st: ST: vincenty formula for Africa

classic Classic list List threaded Threaded
9 messages Options
Reply | Threaded
Open this post in threaded view
|

st: ST: vincenty formula for Africa

Laura Platchkov
Dear all,

I am trying to find out abotu those distances, and apparently somethign went wrong with the conversion and the distmatch formula, so I am thinkin abotu trying the vincenty.

I am not experienced wiht programming,... may I ask if someone knows  how to modify the vincenty command for Africa, using Clarke 1880 ellipsoid value?

From:http://www.movable-type.co.uk/scripts/latlong-vincenty.html

The most accurate and widely used globally-applicable model for the earth ellipsoid is WGS-84, used in this script. Other ellipsoids offering a better fit to the local geoid include Airy (1830) in the UK, International 1924 in much of Europe, Clarke (1880) in Africa, and GRS-67 in South America. America (NAD83) and Australia (GDA) use GRS-80, functionally equivalent to the WGS-84 ellipsoid.
  WGS-84 a = 6 378 137 m (±2 m) b = 6 356 752.3142 m f = 1 / 298.257223563
  GRS-80 a = 6 378 137 m b = 6 356 752.3141 m f = 1 / 298.257222101
  Airy (1830) a = 6 377 563.396 m b = 6 356 256.909 m f = 1 / 299.3249646
  Int’l 1924 a = 6 378 388 m b = 6 356 911.946 m f = 1 / 297
  Clarke (1880) a = 6 378 249.145 m b = 6 356 514.86955 m f = 1 / 293.465
 


Laura


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|

RE: st: ST: vincenty formula for Africa

Roy Wada
> I am trying to find out abotu those distances, and apparently somethign went wrong with the
> conversion and the distmatch formula, so I am thinkin abotu trying the vincenty.


You can try -vincenty- by Austin Nichols from ssc. I am sure he will be able to
help you integrate it into one of his codes.
 
For the record, the outputs produced by -distmatch- has been validated by Austin.
 
Roy
 
_________________________________________________________________
Your E-mail and More On-the-Go. Get Windows Live Hotmail Free.
http://clk.atdmt.com/GBL/go/171222985/direct/01/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|

RE: st: ST: vincenty formula for Africa

Roy Wada
> For the record, the outputs produced by -distmatch- has been validated by Austin.
>
> Roy

 
I was able to say that much sooner than I thought.
 
For those who of us who are not as fortunate, you can still validate your distance
calculations at the following websites:
 
http://www.csgnetwork.com/gpsdistcalc.html
http://gpsvisualizer.com/calculators
 
Roy
 
_________________________________________________________________
Ready for Fall shows? Use Bing to find helpful ratings and reviews on digital tv's.
http://www.bing.com/shopping/search?q=digital+tv's&form=MSHNCB&publ=WLHMTAG&crea=TEXT_MSHNCB_Vertical_Shopping_DigitalTVs_1x1
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|

Re: st: ST: vincenty formula for Africa

Austin Nichols
In reply to this post by Roy Wada
Roy--
Please do not quote me as having validated your output--I have not
done anything of the kind. Quite the contrary--I have requested that
you make a correction in your package.

On Sun, Sep 13, 2009 at 2:04 PM, Roy Wada <[hidden email]> wrote:

>> I am trying to find out abotu those distances, and apparently somethign went wrong with the
>> conversion and the distmatch formula, so I am thinkin abotu trying the vincenty.
>
>
> You can try -vincenty- by Austin Nichols from ssc. I am sure he will be able to
> help you integrate it into one of his codes.
>
> For the record, the outputs produced by -distmatch- has been validated by Austin.
>
> Roy
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|

Re: st: ST: vincenty formula for Africa

Austin Nichols
In reply to this post by Roy Wada
Roy--
Please do not quote me as having validated your output--I have not
done anything of the kind. Quite the contrary--I have requested that
you make a correction in your package.

