the NE township corners were calculated by simply finding the mean coordinate value between township corners’ coordinates (i.e., add together and divide by 2). We will call the midpoint coordinates that were calculated using this approach the average midpoint. The following table displays these coordinates. Issues with Extending a UTM Zone Page 7 UTM Zone 12 Northing UTM Zone 12 Easting Ext. UTM Zone 13 Northing Ext. UTM Zone 13 Easting o 5835865.985 575189.980 5846505.849 169548.777 -109 53’ 17.2000” o 5826169.273 575339.536 5836817.974 168890.375 ∑/2 ∑/2 ∑/2 ∑/2 ∑/2 ∑/2 52 37’ 26.1500” -109 53’ 17.2000” 5831017.629 575264.758 5841661.912 169219.576 Point Latitude Longitude NE36-42-28-W3 52 40’ 03.0700” o -109 53’ 17.2000” NE36-41-28-W3 52 34’ 49.2300” o Average midpoint o o Table 2: Average midpoint NAD83 coordinates Lastly, taking all of these coordinates and reprojecting them into a single projected coordinate system (here we used the Google Mercator system) yields the following map. Figure 4: Map of the Practical Numerical Problem (source: Ryan Brazeal, 2016) As is seen within Figure 4, the distance reported between the average midpoint of the line between the NE township corners in UTM Zone 12 and extended UTM Zone 13 is 0.117m. This value agrees with the SLSA members’ observation, but the question still remains as to what is causing this apparent problem? In an attempt to best explain the problem at hand, I would like to begin by asking the question, what constitutes something being called level? We say that two points within a small project area that have the same elevation coordinates are level, but do we still say that an ocean’s surface is level when in fact we know it is actually curved? The answer relies on how localized our view of the Earth’s surface is and how we choose to represent it (i.e., our frame of reference). Issues with Extending a UTM Zone Page 8 At the core of the problem at hand is a practitioner’s understanding/intuition as to what constitutes a straight line. We like to utilize a planar surface/frame of reference to represent the spatial position of points within a limited region on the Earth (e.g., a UTM zone). Within these frames of reference the concept of a straight line is easy to understand and mathematically define. The definition of a straight line on a planar surface is that it connects two points via the shortest possible path. However, when we expand our view of the Earth and represent it by a reference ellipsoid, the concept of a straight line between two points now becomes unclear. Rather than use the term, straight line, we now need to use the more general term of a geodesic line. The definition of a geodesic line is that it connects two points via the shortest possible path along a surface. Similar to the concept of levelness discussed above, a geodesic line also has curvature. This means that our simple mathematical tools for straight lines in 2D planes are not guaranteed to yield accurate results when trying to answer questions pertaining to geodesic lines. The difference between the UTM Zone 12 and extended UTM Zone 13 positions for the midpoint of the line, is because the UTM coordinates of the township corners were used to mathematically define a 2D straight line between the corners. The simple add together and divide by two mathematics used to calculate the average midpoints were in fact calculating the midpoints along these 2D straight lines. These UTM derived straight lines closely approximated the true geodesic line (i.e., the meridian) between the township corners, but not perfectly. The results of calculating the midpoint using the UTM Zone 12 coordinates were much better than calculating it using the extended UTM Zone 13 coordinates. This is because the township corners involved in this practical problem were near the testing point B we previously discussed. We know that the largest amount of error for UTM derived spatial information within the entire province occurs near testing point B. We will discuss two approaches to solving for the midpoint position of the geodesic line connecting the township corners within both UTM zones. The first approach, which is accurate and relatively easy, involves solving the geodetic coordinates (i.e, latitude and longitude) of the geodesic midpoint and then projecting this geodetic point into each of the respective UTM coordinate systems. Using the reference ellipsoid to relate points of interest is a near perfect way to retain spatial information but it is an underutilized approach because Cartesian coordinates and planar mathematics cannot be used. In order to solve the geodetic coordinates of the geodesic midpoint properly one cannot just simply calculate the mean latitude and longitude coordinates between the township corners (as previously done and shown in Issues with Extending a UTM Zone Page 9 Table 2). This is because there is not a linear relationship between surface distance and angular change on a reference ellipsoid. In general, what needs to be done is to first solve for the ellipsoidal distance between points A and