Geoid Computation Difficulties in the Pacific Northwest
by
Dru A. Smith National Geodetic Survey
Presented at the 1st Official Meeting of the United States Geoid Committee Silver Spring, Maryland
August 18, 1998
Issues arising in Pacific Northwest geoid computations:
CV!-Terrain Corrections at NGS disagree with those at GSD(GC)
( -Terrain may not properly be represented by 30" data ~f~l/ fu -DEMs are not properly referenced to a consistent vertical
datum
qt6.. G96SSS agrees to decimeters with GPS/Benchmarks in PNW, but EGM96 {Bouguer corrected) disagrees to 1 meter with GPS/Benchmarks
Problem #1
TC differences, NGS vs GSD(GC)
1996 Study of Southern British Columbia
·.:
CANADIAN TERRAIN CORRECTIONS (Jan 93) Attemp~ !Q ~produce
- 2 DTEDs: - TOP030 - New Canadian DTED 1995
- 5 Independent TC programs: - ftc.f (FFT, Milbert) - tc01.f (Flat top Prism, Milbert) - tc.f (Flat top Prism, Forsberg) - tcpts01.f (Flat top Prism, Veronneau) - triter4.f (Inclined top Prism, Rupert/Beach)***
- 202 points in soo- 51 o N, 235.5°- 237.5°
*~* = Not fully tested yet . . - .. - . . . ....
~ .. ' .
Ave RMS Min Max
TC DIFFERENCES Jan93(Can Database) MINUS Other TCs
TOP030 1995 Canadian DTED 12 mgals 14 mgals 17 mgals 19 mgals -17 mgals -12 mgals +45 mgals +54 mgals
- 122 non-zero points - Overall stats for all 4 fully-tested programs - Conclusion: The Jan93 TCs {currently still in the Canadian database) are systematically higher than all 8 (4 programs, 2 DTEDs)TC sets at NGS,by a factor of 1.5 to 1.8
TC(Jan93) Minus TC(TOP030/Prlsm/200km)
-124.5 -123.5 -122.5
51.0 51.0 12.4
3.9 5. 9.3
-4.2 10.6 16.3 1.4-0.3 5.0 - 14.0 7
0.9 -2.1 14.6 4.7
-4.4 19.4 13.1
8 .6 13.4 6.5
39.8 9 .9 3 - 8 .8 16.5 23.3 7.3
0.0 0.3
(1) 15.1 11.7 6.4
"'0 8.5 ::J 17.2 +" 50.5 - -~.a> 2.8 50.5 +"
. lL8
ctS fi.Q -5.1 ..J 1&· 9 19.3 28.0 3.6 -4.0 18.5
30.8 22.2 3 .0 1.4 9.0
3.9 1.0
32.2 18.3 11. 3 8.7
·1~~7 4.5 4.0 7.7 -4.3 . 12.5
8 5 1.3 3.0
1.5 -8.3 -2.3
-5.7 -0.7 -2. - 16.6 17.7 17.4
0.1 18 .2 14
15.5 -4.6
4 .2 13.4
50.0 50.0
-124.5 -123.5 -122.5
Longitude
CANADIAN TERRAIN CORRECTIONS (Jan m Preliminary Results
- Unable to reproduce the January 1993 TC's
- Attempts using the old (TOP030) DTED gave results closer to Jan 1993 than the new DTED
- FFT method agrees to within +1- 1 mgal with prism methods, except for large (>30 mgal) spikes, where the FFT is systematically too low by an average of 8 mgals
~fr - Level 1 DTED (3 11X6 11
) unable to get Jan93 TCs! -M'-'1T·~:dr":. .. ...... . . ... ··-·-:--:-:,.,..-····---·-- ,, .:.:~~~~-~·~-... c---. ~ . i:~-.-·· M . , ···- ·-·· ·· ···- ···-MM ----:--~--.. - -~· .... - .... ·- -----,-~_ ... ;r-'::7- ·.~·:_:· --:-~-;-:-:--:-:-~·;-., ·. ' ·-··--~···· -- ··--------.. ····· -..... _. __ ....... .... . . ;.' ' (' ..
· ~·i· ... :. 'tY', . . .. .. ~ .· . _.,;_•.,:,._ . . •. . . ·"" . • • • 0 •
Problem #2
30 .. OEM fails !Q capture full terrain §!gnal
1996 Study of Southern British Columbia
Create Canadian TC's from 30 11 and 3 11 data
Examp~ point:
TC(Canada DB, W TC(Topo30) TC(3 11 X6 11)
(50.66067, 236.88400)
54.6 mgals 14.4 26.0
[This study will be re-investigated in August/September 1998]
Problem #3
No vertical datum consistency in available DEMs
1997/98 DEM/DTED Study
Sources Qf ~(or better) DEMs:
A) Most go back to 1960snos DMA 3 .. data - Old DMA data comes from 1 :250,000 maps
-1 :250,000 maps from old satellite sources - NIMA was updating cell by cell with new photo sources, but has nearly stopped recently
***NO vertical datum documentation or consistency
B) SRTM (expect new DTED in 2003) ***Vertical datum could be consistent!
