ILMF slide presentation on Version 5 (Dave Maune, Denver

advertisement
February 17-19, 2014
Denver, Colorado, USA
ASPRS Accuracy Standards for
Digital Geospatial Data
Dr. David Maune (Dewberry)
Dr. Qassim Abdullah (Woolpert)
Hans Karl Heidemann (USGS)
Doug Smith (ASPRS Photogrammetric Division)
February 17, 2014
www.lidarmap.org/international
Produced by Diversified Communications
PE&RS, December, 2013
Published as DRAFT FOR
REVIEW
Comments due to
committee by Feb 1st
Revised standards to be
submitted to ASPRS
Board for decision
during annual
conference in March
Objectives of New Standards
Replace existing ASPRS Accuracy Standards for LargeScale Maps (1990), designed for hardcopy maps with
published scale and contour interval, and ASPRS
Guidelines, Vertical Accuracy Reporting for Lidar Data
(2004), with new accuracy standards that better address
digital orthophotos and digital elevation data
Establish/tighten horizontal accuracy standards for surveygrade, mapping-grade, and visualization-grade
orthophotos and planimetric maps
Establish vertical accuracy standards for a broad range of
vertical data accuracy classes
www.lidarmap.org/international
Horizontal Accuracy Standards for Digital
Orthophotos
Pixel size can be in centimeters, inches or feet
Class I refers to highest-accuracy survey-grade orthophotos
Class II refers to standard, high-accuracy mapping-grade orthophotos
Class III to Class N refer to lower-accuracy visualization-grade
orthophotos for less-demanding user applications
www.lidarmap.org/international
Horizontal Accuracy/Quality Examples for Digital
Orthophotos (Hi-Res)
www.lidarmap.org/international
Horizontal Accuracy/Quality Examples for Digital
Orthophotos (Mid-Res)
www.lidarmap.org/international
Horizontal Accuracy Standards for Planimetric
Maps
RMSExy values must be in centimeters for all Map Scale Factors
Class I refers to highest-accuracy survey-grade maps
Class II refers to standard, high-accuracy mapping-grade maps
Class III to Class N refer to lower-accuracy visualization-grade maps
for less-demanding user applications
www.lidarmap.org/international
Horizontal Accuracy/Quality Examples for
Planimetric Maps (Large-Scale)
www.lidarmap.org/international
Horizontal Accuracy/Quality Examples for
Planimetric Maps (Medium-Scale)
www.lidarmap.org/international
Vertical Accuracy Standards for Digital Elevation Data
www.lidarmap.org/international
Vertical Accuracy Classes (most demanding)
 Class I, the highest vertical accuracy class, is most
appropriate for local accuracy determinations and tested
relative to a local coordinate system, rather than network
accuracy relative to a national geodetic network.
 Class II, the second highest vertical accuracy class
could pertain to either local accuracy or network accuracy.
 Class III elevation data, equivalent to 15-cm (~6-inch)
contour accuracy, approximates the accuracy class most
commonly used for high accuracy engineering
applications of fixed or rotary wing airborne remote
sensing data.
www.lidarmap.org/international
Vertical Accuracy Classes (LiDAR)
 Class IV elevation data, equivalent to 1-foot contour accuracy,
approximates Quality Level 2 (QL2) from the National
Enhanced Elevation Assessment (NEEA) when using airborne
lidar point density of 2 points per square meter, and Class IV
also serves as the basis for USGS’ 3D Elevation Program
(3DEP). The NEEA’s Quality Level 1 (QL1) has the same
vertical accuracy as QL2 but with point density of 8 points per
square meter (Class III density). QL2 lidar specifications are
found in the USGS Lidar Base Specification, Version 1.1.
 Class V elevation data are equivalent to that specified in the
USGS Lidar Base Specification, Version 1.0.
 Class VI elevation data, equivalent to 2-foot contour accuracy,
approximates Quality Level 3 (QL3) from the NEEA and covers
the majority of legacy lidar data previously acquired for federal,
state and local clients.
www.lidarmap.org/international
Vertical Accuracy Classes (less-accurate)
 Class VII elevation data, equivalent to 1-meter contour
accuracy, approximates Quality Level 4 (QL4) from the
NEEA.
 Class VIII elevation data are equivalent to 2-meter
contour accuracy.
 Class IX elevation data, equivalent to 3-meter contour
accuracy, approximates Quality Level 5 (QL5) from the
NEEA and represents the approximate accuracy of
airborne IFSAR.
 Class X elevation data, equivalent to 10-meter contour
accuracy, represents the approximate accuracy of
elevation datasets produced from some satellite-based
sensors.
www.lidarmap.org/international
Non-vegetated Vertical Accuracy (NVA)