B along with the initial azimuth from point A to point B (solved via the indirect ellipsoid problem). Then, the midpoint’s geodetic position can be computed by starting from point A and going in the direction of the initial azimuth and at half the ellipsoidal distance calculated in the previous step (solved via the direct ellipsoid problem). Both the indirect and direct ellipsoid problems can be solved using NRCan’s INDIR online tool, found here http://webapp.geod.nrcan.gc.ca/geod/tools-outils/indir.php. 1 The table below illustrates the correct midpoint coordinates for the geodesic line between the township corners. For our practical problem, one will notice that there is negligible change in the geodetic coordinates for the midpoint when compared to the coordinates shown in Table 2. This is because of the relatively small ellipsoidal distance between the township corners. Point Latitude Longitude Correct midpoint 52 37’ 26.1506” -109 53’ 17.2000” o o UTM Zone 12 Northing UTM Zone 12 Easting Ext. UTM Zone 13 Northing Ext. UTM Zone 13 Easting 5831017.630 575264.780 5841661.925 169219.482 Table 3: Correct midpoint NAD83 geodetic and UTM coordinates (using first approach) If we were to take these two different sets of UTM coordinates and reproject them into a single projected coordinate system (again we used the Google Mercator system) the following map would result. Figure 5: Map of the Correct Midpoint (source: Ryan Brazeal, 2016) 1 To better understand this concept, use the INDIR online tool to calculate the ellipsoidal distance between the points (30o N, 105o W) and (45o N, 105o W) and o o o o o o o o then between the points (45 N, 105 W) and (60 N, 105 W). You will notice that the point (45 N, 105 W) is not the midpoint between (30 N, 105 W) and (60 o o o N, 105 W). The correct midpoint is (45 01’ 09.6” N, 105 W). o Issues with Extending a UTM Zone Page 10 Notice that there is now only a single correct midpoint displayed. The distances between each of the average midpoints and the correct midpoint are reported as being 0.022m to the UTM Zone 12 average midpoint and 0.096m to the extended UTM Zone 13 average midpoint. These values can be easily calculated by the interested reader by solving the 2D distance between the average midpoint coordinates shown in table 2 and the correct midpoint coordinates shown in table 3 using each respective set of UTM coordinates. The second approach to solving for the midpoint position of the geodesic line connecting the township corners, involves only using the UTM coordinates and the arc-to-chord (also known as the t-T) correction. This approach is technically an approximation. However, it works well within the extents of a regular UTM zone but quickly becomes less accurate when used within an extended UTM zone. In order to best understand how this approach works, one needs to be able to visualize the geodesic line (i.e., a meridian in our problem) in relation to the UTM derived mathematic straight lines. The following figure illustrates how meridians and mathematic straight lines are represented within a transverse Mercator based coordinate system. Figure 6: Meridians, parallels and mathematic straight lines in a transverse Mercator coordinate system (source: Stuifbergen, 2009) As seen in Figure 6, the mathematic straight line (dashed pink) at the central meridian (0o E in this figure) perfectly aligns with the projected meridian. As we move easterly, away from the central meridian, the mathematic straight lines begin to increasingly misalign with the projected meridians; scale and angular Issues with Extending a UTM Zone Page 11 errors also increase. The projected meridians within a UTM coordinate system will always have a concave curvature towards the central meridian (Deakin, 2006). This implies that spatial positions calculated based on UTM coordinates will tend to have a longitudinal (east/west) shift towards the Zone’s central meridian in comparison to their correct geodetic positions. Latitudinal (north/south) shifts can also result, but are significantly smaller. These shifts can be seen in Figure 5, where the UTM Zone 12 average midpoint is shifted to the west of the projected meridian (as the UTM Zone 12 central meridian is at 111o W) while the extended UTM Zone 13 average midpoint is shifted to the east of the projected meridian (as the extended UTM Zone 13 central meridian is at 105o W). The extended UTM Zone 13 average midpoint is also noticeably shifted south. In order to account for these shifts within the UTM derived spatial information, one needs to adjust the position of the calculated average midpoint. The adjustment requires that the arc-to-chord angular correction, the mathematic line distance (i.e., grid distance) and the mathematic line azimuth (i.e., grid azimuth) to be determined. Figure 7 illustrates the arc-to-chord angular correction (t-T) when the geodesic line of interest is a meridian. Figure 7: Arc-to-chord angular correction (source: Ryan Brazeal, 2016) Issues with Extending a UTM Zone Page 12 In this specific case, the t-T angular correction could be