30 july 2004usm 2004 upper pearl survey1 vertical uncertainty total propagated error –error budget...
TRANSCRIPT
30 July 2004 USM 2004 Upper Pearl Survey 1
Vertical Uncertainty
• Total Propagated Error– Error budget using Excel spreadsheets
created for NAVOCEANO by Rob Hare– TPE for every sounding using CARIS HIPS
v5.4
• Cross check analyses using Fledermaus– Multi-beam vs. Multi-beam– Single-beam vs. Single-beam– Multi-beam vs. Single-beam
• PPK/Heights
30 July 2004 USM 2004 Upper Pearl Survey 2
Vertical Uncertainty
• Factors taken into account for error budget calculation– Depth variation: 1-25m– Sea state: 0– River bottom slope: 2o
– Water temperature: 28o C– Salinity: 0– Vessel roll/pitch < 1o
– Maximum swath angle for EM3000: 60o (113 out of 127 beams)
30 July 2004 USM 2004 Upper Pearl Survey 3
Vertical Uncertainty
• Sounding error budget– Static and dynamic draft measurement
uncertainties – Offset measurement uncertainties– Refraction errors– Sounding system limitations– Vessel motion measurement
uncertainties– Tide and vertical datum uncertainties– And…
30 July 2004 USM 2004 Upper Pearl Survey 4
Vertical Uncertainty
• Draft and Offset measurements
Standard Deviation/ RMS Values
Standard Deviation/ RMS Values
Offset Measurement (m)
0.005 -
Draft Measurement (m)
0.037
Dynamic Draft (m) 0.0130.035
SkiffBertram
Vertical Uncertainty (at 95% C.I.)
Vertical Uncertainty (at 95% C.I.)
0.010
0.069
-
0.072
0.025
Hare, R., Error budget analysis for NAVOCEANO hydrographic survey systems, MS, 2001.
30 July 2004 USM 2004 Upper Pearl Survey 5
Vertical Uncertainty
• Refraction
Standard Deviation/ RMS Values
Standard Deviation/ RMS Values
At deepest depth (~25m)
At shallowest depth (~3m)
At deepest depth (~25m)
At shallowest depth (~1m)
SV Measurement (m/s) 1.500 1.500
Temporal SV Variation (m/s)
2.000 2.000
Spatial SV Variation (m/s)
insignificant insignificant
0.030
Vertical Uncertainty (at 95% C.I.)
SkiffBertram
Vertical Uncertainty (at 95% C.I.)
0.090 0.040 insignificant
Post survey analysis.
30 July 2004 USM 2004 Upper Pearl Survey 6
Vertical Uncertainty
• Sounder accuracy and vessel motions
Standard Deviation/ RMS Values
Standard Deviation/ RMS Values
At deepest depth (~25m)
At shallowest depth (~3m)
At deepest depth (~25m)
At shallowest depth (~1m)
Sounding System Accuracy (m)
0.100 0.110 0.120 0.010 0.010 0.010
Heave (fixed) (m) 0.050 - - -
Heave (% of amplitude) 5.000 - - -
Roll (degrees) 0.020 0.030 insignificant - - -
Pitch (degrees) 0.020 insignificant insignificant - - -
SkiffBertram
Vertical Uncertainty (at 95% C.I.)
0.140 0.140
Vertical Uncertainty (at 95% C.I.)
Manufacturer specifications
30 July 2004 USM 2004 Upper Pearl Survey 7
Vertical Uncertainty
• Tide and sounding datum uncertainties (at 95% C.I.)
– Tide gauge measurements 0.02– Datum recovery 0.04– Use of old epoch 0.02– Tidal zoning 0.03
30 July 2004 USM 2004 Upper Pearl Survey 8
Vertical Uncertainty
At 1 mAt 25 m
• IHO S-44 Order-1 requirement : 0.5 m 0.596 m
a=0.5m
b=0.013d=1 to 25m
• Total uncertainty: 0.096 m 0.221 m
22 )( dba
+ DATUM TRANSFER UNCERTAINTY
30 July 2004 USM 2004 Upper Pearl Survey 9
Vertical Uncertainty
• Total Propagated Error– The Fourth Edition of S-44 [IHO, 1998] recommends
that: “All soundings should be attributed with a 95% statistical error estimate for both position and depth.”
– CARIS HIPS v5.4 uses the same approach as in previously mentioned Hare Error Model and propagates the error values to grid nodes of an uncertainty surface, using the actual vessel motion, depth and slope information instead of using average values.
– The same constant values used in spreadsheets were also used in HIPS.