Non-vegetated Vertical Accuracy (NVA), i.e., vertical accuracy at the 95%
confidence level in non-vegetated terrain, is approximated by multiplying the
RMSEz (in non-vegetated land cover categories only) by 1.96.
 This includes survey check points located in traditional open terrain (bare
soil, sand, rocks, and short grass) and urban terrain (asphalt and concrete
surfaces).
 The NVA, based on an RMSEz multiplier, should be used in non-vegetated
terrain where elevation errors typically follow a normal error distribution.
[RMSEz-based statistics should not be used to estimate vertical accuracy in
vegetated terrain where elevation errors often do not follow a normal
distribution for unavoidable reasons.]
www.lidarmap.org/international
Vegetated Vertical Accuracy (VVA)
 Vegetated Vertical Accuracy (VVA), an estimate of vertical
accuracy at the 95% confidence level in vegetated terrain, is
computed as the 95th percentile of the absolute value of vertical
errors in all vegetated land cover categories combined, to include
tall weeds and crops, brush lands, and fully forested.
 For all vertical accuracy classes, the VVA is 1.5 times larger than
the NVA.
 If this VVA standard cannot be met in impenetrable vegetation
such as dense corn fields or mangrove, low confidence area
polygons should be developed and explained in the metadata as
the digital equivalent to dashed contours used in the past when
photogrammetrists could not measure the bare-earth terrain in
forested areas.
 See Appendix C in the full ASPRS standards for low confidence
area details.
www.lidarmap.org/international
Vertical Accuracy/Quality Examples for Digital
Elevation Data
www.lidarmap.org/international
Appendices

Appendix A: Review of prior standards, guidelines, specifications, plus
NEEA and 3DEP

Appendix B: Example accuracy/quality examples above

Appendix C: ASPRS accuracy testing/reporting guidelines

Appendix D: Accuracy formulas and a working example of how to test and
report the vertical accuracy of elevation data for a typical LiDAR dataset
www.lidarmap.org/international
Recommended Number of QA/QC Check Points
Based on Area (km2)
For areas >2500 km2, add 5 additional check points, horizontal and/or vertical, for
each additional 500 km2 area. Each additional set of 5 vertical checkpoints for 500
km2 would include 3 check points for NVA and 2 for VVA.
www.lidarmap.org/international
Error Histogram for sample LiDAR dataset, 20 check
points each x 5 land cover categories
Normal error
distribution, except
two outliers among
100 check points
Weeds
Fully
&
Forested
Crops
www.lidarmap.org/international
Traditional Accuracy Statistics
This demonstrates why the 95th percentile is used (rather than RMSEz x 1.9600) in
vegetated land cover categories.
95th percentile errors will approximate RMSEz x 1.9600 as errors in any land cover
category approach a normal error distribution
Mean errors vary between -2 cm and +4 cm; this is excellent for normal distribution
www.lidarmap.org/international
Comparison of NSSDA, NDEP and new ASPRS
statistics for example dataset
Errors do not approximate a normal distribution in Weeds & Crops, and in Fully
Forested, also causing the Consolidated to fail should we use RMSEz x 1.9600
www.lidarmap.org/international
Comparison of NSSDA, NDEP and new ASPRS
statistics for example dataset
By using the 95th percentile for Supplemental Vertical Accuracy (SVA) and
Consolidated Vertical Accuracy (CVA), the dataset passes, as it should.
The new NVA uses RMSEz x 1.9600 to estimate vertical accuracy at the 95%
confidence level in non-vegetated categories where errors should be normal
The new VVA uses the 95th percentile to estimate vertical accuracy at the 95%
confidence level in combined vegetated categories where errors may not be
normal
www.lidarmap.org/international
Any Questions?
www.lidarmap.org/international
Download