Elvas Tower: The Goode-Homosoline Projection - Elvas Tower

Jump to content

Posting Rules

All new threads will be started by members of the Open Rails team, Staff, and/or Admins. Existing threads started in other forums may get moved here when it makes sense to do so.

Once a thread is started any member may post replies to it.
  • 4 Pages +
  • 1
  • 2
  • 3
  • 4
  • You cannot start a new topic
  • You cannot reply to this topic

The Goode-Homosoline Projection Problem statement Rate Topic: -----

#11 User is offline   Genma Saotome 

  • Owner Emeritus and Admin
  • PipPipPipPipPipPipPipPipPipPipPipPipPip
  • Group: ET Admin
  • Posts: 15,356
  • Joined: 11-January 04
  • Gender:Male
  • Location:United States
  • Simulator:Open Rails
  • Country:

Posted 10 April 2015 - 08:51 PM

 BillC, on 10 April 2015 - 06:05 PM, said:

Dr. Ziegler is the author of TransDem. Here is a link to the WebSite. On the webSite there is an active forum here.


Got it.

Good that it can deal with a number of different projections... bad that it cannot deal with the current U.S. standard (an .img raster file). What he can do is read 1 degree U.S. data from 20 years ago. Fairly crude resolution. Current data has a resolution of 1/9th degree which is vastly more accurate when turned into terrain.

Payware.

Well... it looks much like DEMEX for Tranz.

He's been at it for 10 years now... looks to be slowing down a bit on the upgrades.

#12 User is offline   BillC 

  • Conductor
  • Group: Private - Open Rails Developer
  • Posts: 322
  • Joined: 31-May 11
  • Gender:Male
  • Country:

Posted 10 April 2015 - 09:09 PM

 Genma Saotome, on 10 April 2015 - 08:51 PM, said:

Got it.

Good that it can deal with a number of different projections... bad that it cannot deal with the current U.S. standard (an .img raster file). What he can do is read 1 degree U.S. data from 20 years ago. Fairly crude resolution. Current data has a resolution of 1/9th degree which is vastly more accurate when turned into terrain.

Payware.

Well... it looks much like DEMEX for Tranz.

He's been at it for 10 years now... looks to be slowing down a bit on the upgrades.


Below is an excerpt from the current documentation. Website may have old references. Actively maintains patches, for example the latest patch for 2.51 01/24/15 . Great support usually answers user inquires within a day or 2 .

NED
US Geological Survey DEMs are named National Elevation Dataset NED and are
produced in resolutions of 1 arc sec, 1/3 arc sec and – for some areas – 1/9 arc sec. They
can be downloaded via the USGS National Map Viewer
http://viewer.nationalmap.gov/viewer/ (01/2014). The user is offered a navigable map to
zoom to the area he is interested in. Next, he selects NED DEM data product and
resolution from which the server then composes the download from prefabricated
packages of 1 x 1 degrees (15 x 15 arc min for 1/9 arc sec).
TransDEM can directly read NED in ArcGrid format (= ESRI Binary Grid), in Grid Float
(ESRI float grid) or IMG (ERDAS Imagine). These are selectable data formats offered for
a download. TransDEM internally converts the NED data from Plate Carrée pseudo
projection to UTM with a user adjustable raster width.
Note: 1/9 arc sec resolution is supported by TransDEM from version 2.1 The downloads
will be huge. 9 times as big as for 1/3 arc sec and 81 times as big as for 1 arc sec for an
area of the same size.

#13 User is offline   cjakeman 

  • Vice President
  • PipPipPipPipPipPipPipPip
  • Group: ET Admin
  • Posts: 2,869
  • Joined: 03-May 11
  • Gender:Male
  • Location:Peterborough, UK
  • Simulator:Open Rails
  • Country:

Posted 11 April 2015 - 12:11 AM

How about something radical - using latitude and longitude for position within OR?

This means OR moves away from the tyranny of square tiles. They remain square at the Equator but get narrower towards the poles. Like Google does, we assume the Earth is spherical so we can convert from (lat, lng) to position on a tile with a formula.

If DEM data is always supplied with latitude and longitude values, then OR could use it directly.

#14 User is offline   strawberryfield 

  • Hostler
  • Group: Status: Active Member
  • Posts: 55
  • Joined: 16-February 14
  • Gender:Male
  • Location:Cesena, Italy
  • Simulator:OpenRails, Rail3D
  • Country:

Posted 11 April 2015 - 03:57 AM

 cjakeman, on 11 April 2015 - 12:11 AM, said:

so we can convert from (lat, lng) to position on a tile with a formula.



The formula you need is the projection...

#15 User is offline   dajones 

  • Open Rails Developer
  • Group: Status: Contributing Member
  • Posts: 413
  • Joined: 27-February 08
  • Gender:Male
  • Location:Durango, CO
  • Country:

Posted 11 April 2015 - 05:19 AM

 strawberryfield, on 11 April 2015 - 03:57 AM, said:

The formula you need is the projection...


Actually, you don't need a projection. You can convert latitude, longitude and height above ellipsoid to Earth Centered Earth Fixed (ECEF) coordinates. This requires trig functions so you probably want to store ECEF and only convert to lat/long when needed. You need 8 byte floats for ECEF to get sub mm accuracy on the surface, although you could use 4 byte floats if you store position relative to the center of a tile. Rendering from ECEF only requires a matrix multiplication. I think many flight sims use ECEF. OSGEarth uses ECEF internally and I have the code for an old version of that, so I can find the formula they use for converting lat/long to ECEF.

Tiles probably still need to be defined in terms of latitude and longitude, so tiles won't be square, but that shouldn't be an issue.

