0

Let's say I got $3$ points: $(lat_1, lon_1), (lat_2, lon_2), (lat_3, lon_3)$ and their distances to the point $p$ I want to know are $d_1, d_2, d_3$.

I can also know the distance between the points using Haversine Formula. How can I calculate the latitude and longitude of the point $p$?

Alex M.
  • 35,207

2 Answers2

1

Finally found the solution (didn't know the gis.stackexchange site):

https://gis.stackexchange.com/questions/66/trilateration-using-3-latitude-and-longitude-points-and-3-distances/415#415

Just have to pass the latitude and longitude coordinates to the cartesian ECEF, an then resolve an equation system having 3 circles that had as center one of each points and the distance to the point as radius.

Basically what I was missing was passing to the cartesian ECEF.

0

Since this is very far from what I use to do, take it with care

This problem looks very much to me as a data reconciliation problem (or least-square fit) and, so, I suggest to use the most natural variables.

Name $x$ the latitude and and $y$ the longitude of the point $p$ you are looking for. Compute the distances $d_1^*,d_2^*,d_3^*$ from the assumed latitude and longitude of point $p$ to the known points (each of the $d_i^*$ is a function of $x$ and $y$).

Now, define an error function such as $$\Phi(x,y)=(d_1^*-d_1)^2+(d_2^*-d_2)^2+(d_3^*-d_3)^2$$ and the idea is to minimize it and, hopefully, to get $0$ if the measures are exact. If they are not, this will give the best approximation of the location of point $p$.

  • Well I think the error won't be too big because I am working with distances of 100km more or less. But lowering errors it's always a good idea. –  Jun 23 '15 at 11:05
  • More I think about it, more I think that this is a "simple" least-square fit problem. – Claude Leibovici Jun 23 '15 at 11:07