3OURNALOF GEOPHYSICALRESEARCH VOLUME 65, No.

2 FEBRUARY 1960

SimultaneousAdjustmentof Triangulation and Trilateration

An Investigation of Tellurometer Lengths
L. A. GAL•.

Geodetic Survey of Canada, Ottawa

The present developmentof distance-measur- about (1) the reliability of the Tellurometer
ing instruments suitablefor geodeticsurveys when operated under Canadian field conditions
raisesthe problemof simultaneously
adjusting over a period of 3 or 4 monthswith no recourse
observedlengthsand angles.The basicmathe- to elaborate test equipment; (2) observational
matical toolsfor carryingout suchadjustments procedures suitable for various requirements;
accordingto the methodof least squareshave (3) methodsof applicationin conjunctionwith
existedfor many years.In recentyearsexcel- theodolite measurements.
lentmaterial,writtenin termsof matrixalgebra, It was not economicallyfeasibleto measure
has been publishedon the generaltheory of all the lengths in the triangulation, but the
least squaresadjustmentby such authorsas original aim of measuringat least all the sides
Marchant, Bjerhammar,Wolf, and Gotthardt. of a chain of triangleswasachieved(seeFig. 2).
In some of this literature the approachhas Field tests of observations included one for
beento proceedfrom the generalproblemto consistencyof the sets of measurementsof each
particular casesas opposedto the synthesis line and one in which the measured .lengths
of particularcasesinto a generalmethod.For were checked against the observedangles by
example, the condition equation method and means of the sine law. For the latter test we
the indirect observationequation method are assumed a standard of • = +0.85" for the
treated as specialcasesof the generalmethod. angles and a standard relative error •rs/s =
The type of equationsand the method of q-(I/200,000) for the observedlengths. The
solvingnormalequations usedforanyadjustment discrepanciesAF obtained by using observed
dependon (1) the natureof the problem;(2) the values in the relation
computational facilitiesavailable;(3) the per-
AF = asinB-- bsin A
sonalpreferenceof the computer.
If we have a surplusof observations, and if wherea and b are two sidesof a triangle and A
we accept the method of least squares,our and B are the respectiveoppositeangles,were
main problemis not the development of mathe- comparedwith
matical formulasor toolsfor the adjustmentof
data; it is the old oneof assigning appropriate •F = +a sin B
weightsto the observedquantities.
In an effortto contributetowarda satisfactory
ß (cot +
answerto this problemasit concernsobservations The expressionfor •F is derived from the sine
of directionsand measurementof lengthswith law using the law of propagation of errors.
the Tellurometer,a field party of the Geodetic With the valuesof • and •8/s indicatedabove,
Survey of Canada measuredthirty-five lines 78 per cent of 67 tests fell within the limits
of an east-westarc of first-ordertriangulation given by •r.
extending 137 km westward from Kenora, As part of the investigation, the following
Ontario,to Whitemouth,Manitoba. The lengths five adjustments,in which condition equations
range between 13 and 43 km with a mean of were used, were carried out: (1) corrections
22 kin. Two lines,9-11 and 25-26, weremeasured applied to observeddirectionsonly (Fig. 1);
with the Geodimeter (see Fig. 1). From this (2) correctionsappliedto all observeddirections
operation we also hoped to gain information and lengths (Fig. 1); (3) correctionsapplied
472
SYMPOSIUM ON ELECTRONIC DISTANCE MEASURING 473
474 L. A. GALE

TABLE 1. Some Data and Results of Five Adjustments

Adjustment 1 2 3 4 5

No. of stations to be fixed 16 16 13 13 13

No. of conditions 37 70 39 19 20
No. of directions observed 86 86 54 54 28
No. of lengthsobserved 33 25 5 25
Standard error (weight unity), seconds 0.516 0.519 0.531 0.511 0.532
Averagerelative correctionto lengths,ppm 2.9 3.5 5.4 2.6
Maximum relative correction to
lengths,ppm 12.4 13.0 8.4 12.0
No. of relative correctionsto lengths
greater than 10 ppm 1 2 0 1
Latitude, terminal station 49ø56'55.124" 55.127" 55.122" 55.125" 55.118"
Longitude,terminal station 95ø58'52.442" 52.472" 52.475" 52.461" 52.478"
Terminal azimuth 352ø21'38.992" 39.104" 39.267" 39.685" 39.079"

(•: Adjustment4(k)
0,368m

Fig. 3. Positions
of terminalstationsrelativeto adjustment1.

to all observeddirectionsand lengthsof a chain sideof the chain(Fig. 2); (5) corrections
applied
appliedto to all the observedsidesof the chainof triangles
of triangles(Fig. 2); (4) corrections
the observeddirectionsof the chain of triangles and the directionsfrom stationsalong the south
and to five observedlengthsalong the south sideof the chain (Fig. 2).
 SYMPOSIUM ON ELECTRONIC DISTANCE MEASURING 475

TABLE 2. Lambda Values see that the inclusion of such lengths in the
subsequent adjustmentsdid not appreciablyalter
Adjustment X, meters the standard error of a direction.

