geomatics / surveying iii course: module 1 refraction, module 2 heighting
TRANSCRIPT
≈ 30 Lectures , some Assignments
Recommended Texts:
Electronic Surveying instruments : Reuger
Hydrography for the Surveyor and Engineer : Ingham (rev Abbot)
IHO Manual of Hydrography
+ many other texts
APG3017D
SURVEYING III
Assessment
Assignments 20%
Refraction in Heighting - tut
Precise Levelling - tut
Hydrographic Survey - site
visit
Tests 20%
Exam end of year 60%
Why Surveying III?
Observations in the real world:
- Reduce to plane/ellipsoid
- Not in a vacuum – refraction
- Gravity
Detail of instrumentation, errors, operations etc.
Offshore
Advanced instruments – eg gyrotheodolites
1. Theory of Atmospheric Refraction
2. Heighting
3. Measurement with Electronic Theodolites
4. Electronic Distance Measurement
5. Introduction to Hydrographic Surveying
6. Specialised Instrumentation and
Techniques
Outline of Modules
Retardation of signals
Curvature of light path
Refraction in GPS .... Surveying II
Module 1: Refraction5
RETARDING – affects distance measurement
BENDING – affects direction and distance
measurement
Electromagnetic Spectrum:
Visible and infrared: 0.5mm < l < 1.0mm
Microwave: 5mm < l < 100mm
cf e.g. for 100mm, f 3GHz
Atmospheric Refraction
Electromagnetic spectrum
from: http://www.lbl.gov/MicroWorlds/ALSTool/EMSpec/EMSpec2.html
EDM
InfraredGNSS
Reflectorless
Retardation of EM radiation
Velocity of light in a vacuum: 299792.458km.s-
1 cRefractive Index: n
v
6Modulus of Refraction: N n 1 10
If n = 1.000273 , then N = 273
Fermat's principle:
EM radiation follows the path that takes the
least time -1 n 1
dt min , or: ds ds n ds minv c c
Because n varies along the path, the optical
path is not the geometrical path.
n is a function of air density, which is in turn
a function of pressure, temperature &
humidity
Visible & Infrared:
Modulus of refraction is a function of wavelength
for group velocity. For standard temperature
(To=273.15K) and pressure (Po=1013.25mbar):
0 2 4
4.8864 0.068N 287.604
For other atmospheric
conditions:
o
d d
P eN 0.269578N 11.27
T T
Humidity:
Relationship to Relative Humidity
(RH):
d w we E 0.00066P t t 1 0.00115t mbars
w
24509.73
TE 10 mbars
a
eRH 100
E
d
24509.73
T
aE 10 mbars
Microwave:
No dispersive effect in troposphere, hence no
frequency dependence:
6
2
d d d
P e eN 77.624 0.372 10 12.92
T T T
This is only valid for a ray path that is entirely within
the atmosphere. For measurements to bodies
outside the atmosphere (GPS, VLBI, SLR) a different
approach is followed.
Magnitude of Effects:
Parameter Effect on
Visible/Infrared
Effect on
Microwave
1mbar change
in air pressure
0.3ppm 0.3ppm
1° change in
temperature
1.0ppm 1.6ppm
1° change in td-
tw
0.05ppm 8.0ppm
Measurements to extra-
terrestrial bodies: GNSS, VLBI use microwaves (1.2GHz to
8GHz)
SLR uses visible laser
Approach is to model the Earth's atmospheric
as spherical layers, and to compute the
magnitude of the retarding effect in units of
metres.
