Detail surveying with networkRTK an accuracy research

1
Detail surveying with network-RTK- an accuracy
research

• Diploma thesis
• Robert Odolinski Johan Sunna

2009-04-24, Robert Odolinski Johan Sunna
2
Background and objectives

• Background material
• - proposed control methods for detail surveying
  with network RTK
• - theoretical assumptions of the expected
  accuracy levels
• Objectives
• - estimate if the control methods are
  practically useful
• - can the theoretical assumptions of accuracy
  levels be adjusted to standardised tolerances for
  such control methods?

3
Control methodRevisit with network RTK

• The control method refers to horizontal and
  vertical level
• New fixed ambiguity solution between each
  control point
• At least 20 minutes between each new surveying
  loop (and revisit)
• Surveying in the morning and in the afternoon
  for some revisits

4
Control methodSeparate control with
conventional technique

• The detail points were measured with a total
  station (free stationing) in a local system
• The measured detail points were then compared to
  the network RTK points

5
Proposed accuracy levels
The following accuracy levels are within a
confidence level of at least 95 based on
estimated standard errors

• Revisit network RTK integrated with the
  production measurements or processed alone
• -horizontal deviation ? 55 mm
• -deviation in height ? 100 mm
• Control with total station of point originally
  measured with network RTK
• -unitary transformation unit-weight standard
  error (per coordinate) ? 15 mm
• -translation in height standard deviation ? 35
  mm
• -deviation in distance ? 40 mm
• -deviation in height difference ? 100 mm
• -deviation in height (after estimating a shift)
  ? 70 mm

6
Surveying areas (4 areas)- Teknikparken, Bomhus
and Stigslund
7
Results from area 1-3 (Teknikparken and
Bomhus)- modification of the proposed accuracy
levels
8
Separate study of the centering standard error-
concerning the prism pole for the total station
The centering standard error was estimated to
approximately 14 mm
9
Proposal of new accuracy levels- based on area
1-3 and assumptions

• Prerequisites
• The new accuracy levels are within a confidence
  level of at least 95 based on estimated standard
  errors.
• The estimated standard errors are in addition
  based on results from area 1-3 and from the
  following assumptions
• horizontal standard error for the total station
  of 5 mm
• centering standard error for the rover pole of
  network RTK of
• 14 mm
• tripod used for the prism pole for the total
  station
• ? centering standard error 0 mm

10
Proposal of new accuracy levels, cont. - based
on area 1-3 and assumptions(note that
parenthesis the old proposed accuracy levels)

• Revisit network RTK integrated with the
  production measurements or processed alone
• -horizontal deviation ? 60 mm (55 mm)
• -deviation in height ? 60 mm (100 mm)
• Control with total station of point originally
  measured with network RTK
• -unitary transformation unit-weight standard
  error
• (per coordinate) ? 20 mm (15 mm)
• -deviation in distance ? 45 mm (40 mm)
• -translation in height standard deviation ? 20
  mm (35 mm)
• -deviation in height difference ? 50 mm (100
  mm)
• -deviation in height (after estimating a shift)
  ? 40 mm (70 mm)

11
Results from area 4 (Stigslund)- independent
test of the new accuracy levels
12
Results from area 4, cont.(Stigslund) -
independent test of the new accuracy levels
13
Discussion

• We consider that the control method revisit with
  network RTK is practically useful and efficient,
  because of the integration possibilities with the
  production measurements.
• Separate control with conventional technique
  (total station/leveling instrument) is not
  practically useful to integrate with the
  production measurements, however it is more
  useful when ordered by e.g. a contractor.
• The centering standard errors have more influence
  of the horizontal control than we expected.
• This study achieved a horizontal standard error
  of 10 mm and standard error in height of 15 mm,
  which is comparable with similar studies carried
  out lately (e.g. Edwards et al. 2008). To achieve
  the horizontal accuracy, a tripod of some kind is
  necessary to minimize the influence from the
  centering errors.

14
Discussion, cont.

• It is important to mention that this is an
  accuracy research in Gävle (autumn 2008), and it
  should not be considered as a standard for
  control of detail surveying with network RTK. The
  conditions in this study are considered to be
  very favorable.
• The accuracy of network RTK can be influenced by
  e.g.
• - solar cycle sunspot (maximum in year 2011)
• - distance to reference stations (SWEPOS)
• - location inside our outside the SWEPOS-network
  (the coast)
• - the geoid model (SWEN05 was used, now SWEN08
  is available)
• - the antenna phase center
• - etc
• The accuracy levels can in the future be adjusted
  to standardized tolerances, but more studies at
  different locations and under other conditions
  are then necessary.

15

• The study is available at Lantmäteriets homepage
  as a report LMV-Rapport 20092 Rapportserie
  Geodesi och Geografiska informationssystem.
• Thanks for listening!
• Robert Odolinski robert.odolinski_at_lm.se
• Johan Sunna johan.sunna_at_lm.se

16
(No Transcript)
17
Equipment

• Total station Trimble 5601 DR 200
• GNSS receiver Topcon HiPer
• The GNSS receiver has been configured based on
  recommendations of
• Kortmanual för mätning med SWEPOS Nätverks RTK
  tjänst, with the
• following parameters
• Elevation cut-off angle 15?
• Mean values of 5 observations, with at least one
  second between each observation
• Max antenna height 2 m
• Max PDOP 3

18
Proposed accuracy levels - calculations
Estimated standard errors s0point are given by
a priori standard error ?horizontal 15 mm where
the estimated standard error are scaled by
F-distribution at a risk level of 5 with the
following approximation equation s0point ?
?horizontal(0.96 ö-0.4) ? ?horizontal(0.96
(2n-3)-0.4) and n 20, where -3 are given by
unitary transformation (3 parameters) which
gives s0point ? 20 mm ? s0coord ? 20/??2 ? 15
mm shdiff are given by a priori standard error
?height 25 mm where the estimated standard
error are scaled by F-distribution at a risk
level of 5 with the following approximation
equation shdiff ? ?height (0.96 ö-0.4) ?
?height (0.96 (n-1)-0.4) and n 20
gives shdiff ? 35 mm
19
Proposed accuracy levels - calculations
The following accuracy levels are within a
confidence level of at least 95 based on
estimated standard errors

• Revisit network RTK integrated with the
  production
• measurements or processed alone
• -horizontal deviation ?? 55 mm 2?2Max(s0point
  ) 2,6??position
• -deviation in height ? 100 mm 2?2Max(shdiff)
  2,8??h
• Control with total station of point originally
  measured with
• network RTK
• -unitary transformation unit-weight standard
  error
• (per coordinate) ? 15 mm s0koord
• -translation in height standard deviation ? 35
  mm shdiff
• -deviation in distance ? 40 mm 2Max(s0point)
  2,7?distance
• -deviation in height difference ? 100 mm
  2?2Max(shdiff) 2,8??h
• -deviation in height (after estimating a shift)
  ? 70 mm
• 2Max(shdiff) 2,8?height