Doug

#16 User is offline   WaltN 

  • Open Rails Developer
  • Group: Status: R.I.P. or just Retired
  • Posts: 1,086
  • Joined: 28-February 10
  • Gender:Male
  • Location:Vestal, New York
  • Simulator:Open Rails, MSTS
  • Country:

Posted 11 April 2015 - 06:54 AM

 cjakeman, on 11 April 2015 - 12:11 AM, said:

How about something radical - using latitude and longitude for position within OR?

Take that man to the Tower!

I'll show mercy: I sentence him to life in the Tower where he will develop the arithmetic for TDB Traveller and evaluate performance. Parole available when he's done.

Seriously, take a look at Wikipedia's article on "Spherical Coordinates." (And someone will bellyache that Earth is not spherical.) Formulas become more complex (rather than nearly trivial) and filled with sines and cosines. Now consider the travelers -- two per train -- one moving through the train to compute truck position, many times per frame. Trig functions are expensive for that kind of usage. Today's traveler -- with Cartesian coordinates -- requires NO trig functions, just adds and multiplies.

Now glance at the section of that article headed Kinematics. Take a look at velocity and acceleration. Even our physics math would change!

EDIT: Probably the most important problem is that no graphics API (that I'm aware of) supports spherical coordinates. That means you have to project them to Cartesian sometime.

#17 User is offline   Genma Saotome 

  • Owner Emeritus and Admin
  • PipPipPipPipPipPipPipPipPipPipPipPipPip
  • Group: ET Admin
  • Posts: 15,356
  • Joined: 11-January 04
  • Gender:Male
  • Location:United States
  • Simulator:Open Rails
  • Country:

Posted 11 April 2015 - 08:53 AM

@Bill -- Cool. I was reading some other part of the web page that spoke to inputs and it only mentioned the Space Shuttle data. V. good to see 1/9th arc .img data canbe processed.

But processed into what? I dunno Tranz... no idea what their tile files are like. And then there is the payware issue.

#18 User is offline   WaltN 

  • Open Rails Developer
  • Group: Status: R.I.P. or just Retired
  • Posts: 1,086
  • Joined: 28-February 10
  • Gender:Male
  • Location:Vestal, New York
  • Simulator:Open Rails, MSTS
  • Country:

Posted 11 April 2015 - 10:05 AM

I spent most of yesterday afternoon nosing around trying to discover what projections were supported by Microdem and MapWindow. Neither seems to have such a list, except in their internal dialogs. Nevertheless, here’s what I found:

Microdem
Microdem using one standard projection, UTM (for a specifized zone). It maps source materials (DEMs and shapes) into that projection. UTM is no problem for its major user, the US Army – artillery mapping.

MapWindow
The MapWindow web site feature four relevant open source projects:
  • MapWindow 4.x is their current level desktop GIS.
  • MapWinGIS is a control that can be added to Windows Forms applications to do GIS functions.
  • DotSpatial is a C#/.NET library for GIS applications. It also supports Windows Forms.
  • MapWindow 6 is on-hold, meaning planned, but no active development. The plan was to converge MapWindow to use the DotSpatial library.

They are currently working on MapWindow 5, which is an effort to introduce standard coding practices. (I read that to mean programmer clean-up.)

The only forum I discovered for MapWindow itself had no posts more recent than last August. This I found disturbing. Moreover, I registered so that I could post a query, got an activation email, but the email indicated that I’d have to wait for a moderator’s approval. I’m still waiting. The DotSpatial forum was active this April.

I watched a few video tutorials, including one on projections. I definitely saw UTM. I definitely saw TM (Transverse Mercator) as it rolled by a dialog. MapWindow takes a general approach to projections. You specify a standard projection at project definition time. Then as you access source materials, it notifies you when it detects a projection conflict. It then guides you to convert or otherwise correct the conflict.

I’m tempted to try MapWindow for a taste and then try to do something with DotSpatial, but I don’t want to commit myself just yet.

#19 User is offline   WaltN 

  • Open Rails Developer
  • Group: Status: R.I.P. or just Retired
  • Posts: 1,086
  • Joined: 28-February 10
  • Gender:Male
  • Location:Vestal, New York
  • Simulator:Open Rails, MSTS
  • Country:

Posted 11 April 2015 - 10:36 AM

 dajones, on 11 April 2015 - 05:19 AM, said:

Actually, you don't need a projection. You can convert latitude, longitude and height above ellipsoid to Earth Centered Earth Fixed (ECEF) coordinates. This requires trig functions so you probably want to store ECEF and only convert to lat/long when needed. You need 8 byte floats for ECEF to get sub mm accuracy on the surface, although you could use 4 byte floats if you store position relative to the center of a tile. Rendering from ECEF only requires a matrix multiplication. I think many flight sims use ECEF. OSGEarth uses ECEF internally and I have the code for an old version of that, so I can find the formula they use for converting lat/long to ECEF.

Tiles probably still need to be defined in terms of latitude and longitude, so tiles won't be square, but that shouldn't be an issue.

Doug

Doug, a reference please. I've used ECEF before, because it seemed the obvious way to go for an application (seismic processing), but I didn't know the acronym. I'm interested in the tiling and floating point analysis.

#20 User is offline   WaltN 

  • Open Rails Developer
  • Group: Status: R.I.P. or just Retired
  • Posts: 1,086
  • Joined: 28-February 10
  • Gender:Male
  • Location:Vestal, New York
  • Simulator:Open Rails, MSTS
  • Country:

Posted 11 April 2015 - 11:30 AM

FOUND ONE!

3D Engine Design for Virtual Globes

OpenGlobe

  • 4 Pages +
  • 1
  • 2
  • 3
  • 4
  • You cannot start a new topic
  • You cannot reply to this topic

1 User(s) are reading this topic
0 members, 1 guests, 0 anonymous users