computed by subtracting the grid declination angle (𝛄), (which has not been discussed previously), from the grid azimuth angle (αG). t -T = αG - 𝛄 A general equation for solving the arc-to-chord angular correction for any geodesic that only requires the UTM coordinates of the end points of the mathematic straight line, along with the reference ellipsoid parameters, is given in Deakin (2006) as, t -T , the interested reader is referred to this reference for a full explanation of the terms in this equation and further information on the subject. The grid azimuth and grid distance (dG ) are simply solved via an inverse problem using the UTM projected coordinates of the township corners. Next, the projected geodesic between the township corners is approximated by a circular arc where the midpoint of the circular arc is assumed to align with the correct midpoint of the geodesic. This allows for the distance of the perpendicular line (dP ), which connects the average midpoint to the correct midpoint, to be calculated using the following equation. The perpendicular line distances should agree with the distances reported by the mapping software as shown in Figure 5. Next, the azimuth of the perpendicular line from the average midpoint to the correct midpoint (αP) is calculated by adding or subtracting 90o from the grid azimuth (this depends upon what side of the central meridian the geodesic is on). Lastly, the position of the correct midpoint can be calculated via solving a direct problem starting from the average midpoint’s UTM position, moving in the direction of the azimuth of the perpendicular line, and going the perpendicular line distance (dP ). The following table displays the intermediate values and results of this second approach. It should be noted that the results for the correct midpoint’s UTM coordinates are very close to the same values as previously calculated using the first approach. Only the UTM Northing coordinates vary by a handful of millimetres. Issues with Extending a UTM Zone Page 13 Name UTM Zone 12 Northing UTM Zone 12 Easting Ext. UTM Zone 13 Northing Ext. UTM Zone 13 Easting Average midpoint 5831017.629 575264.758 5841661.912 169219.576 t-T —> —> -1.849” —> —> 8.112” αG —> —> -0 53’ 01.05” o —> —> 3 53’ 16.53” αP —> —> 89 06’ 58.95” o —> —> -86 06’ 43.47” dG —> —> 9697.8653m —> —> 9710.2222m dP —> —> 0.0217m —> —> 0.0955m Correct midpoint 5831017.6293 575264.7797 5841661.9180 169219.4808 Value Value o o Table 4: Correct midpoint NAD83 UTM coordinates (using second approach) In closing, this article brought to surface some important aspects for the SLSA membership to consider when choosing to use UTM coordinates within their practices. Even though UTM coordinates have a comfortable feel to them and allow us to easily calculate spatial information, there is a potentially large accuracy trade-off as a result of this comfort and easiness. Using the reference ellipsoid and working directly with geodetic coordinates can often be used to ‘easily’ tackle geospatial problems, but for some practitioners this may require a new set of calculation tools. Food For Thought Why is today’s official map projection/projected coordinate system for the province of Saskatchewan one that was defined nearly 70 years ago for the purpose of mapping the entire world? Is it believed, that using UTM Zone 13 is the only way that allows geospatial information within Saskatchewan to be relatable to other geospatial data? In the United States each state utilizes a projected coordinate system (or several of them) that best work for that state’s mapping purposes. New Brunswick uses a special Double Stereographic projection for their provincial mapping purposes. Why are we not defining a much better fitting and applicable map projection/projected coordinate system for Saskatchewan? The possibility certainly exists to create a moderate map projection for Saskatchewan that brings about substantial improvements in accuracy, commonality in the geospatial data collected and improved standard operating procedures for practitioners. Just a thought! Issues with Extending a UTM Zone Page 14 References Deakin, R.E., (2006). Traverse Computation on the UTM Projection for Surveys of Limited Extent, School of Mathematical and Geospatial Sciences, RMIT University, March 2006. ISC, (2014). Saskatchewan CAD File & Georeferencing Specifications, Information Services Corporation of Saskatchewan, Plan Processing Service, October 4, 2014. Krakiwsky, E.J., (1973). Conformal Map Projections in Geodesy, Department of Geodesy and Geomatics Engineering, University of New Brunswick, September 1973. NRCan, (2016). The UTM Grid - Map Projections, Natural Resources Canada, Earth Sciences, Retrieved from: http://www.nrcan.gc.ca/earth-sciences/geography/topographic-information/maps/9775 on July 10th, 2016. Stuifbergen, N., (2009). Wide Zone Transverse Mercator Projection, Canadian Technical Report of Hydrography and Ocean Science No. 262. Snyder, J.P., (1987). Map Projections - A Working Manual, U.S. Geological Survey Professional Paper: 1395.
© Copyright 2026 Paperzz