On Sun, Sep 13, 2009 at 2:56 PM, Roy Wada <[hidden email]> wrote:

>> For the record, the outputs produced by -distmatch- has been validated by Austin.
>>
>> Roy
>
>
> I was able to say that much sooner than I thought.
>
> For those who of us who are not as fortunate, you can still validate your distance
> calculations at the following websites:
>
> http://www.csgnetwork.com/gpsdistcalc.html
> http://gpsvisualizer.com/calculators
>
> Roy
>
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|

RE: st: ST: vincenty formula for Africa

Roy Wada
In reply to this post by Austin Nichols
> Roy--
> Please do not quote me as having validated your output--I have not
> done anything of the kind. Quite the contrary--I have requested that
> you make a correction in your package.

Should someone be able to publicly request a correction in someone else's package
when it clearly does not affect that person? Is there a gross scientific falsehood
not consistent with good research? Have I caused an unconscionable loss or damages
to someone as a result of my package? Did I steal codes out of someone's programs
and misrepresented myself as the author?
 
I am trying to understand Austin's position here.
 
What it says in the help file is not important to me.
 
If Austin thinks it will greatly improve the package, then I am open to a revision.
He can post a revised version help file here or send it to me privately.
 
The current version says the following:

    distmatch provides a fast and easy way to perform distance matching based
    on latitudes and longitudes. It mathces location identifiers, attributes,
    or distances from one to another. It also counts the number of
    neighboring locations within a given area, either in circles or bands of
    concentric circles.
 
    Outputs produced by distmatch are typically used in studies of geocoded
    data. These outputs can also be produced by non-Stata softwares such as
    ArcGIS and ArcView.
 
    distmatch implements haversine formula for distance. The default is in
    miles.
 
    Distance matching is computationally intensive. Observations of 3,000 may
    take several minutes to complete.  Other methods typically take days if
    not weeks and requires extensive user-involvement.

Roy
_________________________________________________________________
Ready for Fall shows? Use Bing to find helpful ratings and reviews on digital tv's.
http://www.bing.com/shopping/search?q=digital+tv's&form=MSHNCB&publ=WLHMTAG&crea=TEXT_MSHNCB_Vertical_Shopping_DigitalTVs_1x1
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|

Re: st: ST: vincenty formula for Africa

Austin Nichols
In reply to this post by Laura Platchkov
All--
In response to a private query, I sent Laura a draft revision of
-vincenty-early this morning that will use any of these ellipsoids; in
Stata, type
  net from http://www-personal.umich.edu/~nicholsa/stata
to get it.  She told me she was not using Stata today, so she would
try it out another day.

I doubt the issue is really one of choice of ellipsoidal or spherical
approximation; I suspect she is computing distances from farms to
random points (one on each lake), judging from other questions on the
Statalist.  As I told her privately, that will often be worse than
using the centroids of lakes, but if she has the shapefile she can
compute the distance to the nearest vertex.  That approach would also
work for rivers, if she had a shapefile for rivers.

Perhaps her calculations are failing because the latitudes and
longitudes are not in the same format in both datasets, i.e. signed
decimal degrees--I have not investigated that possibility.

On Sun, Sep 13, 2009 at 8:22 AM, Laura Platchkov <[hidden email]> wrote:

> Dear all,
>
> I am trying to find out abotu those distances, and apparently somethign went wrong with the conversion and the distmatch formula, so I am thinkin abotu trying the vincenty.
>
> I am not experienced wiht programming,... may I ask if someone knows  how to modify the vincenty command for Africa, using Clarke 1880 ellipsoid value?
>
> From:http://www.movable-type.co.uk/scripts/latlong-vincenty.html
>
> The most accurate and widely used globally-applicable model for the earth ellipsoid is WGS-84, used in this script. Other ellipsoids offering a better fit to the local geoid include Airy (1830) in the UK, International 1924 in much of Europe, Clarke (1880) in Africa, and GRS-67 in South America. America (NAD83) and Australia (GDA) use GRS-80, functionally equivalent to the WGS-84 ellipsoid.
>        WGS-84  a = 6 378 137 m (±2 m)  b = 6 356 752.3142 m    f = 1 / 298.257223563
>        GRS-80  a = 6 378 137 m         b = 6 356 752.3141 m    f = 1 / 298.257222101
>        Airy (1830)     a = 6 377 563.396 m     b = 6 356 256.909 m     f = 1 / 299.3249646
>        Int’l 1924      a = 6 378 388 m         b = 6 356 911.946 m     f = 1 / 297
>        Clarke (1880)   a = 6 378 249.145 m     b = 6 356 514.86955 m   f = 1 / 293.465
>
> Laura

