Here's my problem; I have an origin for a grid system - the NYC Dock Department - expressed as a geographic position on the United States Standard Datum.
My thoughts were to select four stations around the origin and do some geodetic inverses to find a scale, shift and rotation to decide whether I could determine the NAD 27 coordinate of the USSD position.
So, I select KV4020 Highwood 2 1867, KU3989 St Pats North Cathedral 1903, KU3592 Manhattan Holy Redeemer 1912, and KV4284 Castle Stevens 1875. They all possess USSD and NAD27 geographic positions.
I run inverses between all the points on Clarke 1866 (which is common to both USSD and NAD27) using the USSD and NAD27 values and find a large residual in azimuths and distances associated with Castle Stevens. The azimuth north towards Highwood 2 1867 for example is 6 arc-minutes different and the distance to St Pats 1903 is 5 meters different between datums.
I pull the datasheet for Castle Stevens 1875 and find that the first position is in the southwest corner of the tower and there is a reference measurement to a flagpole 5 meters east.
If I take the tangent of one foot (0.000 00485) and multiply by the distance to Highwood 2 1867 (2875 meters) and multiply by 360 seconds...I get 5 meters.
SO, it seems the USSD position for Castle Stevens 1875 is the flagpole and the NAD27 position a roofing nail in the southwest corner of the tower which wasn't eccentrically reduced to the flagpole.
My comp sheet.
While you have made some interesting deductions from available data, the 2001 recovery indicates a different building now exists in the site.
The reduction of eccentric observations (either observed point or the point from which observations are made) were common enough to merit mention in the 1930's USC&GS manual of triangulation computation and adjustment.
I would also be very hesitant to say that I found an 1875 punch hole in a 0.5 inch diameter copper nail. As mentioned in an earlier thread on the lost art of triangulation, many intersection stations have changed over time. Flagpoles atop elevated structures are frequently moved or replaced.
If you have the time and inclination, and if there are additional elevated points visible from this site, you could do a three-point resection. Of course, GPS is also an option. The manual mentioned above (and linked in the earlier thread) has a worksheet to perform the reductions. Unfortunately you will need to remember how the use logarithms. They use base10.
Fun, fun, fun.
🙂
I wrote the 2001 recovery note.
Scott: The NGS has all of the old observations in their database. I used to have access to it via telnet to be able to retrieve the obs, but not any longer. However, people inside NGS can still retrieve the data of course. Perhaps you could get what they have and see if there are some eccentric reductions that got left out, or a name was changed inadvertently.
It sure seems like you are right, a lot of times those old networks used spires, flagpoles, cupolas, etc, and of course they were occupied eccentrically. They had routines for "reduction to center".
I was able to make progress on my task at hand by other means.
I have a terrible copy of the Dock Department map.
A detail of the "Title block"?
Most of the rest of it...
So, I calc'ed up some ellipsoid shifts. Theoretically, USSD and NAD27, share a spheroid (Clarke 1866) so the only change should be a re-definition of the "origin" (which isn't one point in USSD, or NAD27 for that matter).
And from an average shift I calculated a coordinate for the Dock Department Origin.
And then I took the Dock Department coordinates for the three Borough Coordinate system origins as shown on their map and did a similarity 6 parameter transformation and was able to compute another coordinate to check my first derived coordinate.
Only 0.635 foot difference!
How's that for darn lucky?!?!?
o.O
if I recall correctly the two datums (USSD and NAD27) have the same origin coordinates at MEADES RANCH, but the starting azimuth was SLIGHTLY different, and there were not as many Laplace stations and taped base lines to control scale and rotation. By the time you get to NYC from the middle of Kansas, the errors accumulate. So, USSD and NAD27 both have a common origin coordinate, but the further away you get the more difference there is. NAD27 used the azimuth to WALDO, not sure about USSD. And I believe the origin coordinates are astronomic, which means deflection of the vertical at the origin was zero. And the geoid separation was also zero.
