diff --git a/CMakeLists.txt b/CMakeLists.txt
index 4e12d0013..80652376c 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -100,6 +100,7 @@ set(SUPPORT
   src/core/textstream.cc
   src/core/usasciicodec.cc
   src/core/vector3d.cc
+  src/core/welzl.cc
   src/core/xmlstreamwriter.cc
 )
 if(${QT_VERSION_MAJOR} EQUAL "6")
@@ -193,6 +194,7 @@ set(HEADERS
   src/core/textstream.h
   src/core/usasciicodec.h
   src/core/vector3d.h
+  src/core/welzl.h
   src/core/xmlstreamwriter.h
   src/core/xmltag.h
 )
diff --git a/GPSBabel.pro b/GPSBabel.pro
index 00bc8d03c..924bd3d2c 100644
--- a/GPSBabel.pro
+++ b/GPSBabel.pro
@@ -111,6 +111,7 @@ SUPPORT = route.cc waypt.cc filter_vecs.cc util.cc vecs.cc mkshort.cc \
           src/core/textstream.cc \
           src/core/usasciicodec.cc \
           src/core/vector3d.cc \
+          src/core/welzl.cc \
           src/core/xmlstreamwriter.cc
 
 versionAtLeast(QT_VERSION, 6.0): SUPPORT += src/core/codecdevice.cc
@@ -190,6 +191,7 @@ HEADERS =  \
 	src/core/textstream.h \
 	src/core/usasciicodec.h \
 	src/core/vector3d.h \
+        src/core/welzl.h \
 	src/core/xmlstreamwriter.h \
 	src/core/xmltag.h
 
diff --git a/kml.cc b/kml.cc
index 695b30d65..94f5058a0 100644
--- a/kml.cc
+++ b/kml.cc
@@ -49,6 +49,7 @@
 #include "src/core/datetime.h"          // for DateTime
 #include "src/core/file.h"              // for File
 #include "src/core/logging.h"           // for Warning, Fatal
+#include "src/core/welzl.h"
 #include "src/core/xmlstreamwriter.h"   // for XmlStreamWriter
 #include "src/core/xmltag.h"            // for xml_findfirst, xml_tag, fs_xml, xml_attribute, xml_findnext
 #include "units.h"                      // for fmt_setunits, fmt_speed, fmt_altitude, fmt_distance, units_aviation, units_metric, units_nautical, units_statute
@@ -364,7 +365,7 @@ void KmlFormat::rd_deinit()
 void KmlFormat::wr_init(const QString& fname)
 {
   char u = 's';
-  waypt_init_bounds(&kml_bounds);
+  kml_points.clear();
   kml_time_min = QDateTime();
   kml_time_max = QDateTime();
 
@@ -814,7 +815,7 @@ void KmlFormat::kml_recompute_time_bounds(const Waypoint* waypointp)
 
 void KmlFormat::kml_add_to_bounds(const Waypoint* waypointp)
 {
-  waypt_add_to_bounds(&kml_bounds, waypointp);
+  kml_points.push_back(gpsbabel::NVector(waypointp->latitude, waypointp->longitude));
   kml_recompute_time_bounds(waypointp);
 }
 
@@ -1682,6 +1683,8 @@ void KmlFormat::kml_write_AbstractView()
     route_disp_all(nullptr, nullptr, kml_add_to_bounds_lambda);
   }
 
+  gpsbabel::Circle bounds = gpsbabel::Welzl::welzl(kml_points);
+
   writer->writeStartElement(QStringLiteral("LookAt"));
 
   if (kml_time_min.isValid() || kml_time_max.isValid()) {
@@ -1704,25 +1707,19 @@ void KmlFormat::kml_write_AbstractView()
     writer->writeEndElement(); // Close gx:TimeSpan tag
   }
 
-// If our BB spans the antemeridian, flip sign on one.
-// This doesn't make our BB optimal, but it at least prevents us from
-// zooming to the wrong hemisphere.
-  if (kml_bounds.min_lon * kml_bounds.max_lon < 0) {
-    kml_bounds.min_lon = -kml_bounds.max_lon;
-  }
-
-  writer->writeTextElement(QStringLiteral("longitude"), QString::number((kml_bounds.min_lon + kml_bounds.max_lon) / 2, 'f', precision));
-  writer->writeTextElement(QStringLiteral("latitude"), QString::number((kml_bounds.min_lat + kml_bounds.max_lat) / 2, 'f', precision));
+  writer->writeTextElement(QStringLiteral("longitude"), QString::number(bounds.center().longitude(), 'f', precision));
+  writer->writeTextElement(QStringLiteral("latitude"), QString::number(bounds.center().latitude(), 'f', precision));
 
