Quick Reference Guide # 22 Global Survey Limited

Purpose: Understanding Coordinates and Coordinate Conversions

Instrument : TPS & GPS 1200

System 1200 makes the passing of data between the TPS1200 and GPS1200 instruments simple and
seamless.

For example, a typical topographic survey may mean using GPS1200 to survey points suitable for GPS
surveying and then put the CF card into the TPS1200 to complete the survey. Or control points are
surveyed with GPS1200 and then maybe these points are used as backsight points for a resection when
using TPS1200.

But this of course is only possible with the conversion of the coordinates of the surveyed points. Since the
mixing of TPS and GPS data is becoming more and more commonplace and will continue to do so in the
future, it is worth to go back to basics and explain the fundamentals of coordinate conversions.

1. Storage of Points within System 1200.

The first and most important thing to remember with regards to the storing of points measured using the
individual TPS1200 and GPS1200 instruments is the following:

Points measured with a TPS1200 instrument are always stored with Local Grid coordinates in the DBX
database. Points measured with a GPS1200 instrument are always stored with WGS84 Geodetic
coordinates in the DBX database.

Note, the two terms local grid and WGS84 geodetic should be considered as coordinate types. It is only
possible to convert between coordinate types if a coordinate system is being used. As described later, a
coordinate system (typically) consists of a transformation, a local ellipsoid and a projection and should be
thought of as the mathematical algorithms which allow the conversion between the two coordinate types.

Note, even during the conversion of the coordinates to other coordinate types, the original measured TPS
points always remain stored as local grid within the DBX and the measured GPS points always remain
stored as WGS84 geodetic within the DBX. The mathematical algorithms allow the other coordinate types
to be computed and viewed, but the original stored coordinate remain the same.

2. The Classical 3D Method of Converting Coordinates.

FROM GPS TO TPS
It is, of course, possible to convert coordinates in both directions - from TPS (local grid) to GPS (WGS84
geodetic) and in the other direction.

Below it is described how coordinates are converted from GPS to TPS since ultimately, local most
surveyors ultimately need their surveyed points to be exported with grid coordinates.

a. STEP 1: WGS84 GEODETIC TO WGS84 CARTESIAN

As mentioned, a point surveyed with GPS1200 in the field is stored in the DBX as a WGS84
geodetic co-ordinate. As with any geodetic coordinate, it is described in terms of Latitude (j,
Longitude(l) and Height (h) above the ellipsoid (in this case the WGS84 ellipsoid). To make the
understanding of the coordinate conversion process a little simpler, we will consider how a point
evolves as it goes through each stage of the coordinate conversion process to get to local grid.
Our point P was surveyed with GPS1200 and has been stored in the DBX with the following
WGS84 co-ordinates:

WGS84 latitude: 48N WGS84 longitude: 10E WGS84 ellipsoidal height: 500m

Page 1 of 1
Quick Reference Guide # 22 Global Survey Limited

Knowing the Ellipsoid and using standard algorithms it is possible to compute the corresponding
Cartesian co-ordinates for this same point (described in terms of X, Y and Z). Applying these
algorithms to our point results in the following Cartesian co-ordinates:

WGS84 X: 4211089.525m WGS84 Y: 742528.701m WGS84 Z: 4717247.902m

Notice how much easier it is to imagine on the Earth where a point with geodetic co-ordinates is
than Cartesian co-ordinates. Would you have known that the co-ordinates 4211089.525m,
742528.701m, 4717247.902m relate to a point on the ground near Heerbrugg in Switzerland?

So now it is clear we need an ellipsoid to convert between geodetic and Cartesian coordinate and
back again. However to get to local grid co-ordinates we firstly need to get to Local Cartesian
coordinates.

b. STEP 2: WGS84 CARTESIAN TO LOCAL CARTESIAN

To get from WGS84 Cartesian to local Cartesian coordinates a transformation is required. A
transformation consists of up to 7 Parameters. A full 7 Parameter Transformation consists of 3
shifts (dX, dY and dZ), 3 rotations (Rx, Ry, Rz) and a scale factor. These parameters may
already be known, or may need to be computed. In some cases, not all of these parameters are
required With the appropriate transformation parameters it is possible to use standard algorithms
to convert between WGS84 Cartesian and local Cartesian co-ordinates.