Correction to range computed using c is given
by:
Microwave: -2.3secq metres
Laser: -2.5secq metres
(both a function of P, T, e and valid for troposphere
only)
Curvature of Light Path:
Affects:
Trigonometrical heighting
Levelling
EDM
Curvature in trigonometrical
heighting
P
Q
z dz
z ds dh
1 dz
ds
dh
sin 90 zds
dhds
cos z
Curvature:
Now:
Hence:
Curvature - 2:
Substituting for ds and dz in the expression for
curvature:
Snell's Law: nsinz = constant
Differentiating: ncosz.dz + sinz.dn =
0
Hence:dn sin z
dzn cos z
1 1 dnsin z
n dh
Now we need to find an expression for the vertical
gradient of the refractive index n
Curvature - 3:
Using No = 293
and
Remembering that n = 1 + N.10-6, we can use the formula for N
to get:
d
dP P0.0342
dh T
6 d0 2
d d
dTdn 1 dP P0.26943 N 10
dh T dh T dh
We then get:
5 d
2
d
dT1 P7.88 10 0.0342 sin z
T dh
S-s
dzdℓ
L
s
S
P
Q
W
Total refraction angle:
L
SW L d
d S s dz
dsdz
But, from the definition of
curvature:
Hence
:
dsd S s
1 S sds
S
W
and:
Total refraction angle - 2:
5 d
2
d
dT1 P7.88 10 0.0342 sin z
T dh
ddT
dh
If the total refraction angle is:
we just need to substitute for P, Td and to get W
1 S sds
S
W
This is not practical, and we need to make further
approximations
and:
Total refraction angle - 3:
Sk (k generally taken as 0.13)
2RW
S
2rW
Rk
r
If the total refraction angle is:
If we use the coefficient of refraction,
1 S sds
S
W
Then:
we may further approximate this by setting the radius of
curvature s to be a constant – r. Then:
(R = radius of Earth)
Curvature in levelling:
5 d
2
d
dT1 P7.88 10 0.0342 sin z but z 90
T dh
b f
eb ef
Provided the temperature gradient is the same, eb = ef and the
refraction error cancels with equal sight lengths
Curvature in levelling - 2:
On a hill, the temperature gradient is steeper close to the ground:
ef > eb , hence (b – f) is too small and hill appears to be too low
Curvature in levelling - 3:
The correction can be modelled as a function of the vertical derivative
of the temperature gradient:
Here:
22
2
d Ts dn de e e f bf b 6 dT dh
d
5
2
d d
dn P7.88 10
dT T
2
d
2
d Tand
dh
can be deduced from temperature measurements on the
staves
Curvature in EDM:
Major effect of refraction on EDM is retardation
But curvature of light path also causes a change
in path length:
desired path length is d1 , but measured path length
is do
d
d
P
Q0
1
Curvature in EDM - 2:
Approximating the ray path by an arc of a circle:
Using a series expansion:r
r
2d1 2
d1
q q
do
1d 2r sin q
oo
dd 2 r
2r q q
o1
d d 2r sin
2r
3
o1 o 2
dd d
24r
3 2
o1 o 2
d .kRSubstituting r , we get d d
k 24R
Curvature in EDM - 3:
Optical coefficient of refraction: k = 0.13
Microwave coefficient of refraction: k = 0.25
The EDM curvature is flatter than that of the Earth, so the ray passes
through different layers of the atmosphere:
23
0 2
k kd ' d
12R
23
0 2
2k kd d
24R
Correction: Combined:
Specifications for height networks
Designing level networks
Selecting a heighting datum and system
Precise levelling
Accuracy estimates for levelling and adjustment
Special techniques of heighting
Module 2: Heighting28
Heighting
Why do we need height networks?
Control for mapping
Control for engineering projects
Deformation measurements
Reduction of distances to the ellipsoid
Reduction of gravity observations to the geoid
Study of variations in Mean Sea Level
Specifications
Order Relative Accuracy Procedure
First 0.5 - 1.0mm per √km geodetic level with parallel plate micrometer or
digital level; invar staves; double run levelling
Second 1.0 - 3.0mm per √km as above, sometimes with only single run
levelling
Third 3.0 - 8.0mm per √km ordinary, digital or geodetic level (no
micrometer); single run levelling
Fourth 8.0 - 40mm per √km ordinary level or trigonometrical height traverse
or GPS heighting
Level network design
nodal point
tide gauge
Closed loops
Connection to tide gauges
Route based upon convenience and economic
need
South African levelling
networks
Vertical datum
Ideally – the geoid. Practically – MSL, measured using tide
gauges. Use a single tide gauge – possible bias wrt geoid,
discrepancies at other gauges
Use all tide gauges and force levelling network to fit –
distortions?
