[R-sig-Geo] Distance between two points

[Roger Bivand]

On Fri, 7 Dec 2012, Sarah Goslee wrote:

> Hi Simon,
>
> I've copied this back to the list, as is encouraged.
>
> On Fri, Dec 7, 2012 at 9:41 AM, O'Hanlon, Simon J
> <simon.ohanlon at imperial.ac.uk> wrote:
>> Hi Sarah, thank you.
>>
>> My data points are located in West Africa, so I think a good projection 
>> would be Lambert Azimuthal Equal Area.
>>
>> This does raise one more question for me. I also have some raster data 
>> that I wish to use as covariates which is also in geographic 
>> coordinates. If I try to simply reproject a raster, I think it would 
>> screw up the regular grid. Does it make sense if I convert the lat-long 
>> raster to a SpatialPointsDataFrame and transform this to LAEA so I 
>> could then create a grid in the desired planar coordinates from scratch 
>> and use over() to assign values from the reprojected spatial points 
>> onto the raster?
>
> A reprojected raster is still a raster, and thus a regular grid. Or am
> I missing something?

Yes, a warped raster is a raster, but a reprojected raster will in general 
be a set of irregular points. So resampling is invoved one way or the 
other. Whether one queries the raster as-is with spatial points, and then 
projects the output, or warps the raster doing spatial query with 
projected points will depend a bit on the support(s) of the data sets. 
Maybe see: ?projectRaster in raster for warping.

Roger

>
> Sarah
>
>
>> Thanks again for your help.
>>
>> Cheers,
>>
>> Simon
>>
>> -----Original Message-----
>> From: Sarah Goslee [mailto:sarah.goslee at gmail.com]
>> Sent: 07 December 2012 14:18
>> To: O'Hanlon, Simon J
>> Cc: r-sig-geo at r-project.org
>> Subject: Re: [R-sig-Geo] Distance between two points
>>
>> Precisely. You should use great-circle distances with lat-lon coordinates, rather than Euclidean distance, because the actual length varies with position on the globe.
>>
>> Converting to UTM or something similar is one solution if your points are not too far apart.
>>
>> There are many other R solutions: searching for "great circle distance" at rseek.org will get you quite a list.
>>
>> Sarah
>>
>> On Fri, Dec 7, 2012 at 8:10 AM, O'Hanlon, Simon J <simon.ohanlon at imperial.ac.uk> wrote:
>>> Dear list,
>>> I am using the package geoRglm to do some predictive mapping. There is a function that calculates the distance between observed data points and the prediction locations using a .C call to a function which eventually calculates the length of the hypotenuse between one location and the other given the vertical and horizontal separation distance of those points.
>>>
>>> My question is, is this method of distance-finding incompatible with long-lat style coordinates? Should I first transform my data and prediction locations into something where the unit of measurement is in metres rather than decimal degrees?
>>>
>>> Many thanks,
>>>
>>> Simon
>>>
>
>
> --
> Sarah Goslee
> http://www.functionaldiversity.org
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>

-- 
Roger Bivand
Department of Economics, NHH Norwegian School of Economics,
Helleveien 30, N-5045 Bergen, Norway.
voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no