   // It turns out the length of the diagonal of the bounding box gives us a
   // reasonable guess for setting the camera altitude.
-  double bb_size = gcgeodist(kml_bounds.min_lat, kml_bounds.min_lon,
-                             kml_bounds.max_lat, kml_bounds.max_lon);
+  // TODO: we can easily calculate a range such that the camera sees a bit beyond the points
+  // for the cases that the range is reasonable.
+  double bb_size = 2.0 * bounds.radius();
   // Clamp bottom zoom level.  Otherwise, a single point zooms to grass.
   if (bb_size < 1000) {
     bb_size = 1000;
   }
-  writer->writeTextElement(QStringLiteral("range"), QString::number(bb_size * 1.3, 'f', 6));
+  writer->writeTextElement(QStringLiteral("range"), QString::number(bb_size * 1.3, 'f', 3));
 
   writer->writeEndElement(); // Close LookAt tag
 }
diff --git a/kml.h b/kml.h
index d324b167f..2533ba4cc 100644
--- a/kml.h
+++ b/kml.h
@@ -33,6 +33,8 @@
 #include "format.h"
 #include "src/core/datetime.h"          // for DateTime
 #include "src/core/file.h"              // for File
+#include "src/core/nvector.h"
+#include "src/core/welzl.h"
 #include "src/core/xmlstreamwriter.h"   // for XmlStreamWriter
 #include "xmlgeneric.h"                 // for cb_cdata, cb_end, cb_start, xg_callback, xg_string, xg_cb_type, xml_deinit, xml_ignore_tags, xml_init, xml_read, xg_tag_mapping
 
@@ -216,7 +218,7 @@ class KmlFormat : public Format
   gpsbabel::XmlStreamWriter* writer{nullptr};
 
   int realtime_positioning{};
-  bounds kml_bounds{};
+  gpsbabel::PointContainer kml_points;
   gpsbabel::DateTime kml_time_max;
   gpsbabel::DateTime kml_time_min;
 