But what really is the difference between local Cartesian co-ordinates and WGS84 Cartesian co-
ordinates? They both describe the location of the same physical point in Cartesian coordinates, so
why are the numbers different? It is because the origin, and/or the orientation of the two reference
frames (the WGS84 and the local reference frames) are different. The transformation parameters
actually mathematically describe the differences in the reference frames.

The diagram below shows two different reference frames with different origins and orientations.
In our example we will use a simple 3 Parameter Transformation where dX=100, dY=-200 and
dZ=300. This results in local Cartesian coordinates of:

Local X: 4211189.525m Local Y: 742328.701m Local Z: 4717547.902m

c. STEP 3: LOCAL CARTESIAN TO LOCAL GEODETIC

As described earlier, an ellipsoid is needed to convert between Cartesian and geodetic coordinates.
Because we are now converting on the local side that is, converting coordinates between local
Cartesian and local geodetic, this can be called the Local Ellipsoid. In this example we will use
the Bessel ellipsoid. Using exactly the same algorithms as in Step 1 (but with a different ellipsoid)
we can compute local geodetic co-ordinates. In our example we obtain the local geodetic
coordinates:

Local latitude: 4800 0.82316N Local longitude: 95949.66165E

Local ellipsoidal ht: 1468.783m

Page 2 of 2
Quick Reference Guide # 22 Global Survey Limited

d. STEP 4: LOCAL GEODETIC TO LOCAL GRID

The final step in the coordinate conversion process is to compute grid coordinates to do this we
need a Projection. Hundreds of projections exist, all of them are used to convert Geodetic co-
ordinates (on a curved surface) to Grid co-ordinates (on a plane surface). As for the local
ellipsoid, the projection which is used for a specific country is the one which fits best to the shape
of the country with the minimum of distortion. For example a Transverse Mercator projection is
suitable for a country, which is long and thin in a north-south direction whereas a Lambert
projection may better suit a more square shaped country. Some larger countries or States within
the US will even use more than one projection to cover the country.

In our example we use the Universal Transverse Mercator 32 and standard algorithms to obtain
the following grid co-ordinates:

Easting: 790830.175m Northing: 319665.347m Ellipsoidal ht: 1486.783m

Now we have completed the conversion of the co-ordinates of our point from WGS84 geodetic co-
ordinates to local grid with the following coordinate types:

REMEMBER
Most important to remember is that points measured with a TPS1200 instrument are always
stored with Local Grid coordinates and points measured with a GPS1200 instrument are always
stored with WGS84 Geodetic coordinates in the DBX database. This never changes using a
coordinate system simply allows the coordinates to be converted, but the points themselves are not
re-stored.

Generally speaking, a coordinate system consists of several elements the Transformation, the
Local Ellipsoid and the Projection and it is these individual elements which allow the coordinate
conversions to be made.

Within LGO and System1200, this combination of transformation, local ellipsoid and projection is known
as a Classic 3D coordinate system. It is clear that the local ellipsoid and projection must be known in order
to use a Classic 3D coordinate system. If the transformation is not known and must be determined, then it is
necessary that all common points used to determine the transformation must be known in position and
height. But what if you do not know the local ellipsoid and projection? Maybe you need to use GPS to
survey an area which uses a completely arbitrary coordinate system - is it still possible to use GPS to
complete this survey? Or what if the common points being used to determine the transformation are not
known in position and height, but only position or height? In these cases, it would be necessary to compute
and use a OneStep transformation. Regardless of which transformation type is used, the basic idea of
matching common points that is, matching the control points on the ground for which the grid coordinates
are known and which have been measured with GPS - is the same.

Page 3 of 3
Quick Reference Guide # 22 Global Survey Limited

3. THE ONESTEP COORDINATE SYSTEM

Imagine you want to use GPS to survey a quarry. For years, the quarry has been measured with TPS where
the origin of this grid system is a survey marker in the corner of a field with the coordinates 1000, 5000
with additional control points positioned around the quarry. These control points have been here for some
years, they fit together pretty good and are of sufficient accuracy for regular surveys of the quarry.
Clearly, in order to use GPS to survey this quarry we need GPS to fit into this grid system and give the
same coordinates for points as if they had been measured with TPS. The Classical 3D approach cannot be
used here there is no local ellipsoid and projection simply an arbitrary grid! The OneStep
transformation is ideal for this situation. Remember, the ultimate goal of all coordinate systems is to
convert coordinates between WGS84 geodetic and grid coordinates. In order to do this with the OneStep
coordinate system, the position and height components of this transformation are treated separately.

a. HOW IT WORKS
POSITION COMPONENT
Imagine you have now measured all the control points with GPS in the area in which you wish to
work and therefore know the coordinates of the common points in both WGS84 geodetic
coordinates (the GPS measured points) and local grid coordinates (the easting and northing of the
control points). It is now possible to match these common points and compute the position
component.