2(x) -0.070
A comparisonof the adjusted lengths (not
3(x ) - o. 068 given here) shows that all the final lengths
4(X) --0.072 resulting from adjustment 2 are greater than
the corresponding lengthsof adjustment1, one
lengthof 3 is lessthan the corresponding
lengths
In all the adjustments,two Geodimeterlines, of adjustment 1, and only three lengths of
one at each end of the arc, were treated as adjustments4 and 5 are shorter than the cor-
fixed base lines. respondinglengthsof adjustment 1. This syste-
In the adjustments, correctionswere applied matic discrepancyis also reflectedin the longi-
to the logarithmsof the lengths,the units being tudes of the terminal station (see Table 1),
M ds/s X 106,with M = log•0e. when we remember that we are dealing with
Since we assumed a constant relative error an east-west arc of which the east end is fixed
for the measuredlengths, the application of in position (see also Fig. 3). We also notice
logarithmiccorrectionsallowed us to use the the tendency toward increasein the length of
same weight for all the logarithmicdistances. the triangulation schemeas the ratio of the
I understand that the U.S. Coast and Geodetic observed lengths with respect to observed
Survey has used ds/(s sin 1") as the unit of directions increases.
correction, which for weighting purposeshas On the assumptionthat a constantsystematic
the sameproperty as the logarithmiccorrection error had occurredin all the length measure-
when a constant relative error is assumed. ments, we incorporateda constantcorrectionX
On the basis of field tests and some small for each length in the equationsof adjustments
preliminaryadjustments,we used a weight 2, 3, and 4; i.e., each observedlength s• was
ratio of 20 to i for directionswith respectto subjectedto a correctionds• and a correctionX
logarithmicdistances.Assuming:t=0.5" for the which was common to all observedlengths in
the standarderror of a direction,this weighting any one adjustment.The 'lambda' adjustments
implies a standard relative error in length of correspondingto adjustments 2, 3, and 4 are
approximatelyq- (1/194,000). designated 2(X), 3(X), and 4(•). The values
A perusalof Table I showsthat the standard obtained for X are shown in Table 2.
errors (weight unity) are approximatelyequal With referenceto Table 3, which showssome
with a value of :t=0.5". In these adjustments of the data and results of the lambda adjust-
this standard error represents the standard ments, we note that the values of the standard
error of a direction. If we recall that no Tellurom- error of an observationof weight unity are all
eter lengths were used in adjustment 1, we approximatelyq-0.5" and are in closeagreement

TABLE 3. Data on Lambda Adjustments

2(X) 3(X) 4(X)

No. of stations to be fixed 16 13 13

No. of conditions 70 39 19
No. of directions observed 86 54 54
No. of lengths observed 33 25 5
Standard error (weight unity), seconds -v0.5O8 4-0.519 -v0.5O5
Averagerelative correctionto lengths,ppm 3.3 3.4 5.3
Maximum relative correctionto length, ppm 11.5 12.0 9.2
No. of relative correctionsgreater than 10 ppm 1 2 0
X, meters -0.070 -0.068 -0.070
Latitude, terminal station 49ø56'55.126" 55.120" 55.122"
Longitude,terminal station 95058'52.453" 52.457" 52.447"
Terminal azimuth 352ø21'39.104" 39.226" 39.539"
476 L.A. GALE

0.539
m• 23

o.?41m

o 588 m

o.'r:•
•):
Adjustment ß

(•) = Adjustment
• = Adjustment
4(•1 _

Fig. 4. Positionsof terminalstationsrelative to adjustment1.

with thoseof the previousadjustments shownin ratio of lengthsto directions is 33:86 and the
Table 1. There are alsono significantdiscrepan- lineardiscrepancy in positionof terminalstation
cies between the average relative corrections 25 is 0.588 meter with respectto adjustment
of the first adjustments2, 3, and 4 and the 1, and the corresponding valuesfor adjustments
corresponding lambdaadjustments. The shorter 4 and 5 are 5: 54, 0.384meter,and 25: 28, 0.750
lengthsof the lambdaadjustments areindicated meter. This apparentovermeasurement could
bythelongitudes fortheterminal station(no.25) be a function of a number of factors, including
shown in Table 3. A comparisonof Figures 3 the instrument used, the terrain, and meteoro-
and 4 showsthe effect of the lambda adjustments logicalcorrections. It might also be ascribed
on the positionsof stationsin the terminal to the directionobservations', but, because of
figureof triangulation.Figure4 alsoindicates the lengthcontrol
at eachend of the arc, the
that the tendencytowardgreaterlengthswhen angle measurements probablycontributevery
observedlengthswere includedin the adjust- little to this tendency.No conclusions can be
ments was not completelyeliminatedby the drawn from a comparisonof Tellurometer
lambda adjustments. lengthsand Geodimeterlengthsfor the two
The results of these adjustments show a lines9-11, and25-26.For line9-11theTelluromo
definite trend toward greater lengths in the eter length was shorter than the Geodimeter
particulartriangulationschemeas the ratio lengthby 0.9 ppm, and for line 25-26 the
of the observed directions Tellurometermeasurement
lengthsto the observed was longerthan the
is increased.For example,in adjustment2 the Geodime[ermeasurementby 1.4 ppm.
 SYMPOSIUMON ELECTRONICDISTANCEMEASURING 477

A comparisonof •he standard errors of an ments, we cannot rely on this instrument to

observationof weight unity indicates,we believe, provide basic length control for first-order tri-
that the weight ratio of directionswith respect angulation schemes.It is our hope that more
to lengthsusedin theseadjustmentsis reasonably experienceand more data obtainedfrom opera-
satisfactory. tions in differentareasby geodeticorganizations
With our present lack of quantitative know- may enable us to assessmore accurately the
ledge about factors that may significantly in- factorsaffectingmeasurements with theTellurom-
fluence the accuracy of Tellurometer measure- eter.