Use all tide gauges, and allow levelling network to "float" to
best fitSouth African levelling datum – Land Levelling Datum
(LLD): Forced fit to four tide gauges: CPT, PLZ, ELN, DBN
15cm to 20cm below current MSL
Height system
Orthometric heights:
Normal heights:
Spheroidal orthometric
heights:
Dynamic heights:
dnH g
g
dnH g
dnH
dnH g
G
NB: these formulae are NOT exact
Benchmarks
Fundamental benchmarks: 50km to 100km spacing;
on bedrock; below ground; with reference
benchmarks
Main benchmarks: 8km to 12km spacing; on
bedrock where possible; with reference
benchmarks
Ordinary benchmarks: 1km to 3km spacing; on
bedrock, in bridges, culverts or concrete
foundations
Tide Gauge benchmarks (TGBM): adjacent to tide
Precise levelling - levels
High magnification (30x to 50x); high sensitivity (
< 0.5")
High accuracy (under ideal conditions: 0.3mm per
√km;
practically: 1mm per √km)
Optical levels with parallel plate micrometers
(obsolete)
Digital levels with bar coded staves
Pairs of invar staves
Digital Levels
Use of barcoded staves
Image of portion of staff captured by CCD
inside the level
Image matching enables precise
determination of position of crosshair on staff
Can also determine distance to staff, to a few
cms
Quick, accurate (no reading error), range of
up to 100m
Needs good lighting conditions
Automatic Compensators
Sources of
Error Earth curvature
Collimation
Overcompensation
Symmetrical
refraction
Change in height of
collimation axis due
to change of focus
Mitigation
Equal length foresight &
backsight
Equal length backsight &
foresight
Provided instrument is level,
equal length backsight &
foresight
Equal length foresight &
backsight
Equal length foresight &
backsight will eliminate the
need to change focus
Sources of Error & their Mitigation
Refraction changing uniformly with time:
Use two staves (A & B) and observe A, then B, then B, then A;
Compute two height differences and take mean.
Steadily sinking (or rising) instrument or staff (single setup):
Use two staves (A & B) and observe A, then B, then B, then A;
Compute two height differences and take mean.
Maladjustment of circular bubble:
Level up pointing in alternate directions – i.e always pointing
towards same staff (e.g. staff A). Only works for even number
of setups and with each pair of setups having similar
sightlengths.
Non-verticality of staff:
Use even number of setups, with each pair of setups having
similar sightlengths.
Sources of Error & their Mitigation
Heating of the instrument (causes changes in collimation):
Shield with an umbrella.
Index error of staves:
Use same staff on every benchmark – i.e. even number of
setups.
Scale error in staves:
Calibrate staves (calibrate entire system for digital level).
Non-symmetric refraction:
Never observe below 0.5m on the staff
Use short sight lengths
Measure temperature gradient and apply correction
Earths Magnetic Field
Daily checking of collimation error
(adjust if greater than 10")
Weekly checking of circular bubble
Weekly checking of staff bubbles
Annual calibration of staves
Observation procedures: Checks
Observation procedures
Equal sight lengths for forward and back (within 1m)
Short sight lengths ( < 60m, but < 30m on long shallow
inclines )
Observe higher than 0.5m on the staff
Same staff (A) on benchmarks (equal number of
setups)
Level up while pointing to the same staff (A) every time
Always observe the same staff (A) first
Observe staves in the sequence: ABBA
(check that agreement is better than 0.5mm)
Observation procedures - 2
Keep instrument shaded
Use support rods for staves (do not lean on them), and
place footplates securely
Level at the time of day when shimmer is least (early
morning)
Level each section in both directions, at different times
Corrections to levelling
Scale correction:
Requires determination of scale factor for staves –
Conventional invar staves are calibrated using a
laser interferometer (see later) and either an
average scale factor or a scale factor per
graduation are applied.