Note, when matching points in order to compute the OneStep transformation it is possible to
match points by position only so even if the height of a control point is not known, the control
point can still be used, with only the easting and northing of the point being used. The position
component of the OneStep coordinate system can be thought of as being computed in 2 steps
(these 2 steps are invisible to the user).

The first step is that the WGS84 geodetic coordinates of the points are converted to grid
coordinates using a temporary Transverse Mercator projection. The central meridian of this
projection passes through the centre of gravity of the common points. This results in preliminary
grid co-ordinates for the GPS measured points. These preliminary grid coordinates are never seen
on System1200 or in LGO (they are of no interest to a user). The second step is to match these
preliminarygrid co-ordinates with the local grid control points and compute the best-fitting
easting and northing shifts, rotation and scale factor between these two sets of points. The
positional component of the transformation is now computed. What this means is that the GPS
coordinates are squeezed to fit into the local grid coordinates.

Note, it is possible to compute the positional component with only one common point being
matched. In this case, the rotation is zero (grid north will point in the same direction as WGS84
north) and the scale is 1. The conversion process is now modified as shown below.

Page 4 of 4
Quick Reference Guide # 22 Global Survey Limited

Note that different to the Classic 3D transformation, there are no local geodetic or local Cartesian
computed. So now we have the position component of the OneStep transformation (this is actually
nothing more than a 2D Helmert transformation).

HEIGHT COMPONENT
Now we need to compute the height component of the OneStep transformation. Again, the
common points have been matched and so the height component can be computed. Similar to the
way that points could be matched in position only, it is also possible to match points by height
only. It is even possible to compute a OneStep transformation without knowing the height of any
of the control points (in this case the height of the computed local points have the same height as
the WGS84 coordinates).
If only one point is matched in height, then the WGS84 heights are simply shifted to fit to thatone
local height control point.

If two or three points are matched in height then a plane is fitted to these points. If three or more
points are matched in height, a best fitting tilted plane is computed to approximate the local
heights.

b. WHAT DOES ALL THIS REALLY MEAN?

So now you know the theory of the OneStep transformation, but what really happens when you
then attach this coordinate system to a job and survey the quarry with GPS? Basically GPS
measured coordinates now fit to the real world where the real world in this case is the existing
quarry grid system.

In our quarry, the control points of course do not fit perfectly together this can be seen when
examining the residuals of the matched points during the determination of the coordinate system
and is graphically shown below for position.

It could be said that the accuracy of GPS is actually too good for the quarry! But the aim was to
retain the original TPS grid system (including any errors) such that when any point is now
measured with GPS it would give the same coordinates as if it had been measured with TPS. This
is now the case.

So the OneStep transformation sounds wonderful! It can be used when the local ellipsoid and
projection is not known and can be used with control points where only the positions or heights of
the points are known. Why is it not used everywhere for every survey?

There is of course a limit. The main disadvantage of the OneStep transformation is that it is
limited to areas of about 10km square. This is because the WGS84 geodetic coordinates are
projected to the preliminary grid co-ordinates using a Transverse Mercator projection with a
scale of 1 with the central meridian passing through the centre of gravity of the common points. It

Page 5 of 5
Quick Reference Guide # 22 Global Survey Limited

is extremely unlikely that the control points in the grid system were also originally surveyed using
the same scale factor which results from the OneStep Transverse Mercator. Differences will
therefore quickly grow the further you are away from the centre of the common points. How big
can the errors grow if the area is extended?

This is very hard to answer and mainly depends on how quickly (also if) the scale factor of the
local points change within the area. Errors may sometimes quickly reach several centimetres. This
question will be looked at in more detail in a future newsletter.