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|

Re: st: ST: vincenty formula for Africa

Austin Nichols
In reply to this post by Roy Wada
Roy--
My objection was to the last sentence, as I made clear in my first post:
  "Other methods typically take days if
   not weeks and requires extensive user-involvement."
in the last quoted paragraph seems clearly false to me.

Why not simply remove that sentence and leave the package otherwise unchanged?

And yes, someone should be able to request a correction of an author.
More reasonable requests may be honored, slightly less reasonable
requests may be politely refused, and unreasonable requests may be
ignored, at the discretion of the author, of course.

Evidently we have a deep and abiding disagreement as to what
constitutes "reasonable" (possibly along several dimensions).

I certainly find you quoting me as having validated your package as an
unreasonable statement, and a gross misrepresentation.  I did nothing
of the sort.

On Sun, Sep 13, 2009 at 3:43 PM, Roy Wada <[hidden email]> wrote:

> Should someone be able to publicly request a correction in someone else's package
> when it clearly does not affect that person? Is there a gross scientific falsehood
> not consistent with good research? Have I caused an unconscionable loss or damages
> to someone as a result of my package? Did I steal codes out of someone's programs
> and misrepresented myself as the author?
>
> I am trying to understand Austin's position here.
>
> What it says in the help file is not important to me.
>
> If Austin thinks it will greatly improve the package, then I am open to a revision.
> He can post a revised version help file here or send it to me privately.
>
> The current version says the following:
>
>    distmatch provides a fast and easy way to perform distance matching based
>    on latitudes and longitudes. It mathces location identifiers, attributes,
>    or distances from one to another. It also counts the number of
>    neighboring locations within a given area, either in circles or bands of
>    concentric circles.
>
>    Outputs produced by distmatch are typically used in studies of geocoded
>    data. These outputs can also be produced by non-Stata softwares such as
>    ArcGIS and ArcView.
>
>    distmatch implements haversine formula for distance. The default is in
>    miles.
>
>    Distance matching is computationally intensive. Observations of 3,000 may
>    take several minutes to complete.  Other methods typically take days if
>    not weeks and requires extensive user-involvement.
>
> Roy
>

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
Reply | Threaded
Open this post in threaded view
|

RE: st: ST: vincenty formula for Africa

Roy Wada
Done. That last sentence is out.
 
I take it you don't want to see me put your name in the acknowledgement section of
the help file?
 
Roy
 
 

> I certainly find you quoting me as having validated your package as an
> unreasonable statement, and a gross misrepresentation. I did nothing
> of the sort.
>
> On Sun, Sep 13, 2009 at 3:43 PM, Roy Wada wrote:
>> Should someone be able to publicly request a correction in someone else's package
>> when it clearly does not affect that person? Is there a gross scientific falsehood
>> not consistent with good research? Have I caused an unconscionable loss or damages
>> to someone as a result of my package? Did I steal codes out of someone's programs
>> and misrepresented myself as the author?
>>
>> I am trying to understand Austin's position here.
_________________________________________________________________
Ready for Fall shows? Use Bing to find helpful ratings and reviews on digital tv's.
http://www.bing.com/shopping/search?q=digital+tv's&form=MSHNCB&publ=WLHMTAG&crea=TEXT_MSHNCB_Vertical_Shopping_DigitalTVs_1x1
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/