diff --git a/reference/LineStyles.kml b/reference/LineStyles.kml
index 77c81ca37..79ad33378 100644
--- a/reference/LineStyles.kml
+++ b/reference/LineStyles.kml
@@ -3,9 +3,9 @@
   <Document>
     <name>GPS device</name>
     <LookAt>
-      <longitude>-122.275670</longitude>
-      <latitude>37.523278</latitude>
-      <range>5908.813619</range>
+      <longitude>-122.276195</longitude>
+      <latitude>37.523328</latitude>
+      <range>5678.471</range>
     </LookAt>
     <!-- Normal track style -->
     <Style id="track_n">
@@ -159,7 +159,7 @@
       </Placemark>
       <Placemark>
         <name>NOTE 1</name>
-        <description>12 test lines - see notes with GPX sample!!
four line widths for solid red
two line widths for dashed red
seven colors for medium width
Note that line names have nothing at all to do with line style attributes! These are named for convenient cross-reference,
   but keep in mind that &quot;red med&quot; is a line label and &quot;Red Medium&quot; is a style label.
Waypoints are generated by Topo to exactly match the first track point.
Apparently only GPX and KML output from GPSBabel retain track descriptions??</description>
+        <description>12 test lines - see notes with GPX sample!!
four line widths for solid red
two line widths for dashed red
seven colors for medium width
Note that line names have nothing at all to do with line style attributes!These are named for convenient cross-reference,
   but keep in mind that &quot;red med&quot; is a line label and &quot;Red Medium&quot; is a style label.
Waypoints are generated by Topo to exactly match the first track point.
Apparently only GPX and KML output from GPSBabel retain track descriptions??</description>
         <styleUrl>#waypoint</styleUrl>
         <Point>
           <coordinates>-122.277670,37.522060</coordinates>
diff --git a/reference/bounds-test.kml b/reference/bounds-test.kml
index 65298dd15..3c2023c36 100644
--- a/reference/bounds-test.kml
+++ b/reference/bounds-test.kml
@@ -3,9 +3,9 @@
   <Document>
     <name>GPS device</name>
     <LookAt>
-      <longitude>-117.144015</longitude>
-      <latitude>36.438270</latitude>
-      <range>87257.203848</range>
+      <longitude>-117.155510</longitude>
+      <latitude>36.443045</latitude>
+      <range>68977.917</range>
     </LookAt>
     <!-- Normal route style -->
     <Style id="route_n">
diff --git a/reference/earth-expertgps-track.kml b/reference/earth-expertgps-track.kml
index 207b78b3f..6bbf01d9b 100644
--- a/reference/earth-expertgps-track.kml
+++ b/reference/earth-expertgps-track.kml
@@ -7,9 +7,9 @@
         <begin>2001-06-02T00:18:13Z</begin>
         <end>2002-05-25T19:05:57Z</end>
       </gx:TimeSpan>
-      <longitude>-81.356770</longitude>
-      <latitude>36.257086</latitude>
-      <range>2979727.763965</range>
+      <longitude>-82.179826</longitude>
+      <latitude>36.694478</latitude>
+      <range>2974869.803</range>
     </LookAt>
     <!-- Normal route style -->
     <Style id="route_n">
diff --git a/reference/earth-expertgps.kml b/reference/earth-expertgps.kml
index fd0b5347a..6c474017b 100644
--- a/reference/earth-expertgps.kml
+++ b/reference/earth-expertgps.kml
@@ -7,9 +7,9 @@
         <begin>2001-06-02T00:18:13Z</begin>
         <end>2002-05-25T19:05:57Z</end>
       </gx:TimeSpan>
-      <longitude>-81.356770</longitude>
-      <latitude>36.257086</latitude>
-      <range>2979727.763965</range>
+      <longitude>-82.179826</longitude>
+      <latitude>36.694478</latitude>
+      <range>2974869.803</range>
     </LookAt>
     <!-- Normal route style -->
     <Style id="route_n">
diff --git a/reference/earth-gc.kml b/reference/earth-gc.kml
index 94d1bfde0..da0f17ed0 100644
--- a/reference/earth-gc.kml
+++ b/reference/earth-gc.kml
@@ -7,9 +7,9 @@
         <begin>2002-08-15T07:00:00Z</begin>
         <end>2003-06-29T07:00:00Z</end>
       </gx:TimeSpan>
-      <longitude>-79.930833</longitude>
-      <latitude>41.027500</latitude>
-      <range>2109328.437865</range>
+      <longitude>-80.472308</longitude>
+      <latitude>41.234430</latitude>
+      <range>2106968.144</range>
     </LookAt>
     <!-- Normal waypoint style -->
     <Style id="waypoint_n">
@@ -309,7 +309,7 @@ PICTURES of points and of the panels will follow soon. YOU CAN ONLY LOG ONE POIN
 Good luck!]]></value>
           </Data>
           <Data name="gc_logs">
-            <value><![CDATA[<p><b>Found it</b> by Christopher R & Pooh B 2005-07-12<br />This marker is not in Quebec but it is a Geodesic marker in Clarenville, Newfoundland, Canada!
+            <value><![CDATA[<p><b>Found it</b> by Christopher R & Pooh B2005-07-12<br />This marker is not in Quebec but it is a Geodesic marker in Clarenville, Newfoundland, Canada!
 
 Found this one while hunting a traditional cache and thought of this cache right away! 
 
@@ -319,7 +319,7 @@ Smiles Pooh Bear
 
 Ce marqueur n'est pas au Québec mais c'est un marqueur géodésique dans Clarenville, Terre-Neuve, Canada!  
 
-A trouvé celui-ci tandis que chasse une cachette traditionnelle et pensé à cette cachette tout de suite!  Elle est située sur la montagne nue dans Clarenville - il y a aactually deux marqueurs à moins de 15 pieds d'un des autres sur la montagne nue...  Ours De Pooh De Sourires</p><p><b>Found it</b> by TravelBen 2005-06-26<br />[:D] 14h22
+A trouvé celui-ci tandis que chasse une cachette traditionnelle et pensé à cette cachette tout de suite!  Elle est située sur la montagne nue dans Clarenville - il y a aactually deux marqueurs à moins de 15 pieds d'un des autres sur la montagne nue...  Ours De Pooh De Sourires</p><p><b>Found it</b> by TravelBen2005-06-26<br />[:D] 14h22
 
 Marqueur du Service de la Géodégie (c'est bien un &quot;g&quot; pas un &quot;s&quot;) du Québec.
 
