4 replies to this topic

[MrTheSpoon]

Hi,

I'm developing a GIS that will calculate distances between two points in Tunisia.

While I can use a great curve formula to calculate the distance, processing speed is a key factor and I would rather use the pythagoras theorem. For this prupose, I need to project my maps of Tunisia into an azimuthal equidistant projection, trouble is I don't know which one to use.

I have found a few projections based on Tunisia: Carthage_UTM_Zone_32N, Carthage_TM_11_NE, Nord_Tunisie, Sud_Tunisie but I don't know if any of them allow you to measure shortest distances with a straight flat line.

Help please!

Many thanks.

[Jimbo]

Hi,

I'm developing a GIS that will calculate distances between two points in Tunisia.

While I can use a great curve formula to calculate the distance, processing speed is a key factor and I would rather use the pythagoras theorem. For this prupose, I need to project my maps of Tunisia into an azimuthal equidistant projection, trouble is I don't know which one to use.

I have found a few projections based on Tunisia: Carthage_UTM_Zone_32N, Carthage_TM_11_NE, Nord_Tunisie, Sud_Tunisie but I don't know if any of them allow you to measure shortest distances with a straight flat line.

Help please!

Many thanks.

If your developing a GIS do you have a website that you could show us to give out more details about what it is, how it will be used, unique qualities that are different from other GIS, etc.

Also in regards to your question it may help to know what your programming platform is before anyone can help you with that question.

[MrTheSpoon]

If your developing a GIS do you have a website that you could show us to give out more details about what it is, how it will be used, unique qualities that are different from other GIS, etc.

Also in regards to your question it may help to know what your programming platform is before anyone can help you with that question.

The GIS is internal to my company so I can't publish any screenshots I'm affraid.
The data I have is an ESRI polygon shapefile of the delegations (counties) of Tunisia. There is also another shapefile of points denoting various locations in Tunisia.
The resulting application will generate thousands of points throughout Tunisia and calculate the distance from the generated points to the shapefile points.

Initially I'll just be using ArcGIS, then later the program will be built in C#

The only bit that I need is to find a good equidistant projection for Tunisia.

[Crischan]

UTM, TM and the two Lambert projections are conformic. So you'll have to choose a equidistant projection yourself to define a working CRS for your application. Have a look at http://www.radicalca.../?projectionref which has a very nice overview of projections for different scales and purposes.

Cheers,
Crischan

[MrTheSpoon]

Thanks for the link Crischan. That website really helps me visualise the various projections.
I posted this same question on the ESRI forums and Bill Huber replied with:

An azimuthal equidistant projection will give you correct distances only from the central point of the projection, Ian. Between any other pair of points the answer will only be approximate.

You are better off using a good projection adapted for Tunisia. It is such a small region that you can limit the distance errors to a few hundred parts per million. For example, a TM projection with central meridian at 9.55 degrees and scale factor of 0.99977 has a maximum scale distortion of +-220 parts per million. An azimuthal equidistant projection centered in Tunisia (9.55 degrees east, 33.8 degrees north) has a larger maximum scale distortion of +-320 parts per million. Moreover, it is not conformal, implying the distortion varies with the orientation of the points.
--Bill Huber
Quantitative Decisions (http://www.quantdec.com )

I'm going to follow his advice and create a Transverse Mercator projection onver Tunisia. As he states, the distortion should be minimal given the small size of Tunisia.