Bar-coded staves must be calibrated together with
the digital level, and the process requires (in
addition to a laser interferometer) a special
carriage to move the staff vertically through the
field of view of the level.
Corrections to levelling - 2
Temperature correction:
Not normally applied, as invar is not temperature
sensitive.
Non-symmetric refraction:
Requires measurement or modelling of temperature
gradient and its derivative at every setup. Seldom
done.
22
2
d Ts dn de e e f bf b 6 dT dh
d
5
2
d d
dn Pwhere: 7.88 10
dT T
Corrections to levelling
- 3 Earth Tides:
Attraction of sun and moon causes small changes in
direction of gravity and hence in direction of collimation
axis (effect is ± 0.02").
Generally ignored (also indirect effect of ocean
loading).
Few cm over 1000km leveling
Influence of Gravity:
Spatial variations in gravity cause changes in shape of
equipotential surfaces. Derivation is covered in
Geodesy course. Effect is of the order of cms over
100km.
Error propagation in
levelling
Assuming random tilt error a and equal sight lengths s, then the error in
the backsight is b = s.a and the error in the foresight is f = s.a. The error in
the height difference is:
For a line of levelling of length L:
s s
b f
a
2 2 2 2 2
h b f 2s a
1 2 3 n
2 2 2 2 2
L h h h h....
If all sight lengths are the same ( = s) then:
and:
1 2 3 n
2 2 2 2
h h h h.....
2 2 2
L n.2s a
But if all sight lengths are equal to
s:
Ln
2s
Therefore:2 2 2 2
L
L.2s L.s.
2sa a
L s. . La Hence:
L 1 L
(error in a section of levelling is proportional to the square root of the length of the
section)
Error propagation in levelling -
2
1 k s a
Precision estimates
1 (standard deviation per root km) can be determined as follows:
Manufacturer's specifications: generally too
optimistic
From comparison of forward and back levelling over
sections:
From loop closures:
2
nf b2 i
1
i 1 i
n n1
4n L
2n2 i1
i 1 i
1 h
n L
Precision estimates - 2
1 derived from loop closures is generally larger
than that from comparison of forward and back
levelling.
This implies that errors do not propagate strictly
according to √L
If L is larger, then s1 is larger
This further implies that: where a >
0.5 L 1 La
Precision estimates - 3
Alternative model (Vignal), assumes that errors
propagate randomly, but with different values for s1,
depending upon L:
L < 5km, 1 = h
5km < L < Z, 1 = tL
L > Z, 1 = t
In South Africa: h = 0.6mm, Z = 25km, t = 1.1mm (optimistic)
Level network adjustment
Least squares adjustment by parametric or
condition equations
Free network or minimum constraint or constrained
Weighting of observations:
weight inversely proportional
to distance:
weight inversely proportional
to variance:
variance based upon:
L
kp
L
L 2
L
kp
2
L L 1 L
L 1 La
or Vignal's approach
Valley crossing
Reciprocal Levelling:
At A: HA = (bA + eS + rS +cS) - (fA + eL+S + rL+S + cL+S)
At B: HB = (bB + eL+S + rL+S + cL+S) - (fB + eS + rS + cS)
Reciprocal levelling
Collimation error can be eliminated if the two
levels are swopped and measurements repeated
Collimation error can be eliminated if a single
level is used, provided refraction conditions do
not change between setups
Or, specialised pairs of levels can be used, with
autocollimation
from: F. Deumlich – Surveying Instruments
Reciprocal levelling
Need special targets, as staff graduations cannot
be resolved beyond about 100m range:
Or, with digital level, use enlarged version of
bar code
from: http://www.fig.net/pub/athens/papers/pdf/ts_03_5_takalo_rouhiainen_ppt.pdf
Reciprocal zenith angles
Where the height differences are more than a few
metres, or the range exceeds 2km, reciprocal
levelling cannot be used.