30 July 2004 USM 2004 Upper Pearl Survey 10
Vertical Uncertainty
• Total Propagated Error– 1m grid resolution used for this survey
– The resulting grid then was exported to an ASCII file
– HIPS doesn’t have a device model for Knudsen 320 (only some multi-beam systems are supported) so TPE could only be calculated for the multi-beam survey
30 July 2004 USM 2004 Upper Pearl Survey 11
Vertical Uncertainty
• TPE
30 July 2004 USM 2004 Upper Pearl Survey 12
Vertical Uncertainty
• PPK/GPS Heights– PPK/GPS data was collected both on
Bertram and Skiff throughout the survey.• Alternative to classical tide measurement for
reducing the soundings to chart datum• Tie the soundings directly to the ellipsoid for
a seamless datum
– Noisy due to obstruction of trees, especially for Skiff which ran lines across the river.
30 July 2004 USM 2004 Upper Pearl Survey 13
Vertical Uncertainty
• PPK/GPS Heights needed filtering– Epochs that had height > 0.25m
filtered, then gaps interpolated,– Furthermore, epochs that had heights
> 1.5*mean(height) filtered iteratively until mean(height) < 1m
30 July 2004 USM 2004 Upper Pearl Survey 14
Vertical Uncertainty
• PPK/GPS Heights
23 June 2004, PPK/GPS Heights, Skiff
-35
-30
-25
-20
-15
-10
25000 30000 35000 40000 45000 50000 55000
Time (GPS seconds)
He
igh
t (m
)
Raw
Filtered
23 June 2004, PPK/GPS Heights, Skiff
-25
-24.5
-24
-23.5
-23
-22.5
-22
25000 30000 35000 40000 45000 50000 55000
Time (GPS seconds)
He
igh
t (m
)
Raw
Filtered
23 June 2004, PPK/GPS Heights, Skiff
-24
-23.95
-23.9
-23.85
-23.8
-23.75
-23.7
-23.65
-23.6
-23.55
-23.5
30000 31000 32000 33000 34000 35000 36000 37000 38000 39000 40000
Time (GPS seconds)
He
igh
t (m
)
Raw
Filtered
Mean : -23.424 mMedian : -23.768 mSt.Dev. : 1.773 m
Mean : -23.77 mMedian : -23.786 mSt.Dev. : 0.11 m
0.5m
30 July 2004 USM 2004 Upper Pearl Survey 15
Vertical Uncertainty
Range: 33 cm
Range: 5 cm
Range: 2 cm
23 June 2004, Bertram
30 July 2004 USM 2004 Upper Pearl Survey 16
Vertical Uncertainty
• Cross check analysis– Due to relatively big size of Bertram it wasn’t
possible to run systematic multibeam cross check lines,
– Main singlebeam lines served as check lines for multibeam data => ~750 cross check lines, Order-1 requirement satisfied!
– Fledermaus Cross Check Analysis tool used for the analyses,
– All soundings used for the reference DTM instead of a decimated shoal/deep biased sounding set
30 July 2004 USM 2004 Upper Pearl Survey 17
Vertical Uncertainty
• Multibeam vs.
Multibeam# of points compared 1859102Mean 0.003Median 0.005Std. Deviation 0.05695% 0.109Order 1 Error Limit 0.50 Order 1 - # Rejected 477(99.97% ACCEPTED)
Order 1 Survey ACCEPTED
30 July 2004 USM 2004 Upper Pearl Survey 18
Vertical Uncertainty
• Singlebeam vs.
Singlebeam# of points compared 10603Mean 0.016Median 0.023Std. Deviation 0.25395% 0.496Order 1 Error Limit 0.51Order 1 - # Rejected 457(99.96% ACCEPTED)
Order 1 Survey ACCEPTED
30 July 2004 USM 2004 Upper Pearl Survey 19
Vertical Uncertainty
• Multibeam vs.
Singlebeam# of points compared 716514Mean -0.013Median -0.048 Std. Deviation 0.25495% 0.498Order 1 Error Limit 0.51Order 1 - # Rejected 24375(96.6% ACCEPTED)
Order 1 Survey ACCEPTEDBUT…
30 July 2004 USM 2004 Upper Pearl Survey 20
Vertical Uncertainty
• Cross check summary (Estimated max. error (95% C.I.): 22 cm)
– MB vs. MB : 11 cm– SB vs. SB : 50 cm
• Initial cross check analysis between MB and SB datasets showed a mean offset of 25cm (SB deeper) due to– Bad channels of EM3000 transducer– Accepted Bertram as it is, no ground truth of soundings
• Did not perform bar-check on Bertram• Did not measure static draft
• This could be seen only after applying the tides because PPK/GPS heights were noisy.
30 July 2004 USM 2004 Upper Pearl Survey 21
Vertical Uncertainty
• On the other hand, Skiff was calibrated daily using bar check.
• Furthermore soundings from Skiff matched to lead line measurements within 10 cm (95%)
• As a result, mean difference between the two data sets was applied as static draft to MB data.
30 July 2004 USM 2004 Upper Pearl Survey 22
Vertical Uncertainty
• Lessons learned– Make sure that you thoroughly know
the equipment/software you use.– Use every possible way to ground
truth your soundings; a simple lead line measurement may be as important as a complicated patch test.
Processing: Horizontal Uncertainty
Debbie Mabey