ECEF to lla (lat,lon,alt) in java

dbkoren picture dbkoren · Aug 15, 2013 · Viewed 18.8k times · Source

Ive looked through the posts in the site and haven't found my problem ... as the head line says im trying to convert from ecef to lla .

I'm using this document : Conversion article in the direct formula , not the iterate formula and this site for result comparison : ECEF2LLA

Im developing in java so my code is as follows :

public static final double a = 6378137;
public static final double f = 1/298.257223563;
public static final double b = a*(1-f);
public static final double e = Math.sqrt((Math.pow(a, 2)-Math.pow(b, 2))/Math.pow(a, 2));
public static final double e2 = Math.sqrt((Math.pow(a, 2)-Math.pow(b, 2))/Math.pow(b, 2));
public static double[] ecef2lla(double x , double y , double z){
    double[] lla = {0,0,0};
    double lon,lat,height,N;
    double p = Math.sqrt(Math.pow(x, 2)+Math.pow(y, 2));
    double theta = Math.atan((z*a)/(p*b));
    lon = Math.atan(y/x);
    lon = lon*180/Math.PI;

    lat = Math.atan(((z+Math.pow(e2, 2)*b*Math.pow(Math.sin(theta), 3))/((p-Math.pow(e,2)*a*Math.pow(Math.cos(theta), 3)))));
    lat = (lat*180)/Math.PI;

    N= a/(Math.sqrt(1-Math.pow(e*Math.sin(lat), 2)));
    height = (p/Math.cos(theta)) - N;
    lla[0] = lon;
    lla[1] = lat;
    lla[2] = height;
    return lla;

}

I'm getting wrong height data. I've tried to move to radians and degrees and what not .

Thank you in advance !

Answer

Kyle Melton picture Kyle Melton · Aug 21, 2014

I found this post and was ready to use the "accepted answer" in part of my application but I decided to run a couple of tests first to validate the algorithm. I used sample data generated by an online conversion calculator (http://www.sysense.com/products/ecef_lla_converter/index.html). I got less than stellar results as shown by the output below.

-----Test 1---------
Inputs:   -576793.17, -5376363.47, 3372298.51
Expected: 32.12345, -96.12345, 500.0
Actuals:  32.12306332822881, 83.87654999786477, 486.5472474489361
-----Test 2---------
Inputs:   2297292.91, 1016894.94, -5843939.62
Expected: -66.87654, 23.87654, 1000.0
Actuals:  -66.876230479461, 23.87653991401422, 959.6879360172898

Then I reran the same tests using the code from the following post (https://gist.github.com/klucar/1536194) and got much better results as shown by the output below.

-----Test 1---------
Inputs:   -576793.17, -5376363.47, 3372298.51
Expected: 32.12345, -96.12345, 500.0
Actuals:  32.12345004807767, -96.12345000213524, 499.997958839871
-----Test 2---------
Inputs:   2297292.91, 1016894.94, -5843939.62
Expected: -66.87654, 23.87654, 1000.0
Actuals:  -66.87654001741278, 23.87653991401422, 999.9983866894618

I didn't take the time to find the error in the solution provided in the "accepted answer" but my suggested answer: Use this code...

/*
*
*  ECEF - Earth Centered Earth Fixed
*   
*  LLA - Lat Lon Alt
*
*  ported from matlab code at
*  https://gist.github.com/1536054
*     and
*  https://gist.github.com/1536056
*/

// WGS84 ellipsoid constants
private final double a = 6378137; // radius
private final double e = 8.1819190842622e-2;  // eccentricity

private final double asq = Math.pow(a,2);
private final double esq = Math.pow(e,2);

private double[] ecef2lla(double[] ecef){
  double x = ecef[0];
  double y = ecef[1];
  double z = ecef[2];

  double b = Math.sqrt( asq * (1-esq) );
  double bsq = Math.pow(b,2);
  double ep = Math.sqrt( (asq - bsq)/bsq);
  double p = Math.sqrt( Math.pow(x,2) + Math.pow(y,2) );
  double th = Math.atan2(a*z, b*p);

  double lon = Math.atan2(y,x);
  double lat = Math.atan2( (z + Math.pow(ep,2)*b*Math.pow(Math.sin(th),3) ), (p - esq*a*Math.pow(Math.cos(th),3)) );
  double N = a/( Math.sqrt(1-esq*Math.pow(Math.sin(lat),2)) );
  double alt = p / Math.cos(lat) - N;

  // mod lat to 0-2pi
  lon = lon % (2*Math.PI);

  // correction for altitude near poles left out.

  double[] ret = {lat, lon, alt};

  return ret;
}