Reciprocal zenith angles, with simultaneous
observations, provide an alternative method,
albeit with a loss of accuracy:
Reciprocal zenith angles
2 2B AB A A A A B
H H 1 kAt A: H H 1 . s.cot z .s cosec z i m
2R 2R
2 2B AA B B B B A
H H 1 kAt B: H H 1 . s.cot z .s cosec z i m
2R 2R
2 2 2B AB A A B A B A B B A
Subtracting:
H H 1 k 2 H H 1 s cot z cot z s cosec z cosec z i m i m
2R 2R
Reciprocal zenith angles
2 2 2B AB A A B A B A B B A
H H 1 k2 H H 1 s cot z cot z s cosec z cosec z i m i m
2R 2R
2 2
A B A A B BNow: cosec z cosec z and m i m i
B AB A A B A B
1 H HHence: H H 1 . s. cot z cot z i i
2 2R
(basic equation of simultaneous reciprocal zenith angles)
Note that the distance s is the distance on the ellipsoid
Error propagation
2 42 2 2 2 2
H s k4 2
s scot z
sin z 4R a
If we differentiate the trigonometrical heighting equation, we
get the model for error propagation:
Provided that z is close to 90°, the effect of a distance error on the
height difference is small. E.g., for z = 80° and s = 10mm, the effect is
2mm. The effect of errors in the coefficient of refraction is much larger.
Putting estimates of s = 10mm, a = 2" and k = 0.04, we get the
simplified expression (assuming z equal to 80°):
2 2 4
H
H
4 100s 9.9s
in units of mm, s in units of km
Error propagation - 2
2 2 4
H 4 94s 9.9s
Distance Effect of a Effect of k Combined
effect
1km 10mm 3mm 11mm
2km 20mm 12mm 24mm
3km 30mm 28mm 41mm
5km 50mm 79mm 93mm
Error propagation - 3
2 2 4
H 4 94s 9.9s
0
200
400
600
800
1000
1200
1400
0 2 4 6 8 10 12 14 16 18 20
Distance (kms)
Err
or
(mm
s)
Hydrostatic levelling
Use undisturbed water to define a level surface
Water contained in a flexible tube, with vertical
glass tubes with scales at each end:
Connected to benchmarks at each end
Corrections for temperature and pressure
Hydrostatic levelling - 2
Time comsuming, but can achieve high accuracy
over long distances (a few mm over 10km)
Used to cross water bodies (pipe laid under water)
– e.g. in Denmark and Holland
On a smaller scale (20m to 100m) used for
deformation monitoring in structures (see
APG4005F)
Hydrodynamic levelling
Uses the open sea as a "level"
Tide gauges at each end eliminate/reduce the
effects of waves, seiches and tides
Effects of currents, winds and differential air
pressure must be modelled and corrected
Used to connect Uk and French level networks
across the English Channel – estimated accuracy
of 1.5cm over 70km
Trigonometrical height traversing
An extension of the simultaneous reciprocal
vertical angle approach, using two electronic
tacheometers:
z
z
A
B
S
AB A B
SH cos z cos z
2
S is the slope distance, and DHAB is the height difference
between the collimation axes of the two telescopes
Trigonometrical height traversing
Connection to benchmark involves a single staff
and reading of zenith angles to a minimum of two
graduation lines:
2 1 1 2
1 2
cot z cot zh
cot z cot z
z1
z2
h
ℓ1
ℓ2
Trigonometrical height traversing
Residual refraction effects (k not the same at
both ends of the line) means that sight lengths
should be kept to less than 250m
Accuracy of 3 – 5mm per km can be achieved
Faster than conventional levelling: 2km per hour
vs 0.5km per hour
More flexible – can traverse rough, steep terrain
Motorised levelling
Can be used to speed up levelling (conventional,
digital, or trigonometrical height traversing).
Requires up to three vehicles, which must be
specially modified.
Can improve speed of conventional levelling by
up to 50%