Your analysis seems to be pretty solid. Around Pittsburgh there exists a harbor line system, done 1895-1905 by the USED (now USCE), which basically triangulated and traversed up both sides of the rivers. The origin is gone (part of the wall fell in the river), but I found City notes where they tied in the origin point. Miraculously I found the point they made the tie from (survey disk in a cast iron box at an old train station), and also the backsight point they used (on top of a bridge tower for a large bridge). So, I gps'd them, and using their notes recreated the origin in NAD83. Over the years I have recovered quite a few harbor line monuments and these check so-so, probably about 10 feet floating around in there because they were open ended chains of quadrilaterals, with no control at the far ends, going around 50 miles with short lines (1000-2000 feet) making up the chains.
I think USSD used Principio and azimuth to Turkey Point.
Memory.. Hope the data banks aren't too corrupt.
Edit; nope I'm wrong New England datum used Principio, USSD used Meades Ranch.
I was about to admit maybe I was wrong, that it was NAD (just NAD) that used MEADES RANCH, but I think USSD and NAD are the same.
Scott Zelenak, post: 364521, member: 327 wrote: 🙂
I wrote the 2001 recovery note.
Obviously having too much spare time, I have chosen to be annoying and pursue the matter of whether one can verify the point recovered. To that end, I examined the siteÛªs data sheet.
I may have unwarranted confidence that such an error would be hardly likely given the way the USC&GS operated.
I note that you reported the ÛÏmark not foundÛ in 2001. You also wrote in the recovery
ÛÏSite of mark is now a large building on campus Û?Û My interpretation was that the former castle was replaced with a ÛÏlarge building.Û My interpretation seems reasonable to me.
That said, I defer to your superior knowledge of local landmarks and changes to the area. Of course neither of us were around in 1932, much less 1875.
But more to the point, I wanted to bring to the attention of those reading this thread that the NGS data sheets for triangulation stations include a useful section known as the ÛÏbox score.Û It contains geodetic azimuths and distances to points visible from the site (at least at the time the station was positioned).
To resolve the issue of whether this is truly CASTLE STEVENS 1985 one can set up on the site and turn angles to the points in the box score. In the case of this site, it appears that there are a number of nearby intersection stations. Good geometry from the site would permit a three-point resection.
I have copied the relevant sections of the data sheet below.
KV4284 HISTORY - 20010319 MARK NOT FOUND PORANY
KV4284 STATION RECOVERY (2001)
KV4284
KV4284'RECOVERY NOTE BY PORT AUTHORITY OF NY + NJ 2001 (SSZ)
KV4284'SITE OF MARK NOW A LARGE BUILDING ON CAMPUS OF STEVENS TECHNICAL
KV4284'COLLEGE.
KV4284 ***********************************************************************
KV4284 DESIGNATION - CASTLE STEVENS 1875
KV4284 PID - KV4284
KV4284 STATE/COUNTY- NJ/HUDSON
KV4284 COUNTRY - US
KV4284 USGS QUAD - JERSEY CITY (1981)
KV4284
KV4284 *CURRENT SURVEY CONTROL
KV4284 ______________________________________________________________________
KV4284* NAD 83(1996) POSITION- 40 44 41.51640(N) 074 01 26.51663(W) ADJUSTED
KV4284* NAVD 88 ORTHO HEIGHT - 32. (meters) 105. (feet) SCALED
KV4284 ______________________________________________________________________
KV4284 GEOID HEIGHT - -31.870 (meters) GEOID12B
KV4284 LAPLACE CORR - 6.15 (seconds) DEFLEC12B
KV4284 HORZ ORDER - THIRD
KV4284
KV4284.The horizontal coordinates were established by classical geodetic methods
KV4284.and adjusted by the National Geodetic Survey in September 1999.
>>>>>>>>> from file dsdata.txt <<<<<<<<<<<<
DATA ITEM: Box Score
DISPLAYED: When available for Old Horizontal Control marks.
