aboutsummaryrefslogtreecommitdiff
path: root/src/conversions
diff options
context:
space:
mode:
authorKristian Evers <kristianevers@gmail.com>2020-02-05 07:32:32 +0100
committerKristian Evers <kristianevers@gmail.com>2020-02-06 17:53:11 +0100
commit55112033a4131cc9e5a0f6b2a40cec555d87955a (patch)
treeb5faf4c6cc02578a00cb640c8a57612a20b58610 /src/conversions
parentbd6148db5f5b9a0a13f3dc2d2c35a066c15a57ed (diff)
downloadPROJ-55112033a4131cc9e5a0f6b2a40cec555d87955a.tar.gz
PROJ-55112033a4131cc9e5a0f6b2a40cec555d87955a.zip
cart: Avoid discontinuity at poles in the inverse case
This should avoid issues with numerical stability as uncovered in https://github.com/OSGeo/PROJ/issues/1906. Practically speaking this change isn't going to affect real life scenarios since the position of the center of the Earth is rarely expressed in geodetic coordinates.
Diffstat (limited to 'src/conversions')
-rw-r--r--src/conversions/cart.cpp5
1 files changed, 3 insertions, 2 deletions
diff --git a/src/conversions/cart.cpp b/src/conversions/cart.cpp
index a7817443..537fc29f 100644
--- a/src/conversions/cart.cpp
+++ b/src/conversions/cart.cpp
@@ -165,8 +165,9 @@ static PJ_LPZ geodetic (PJ_XYZ cart, PJ *P) {
if( fabs(lpz.phi) > M_HALFPI ) {
// this happen on non-sphere ellipsoid when x,y,z is very close to 0
// there is no single solution to the cart->geodetic conversion in
- // that case, so arbitrarily pickup phi = 0.
- lpz.phi = 0;
+ // that case, clamp to -90/90 deg and avoid a discontinuous boundary
+ // near the poles
+ lpz.phi = copysign(M_HALFPI, lpz.phi);
}
lpz.lam = atan2 (cart.y, cart.x);
N = normal_radius_of_curvature (P->a, P->es, lpz.phi);