1. Comparisons of GNSS and Leveling- Derived Orthometric Heights Using GIS Kevin M. Kelly, Esri Jay Satalich, Caltrans Kelly, K. M. and J. Satalich (2012). Comparisons of GNSS and leveling-derived orthometric heights using geographic information system software, AGU Fall Meeting, San Francisco, December 3-7, 2012, Abstract G13A-0940.

2. Outline NGS geoids Problem statement Data and results Conclusions

3. NGS Geoids Six models since 1990 - GEOIDYY: YY = [90], [93], 96, 99, 03, {06}, 09, 12A Hybrid geoids 96 through12A - intended to recover NAVD88 H from specific realizations of NAD83(YY) h - Corrector surface to the corresponding gravimetric geoid, USGGYYYY - This surface biases gravimetric geoid to fit NAD83(YY) ellipsoidal heights, h surveyed on NAVD88 benchmarks (GPSBM)

4. Problem Statement How well do NGS hybrid geoids perform locally? Problem? In the Palos Verdes Peninsula of Southern California, discrepancies of up to 8 cm persist between published NAVD88 heights and those derived from GPS on benchmarks using the last four geoid models: GEOID99, GEOID03, GEOID09 and GEOID12A. So What? Local accuracy is the workhorse of the geoid model; local projects exact high demands from geoid model performance. Solution Research and test potential causes Collect new gravity data to help verify existing holdings – GRAV-D Densify GPSBM data? Improve data screening and quality control Benefits? NAVD88 usage will continue even after the new datums. Geoid modeling and data quality implications/improvements for forthcoming models. Solutions for other problematic local areas.

5. Environmental Setting Key geophysical features influencing benchmark heights near the Palos Verdes Peninsula, California. Palos Verdes Peninsula o Made of sandstone and basalt o Active landslides o Historic landslide areas Los Angeles Basin o Made of alluvium o Wilmington Anticline o Wilmington/Torrance oil fields

6. The Data Story Steady improvement from GEOID99 to GEOID12A but ~8 cm discrepancies persist Hybrid geoid surface lies consistently above the surface defined by NAVD88 REDONDO – monument disturbance from construction T1217, T1053 – 4 cm elevation differences between 1960 and 1994 T1217 – ellipsoid height stable to 1 cm between 1997 and 2012 The main purpose of this study is to show that these disagreements are not due to shortcomings in the control data – that is, the leveling and GPS.

7. Benchmark stability over time (1) L25468/14 (1994) minus L17850 (1960) (±1.17 ±2.31) New - Old: ±2.59 mm New-minus-old comparison of geodetic leveling shown as a component of error propagation ± numbers are a posteriori standard error for 1 km of single-run leveling in mm Error propagation:

8. Benchmark stability over time (2) L25468/14 (1994) minus L24301/1 (1978) (±1.17 ±0.93) New - Old: ±1.49 mm New-minus-old comparison of geodetic leveling shown as a component of error propagation ± numbers are a posteriori standard error for 1 km of single-run leveling in mm Error propagation:

9. Benchmark stability over time (3) L25468/14 (1994) minus L25180 (1989) (±1.17 ±0.99) New - Old: ±1.53 mm New-minus-old comparison of geodetic leveling shown as a component of error propagation ± numbers are a posteriori standard error for 1 km of single-run leveling in mm Error propagation:

10. Benchmark Stability - Tabular New-minus-old comparison of orthometric heights in meters. Benchmarks shown with an asterisk indicate historic vertical movement beyond 2 cm between 1994 and 1960. BENCHMARKPID RUNNING DISTANCE (km) 1994 minus 1989 1994 minus 1978 1994 minus 1960 941 0660 TIDAL 8DY10830.00FIXED Z 1217DY11188.950.003580.00138none T 1053*DY099413.300.013750.008140.04157 Y 1312DY249918.380.009120.00541none S 1053DY101719.890.004320.004950.00038 U 1217DY122721.940.00208-0.00321none E 169DY123023.970.00421-0.011580.00857 T 1217*DY123124.66-0.00595-0.02714none 34 33 ADY905024.770.00115-0.00636none T 1313DY249824.970.00330-0.01071none F 788DY123325.54-0.00270-0.00917-0.00741 S 1217DY123727.010.00368-0.00131none 34 39DY904529.08-0.00103-0.01846none TORRANCE F 10 RM 1*DY121730.12-0.00447-0.03458-0.05578 REDONDO*DY121231.04-0.00178-0.11205-0.19149

11. GPS Ellipsoidal Height Repeatability New-minus-old comparison of ellipsoidal heights in meters GNSS STATION 2012 minus 2011 2012 minus 1997 2012 minus 1994 2012 minus 1992 PALOS VERDES ARIES 7268FIXED 941 0660 TIDAL 80.007none0.0140.012 Y 1312none -0.003none S 1053none0.000none U 1217none0.004none T 1217none0.010none T 1313none0.0040.008none F 788none-0.006none 34 39none0.009none TORRANCE F 10 RM 1none-0.0020.003none REDONDO*none0.036none

12. Conclusions Discrepancies of up to -8 cm persist between published NAVD88 heights and GPSBM for all NGS hybrid geoids from 1999 to 2012 Ellipsoidal heights exhibit repeatability at the 1 cm level over four epochs spanning 20 years Differential leveling over four epochs spanning 34 years reveals that benchmarks are stable at ±2 cm or better level The data support an apparent shortcoming in NGS geoid models in the Palos Verdes area

13. A Real Possibility Fig. 12 The effect of significant gravity biases on the geoid Saleh, J., X. Li, Y.M. Wang, D.R. Roman and D.A. Smith. (2013). Error analysis of the NGS surface gravity database, J Geod 87:203-221.