COMMENTS : Distance may be blank; PID may be blank.
There may be unadjusted marks not shown that are
in the vicinity of the Old Horizontal Control mark.
Contact NGS regarding their information.
<<<<<<<<>>>>>>>>>>>
>>>>>>>>> BOX SCORE for KV4284 <<<<<<<<<<<<
KV4284: Primary Azimuth Mark Grid Az
KV4284:SPC NJ - MANHATTAN 23 ST FERRY BLDG FP 068 11 21.2
KV4284:SPC NY L - MANHATTAN 23 ST FERRY BLDG FP 068 30 56.2
KV4284:UTM 18 - MANHATTAN 23 ST FERRY BLDG FP 067 51 46.2
KV4284
KV4284|---------------------------------------------------------------------|
KV4284| PID Reference Object Distance Geod. Az |
KV4284| dddmmss.s |
KV4284| KV4105 WEEHAWKEN HACKENSACK WATER TWR APPROX. 3.3 KM 0052503.0 |
KV4284| KV4108 WEEHAWKEN DIA LINES PIER K TK APPROX. 2.9 KM 0144148.9 |
KV4284| KU3962 MANHATTAN C GAS 45TH ST HOLDER APPROX. 3.0 KM 0461956.2 |
KV4284| KV4366 MANHATTAN 23 ST FERRY BLDG FP APPROX. 1.3 KM 0682959.6 |
KV4284| KV4370 MANHATTAN NATL BISCUIT CO FLAG APPROX. 1.3 KM 0962917.1 |
KV4284| KU3594 MANHATTAN CONSOL G E CO CLKTWR APPROX. 3.2 KM 1113632.2 |
KV4284| KU3582 MANHATTAN 1 FIFTH AVE BLDG TWR APPROX. 2.8 KM 1212733.4 |
KV4284| KV4388 MANHATTAN DWNTWN ATHLETIC C FP APPROX. 4.4 KM 1703627.5 |
KV4284| KV4267 HOBOKEN ST PETERS AND PAULS CH APPROX. 0.6 KM 2264024.5 |
KV4284| KV4287 JERSEY CITY K AND E HIGHEST TK APPROX. 0.8 KM 2561537.0 |
KV4284| KV4123 JERSEY CITY ST ANNES CATH CH APPROX. 2.5 KM 2913605.3 |
KV4284| KV4112 HOBOKEN H R MALLINSON TANK APPROX. 1.9 KM 3224236.5 |
KV4284| KV4097 UNION CITY ST JOSEPHS CATH CH APPROX. 2.3 KM 3252645.3 |
KV4284| KV4120 HOBOKEN LIPTON TEA CO CHIMNEY APPROX. 1.1 KM 3503005.3 |
KV4284|---------------------------------------------------------------------|
Three things.
It could have been two stations with the same name. If you read the XYHt story linked above its an issue in the USSD to NAD83 transformation.
Inverse between the stations I've identified above by PID. Castle Stevens has an excessive difference between USSD and NAD27 of 15 meters in length and 6 arc minutes of azimuth when the others are around 1 to 2 meters. Unless I've done something wrong.
And finally, if we had the 1875 and 1932 observation data and surviving stations from both sets, that'd be the only way to prove a difference.
A copy of my inverses on Clarke spheroid.
Castle Stevens really upsets the apple cart.
Considering most of these stations are second and third order horizontal, some of the deviations are remarkably tight. IF these were perfect stations, I would assume the relative geometry between points would remain identical from USSD to NAD27 with only the geographic values changing. Include the fact that the USSD values were "determined" over a span of 45 years and the NAD values were most likely re-observed within a season, and a meter or 40 arc-seconds change seem reasonable over some of these distances.
So, is the large distance and azimuth value change at Castle Stevens statistically significant?
Key map to station location.
It seems highly coincidental that the distance deviations are in the apparent axis of the roof nail station and the flagpole (i.e. NE/SW) and that the largest angle deviations are at right angles to that axis.