C) USGS is digitizing 1 :24,000 maps onto 1 0 and 30 meter UTM grids (90°/o of west US done, 30o/o of East US done) ***Vertical datum well defined as NGVD29
/ , • ;:::. \ 1.1 ·.J II (? 0WJ ~q fr D ~ L-P () p \; , (:::.{J1/ i/v1 '\ -ti. vv
~- I -t Vc1{~'-0~ + Sb0 fFt1.JPr-f J 0(1\C{
· OEM Differences DMA 1998 DTED minus USGS 3 ..
(44° to 49° N, 237° to 243° E) .
RED = +25 m or greater differences MAGENTA= -25m or lower differences
Geoid Undulation changes due to random +200 meter error in one 1 °X1 o OEM (of 2'x2' elevations)
Red = + 1 em or greater change Magenta = -1 em or lower change
Geoid Undulation changes due to systematic +20 meter error in one 1 °X1 o OEM (of 2'x2' elevations)
Red = + 1 em or greater change Magenta = -1 em or lower change
Problem #4
EGM96 bust in Pacific Northwest
1996/97 GEOID96 computation and validation
G96SSS minus EGM96(Bouguer Corrected)
Red=+ 1.5 meters and greater differences Magenta=-1.5 meters and lesser differences
EGM96 and G96SSS vs. GPS on Benchmarks
-Compared both models to ITRF94/NAVD88 data
- National average residuals of: EGM96/GPS/BM: +41 em G96SSS/GPS/BM: +43 em
In PNW (44°-49°, 237°-243°), with average removed: EGM96/GPS/BM: +94 em + 28 em G96SSS/GPS/BM: -12 em ± 19 em
Q)
"C ::J ..... ..... co ..J
GPS/BM/G96SSS residuals (about 43 em ave)
-123.0 -120.0 -117.0 49.0,_~~~===========-========~==========~~~~============~~~=====r
-3c.8 • 49.0
•
-25 .5 •
-\1.4 -18.2 -15._58 ~
-~.2 . .
14.4
21.4 •
26.8
: ~~·ia86 ••• 30.1
•
29.6
- 1.6 • - n._ 3
J!.:a!
•
•
16.6
33.6
-42.9 •
• -3 0.3 • -38 .4
• -27.0 -32.1 •
• :41.8 .. ~~~~i
-3'7~0.9 • ·-27.5
-29.4 • • _32.5 .-.s.OfeS~~~j2~.a21.2
• 2A 2 - .-. z4.9 - 11 .5
:2J1~11:tf!!l-1~~.6 :21.6 : 14.7• • • ·;t:f~~6.4 • :·4 • .. ~
~ • -'tt;.9 • 5.7 -7·3 .-3.9 1~
.: 1.~ :-~?:J: 1 ~. 9 ..5 1 • -If. .0
1.9 4o.r· - 3 1 • • Q:§ .O . . . . • 6.7
• - 3.3
•
5.0 •
19.9 • 25 .0
•
• •
-50.'1 •
-46.4 •
-53. •
50 .0 •
-41.6 ._44. 7 -27.2 •
-38.1
- 23.7 -25.4 - 3 .0 • •
-20.6 •
-17.7 •
- 15.5 •
-17.6.-20.5 •
-14 .6 •
-4.9 •
-22.1 •
-27.1 •
32.4 • 38.0 •
44.0~--------------------~----------------~---------------------r--------------------~--------------------~-----------------+ 44.0 -117.0 -123.0 -120.0
Longitude
Q)
"C :::J ...., ·-...., as ~
GPS/BM/EGM96 residuals {about 41 em ave)
-123.0 - 120.0 -117.0 49.01-~~~q===~==========~==========~==--1~0=1.~9~d6==========d===.~~====~ 49.0
109 .9
148.3 •
97.7
: e~bm •••
90.2 •
16.6 •
w.o 52.9 .101.4~~-7
50.5 • .s!.\8
157.9 •
82.1 •
•
• 118.8 48.9 • 70.1 • • 104.2
63.9 • • • 63.4 •• 1~i~~
• 93.~8.6 • H~:o •
115.7
28.9 • 80.9 •
97.1 •• 8.~,8~~55p 90.2
• • 1 H~3 ~C\1A 4$~2 • 82 ,1 109t1 101•8
•• ~89~fai9~· 4.5 • • • • 411
••• ;c .. .m.6 ~ • 0 7'
• 115.1 1,2.5 .• 112.0 7.
: 153.0 •• ;~ .. 4~~% 4:1.5 1.15. .
• 166.-61.72 .~ 113.2 •• 1.1 ~8.1
•
•
• 94.7
•
• 80.5
•
66 .4 •
91.6 •
104.8 •
• 91.6
61.1
64. •
• 87.6
• 92.2 •
92. •
60.4
111.5 106.3 11b.4 .3 • • 105.4
• 91 .7 •
108.5120.3 •
109.8 112.2 • 130.2 •
• 111.0 •
121.6
107.1 •
44.0~--------------------~--------------------~--------------------~-----------------,--------------------~--------------------~ 44.0 -117.0 -123.0 - 120.0
Longitude
CONCLUSIONS
- Most geoid research effort is concentrating on the Pacific Northwest
- Many problems stem directly from unreliable high resolution DEMs
- GPS on Benchmarks provide a useful independent check on geoid models
- Additional research being done on downward continuation and long wavelength terrain effects