REMEMBER
A OneStep transformation can be computed even when the local ellipsoid or projection is not
known. Common points can be matched in position and height, in position only or even in height
only. The OneStep transformation treats the heights and position components of points separately.
There is a limit over the area in which a OneStep transformation can be used this is due to the
distortions resulting from scale errors. The errors increase depending on the distance from the
centre of the common points.

4. THE FINAL TRANSFORMATION TYPE THE TWOSTEP

This covers the theory of the TwoStep transformation.

QUICK RECAP
THE CLASSIC 3D TRANSFORMATION
The advantage of the Classical 3D transformation is that it is the most rigorous transformation
type - it is a similarity transformation, which keeps the full geometrical information.

The disadvantage is that that knowledge of the local ellipsoid and the map projection is required
and all common points have to be known in position and height.

THE ONESTEP TRANSFORMATION

The advantage of the OneStep coordinate system is that it is not necessary to know the local
ellipsoid and projection, which makes it ideal to use in areas which use a completely height.

The disadvantage of the OneStep transformation is that it is limited to areas of about 10km
square.arbitrary coordinate system. Additionally, it is possible to use common points for which the
coordinates are known in only position or

THE TWOSTEP TRANSFORMATION

The TwoStep transformation combines the advantages of the two approaches. It allows common points to
be used which are known only in position or height, but is not restricted to smaller areas. However, as for
the Classic 3D transformation, it is necessary to know the local ellipsoid and map projection.

a. HOW DOES THE TWOSTEP WORK?

As the name would suggest, there are two steps in converting coordinates from WGS84 to local
coordinates. In the first step the WGS84 Cartesian coordinates are shifted closely to the local
datum using a given Classical 3D Pre- Transformation to give local Cartesian coordinates.
(This pre-transformation is entered by the user - see later).

The local Cartesian coordinates are then converted to local geodetic coordinates using the known
local ellipsoid and then converted to preliminary grid, but unlike the OneStep transformation
which uses an arbitrary Transverse Mercator projection, the TwoStep transformation uses the true
map projection on which the local points are based. So far this is basically the same as converting
coordinates from WGS84 to local grid. So far we have converted our GPS measured points to
local grid we can now match these grid coordinates to the known grid coordinates of the local
control points this is the second step of the TwoStep coordinate system.

Page 6 of 6
Quick Reference Guide # 22 Global Survey Limited

The two sets of grid coordinates are matched in exactly the same way as with the OneStep
transformation. The final part of the transformation is therefore a 2D positional transformation
and height shift. The full TwoStep coordinate conversion process is shown in the following
diagram.

b. WHY USE A TWOSTEP?

This probably sounds all very complicated, sowhy use a TwoStep transformation?

Compared to the OneStep transformation, the first step of the TwoStep transformation (when the
WGS84 coordinates are converted to the preliminary grid) avoids any distortions due to the fact
that the preliminary grid co-ordinates are built on a different ellipsoid to the local points. Even
more importantly the influence of the scale factor of the map projection is now taken into account
before the final 2D transformation is done. For these reasons the transformation will fit much
better over larger areas than a OneStep transformation. The height part is independent of the
position transformation and is identical to the approach taken for the OneStep.

Note, when using a TwoStep transformation it is not possible to see the preliminary grid
coordinates either in LGO or on the System1200 instruments. Only the final local grid coordinates
are shown. This is correct since the preliminary grid coordinates have no practical use.

c. WHICH PRE-TRANSFORMATION TO USE?

So far, we have only briefly mentioned the pretransformation as shown in the diagram above,
this is the 7 parameter transformation which is used to convert from WGS84 Cartesian coordinates
to local Cartesian coordinates. The actual parameters may be known and can then simply be
entered as a transformation on both System1200 instruments and LGO.

However, even if transformation parameters are not known it is still possible to use the TwoStep
transformation. Create and select anull transformation (zero shifts, rotations and scale) this
then gives you the advantage from the benefits of using the TwoStep transformation!

REMEMBER
The main advantage of the TwoStep transformation is that unlike the OneStep transformation, it is
not limited to smaller areas and common points can be matched in position and height, in
position only or height only to compute the transformation. It is necessary to know the local
ellipsoid and projection and also necessary to select a pretransformation to use (although this
may be a null transformation).

Page 7 of 7