@@ -329,7 +329,7 @@ UTM: 18T E 582877 N 5033250
 
 Ce marqueur se trouve dans le ville de Senneville, sur un monument décrivant une page d'histoire du Québec, sur le bas côté avant droit.
 
-Près de la cache:  Exo-07 La Jumelle de Loudiver (GCP3VE)</p><p><b>Found it</b> by etasse 2005-06-03<br />MRN marker 94K4731 in Gatineau, QC. corner of Du Rhone and Gatineau Ave.
+Près de la cache:  Exo-07 La Jumelle de Loudiver (GCP3VE)</p><p><b>Found it</b> by etasse2005-06-03<br />MRN marker 94K4731 in Gatineau, QC. corner of Du Rhone and Gatineau Ave.
 
 Position Average
 N 45° 29.5247  W 075° 43.0049  59.49m
@@ -341,7 +341,7 @@ UTM 18T 0443996 5037868
 
 This pole has everything:  An underground cable warning, a geodesic mark, a bus stop and a garage sale sign.
 
-Judging by the coordinates it looks like the coords should be 45°29'31.5&quot; -75°43'0&quot;  I placed the GPS antenna right against the marker, to no avail.</p><p><b>Found it</b> by Katou 2005-06-03<br />Un bo point géodésique a Lotbinière..en allant faire une nouvelle cache a l'île richelieu ;-)</p><p><b>Found it</b> by Gps_Gulliver&DauphinBleu 2005-05-29<br />Point Geodesique situe near Port de Plaisance de Longueuil
+Judging by the coordinates it looks like the coords should be 45°29'31.5&quot; -75°43'0&quot;  I placed the GPS antenna right against the marker, to no avail.</p><p><b>Found it</b> by Katou2005-06-03<br />Un bo point géodésique a Lotbinière..en allant faire une nouvelle cache a l'île richelieu ;-)</p><p><b>Found it</b> by Gps_Gulliver&DauphinBleu2005-05-29<br />Point Geodesique situe near Port de Plaisance de Longueuil
 sur le bord du fleuve st-laurent.
 Il y a des sentiers et une grande piste cyclable
 Enjoy !</p>]]></value>
@@ -426,7 +426,7 @@ Now that it's intuitively obvious to even the most casual observer where the cac
 <img SRC="http://www.mtgc.org/mtgc_member-banner.gif" WIDTH="500" HEIGHT="40" ALT="Member of Middle Tennessee GeoCachers Club [www.mtgc.org]" BORDER="0"></a></p>]]></value>
           </Data>
           <Data name="gc_logs">
-            <value><![CDATA[<p><b>Found it</b> by littlepod 2005-07-03<br />Enjoyed the puzzle. We seemed to be about 50ft off though. TFTC.</p><p><b>Write note</b> by robertlipe 2005-04-29<br />TB Drop to show he's hanging out in Nashville until we blast off for Geowoodstock in a few weeks.</p><p><b>Found it</b> by Big Bumblebee 2005-04-18<br />Found it a while ago.  Thanks.</p><p><b>Write note</b> by robertlipe 2005-03-27<br />I had to renew my permit with the CDC and in doing so, I trolled out here verified that the infectious ooze is fully within specification and industry accepted tolerance.   Ooze On!</p><p><b>Found it</b> by Virtual Babe 2004-12-27<br />This was a great cache, however on this day I considered it a FIFM cache (Fun, Invigorating, Frustrating and Maddening), especially when the cache was not replaced in the proper spot by the previous cacher!  Thanks anyway!!</p><p><b>Write note</b> by robertlipe 2004-01-12<br />I got a complaint from the CDC about oozy rat this weekend.     I went out tonight in the dark and verified that the infectious ooze is fully within specification and industry accepted tolerance. (Although I realize now I did misstate the cache container to the reporting officer when confronted.   It's, uuuuh, smaller than I said.)</p><p><b>Write note</b> by robertlipe 2003-10-04<br />In the expectation that this cache will get some traffic in the next 48 hours, Ryan and I checked it earlier today.   The Rat is Oozing just as we planned it.</p><p><b>Write note</b> by robertlipe 2003-07-03<br />It won't earn him a smiley face, but I've confirmed that rickrich would have indeed sunk the battleship!      Thanx for playing. You get a copy of the home game and some rice-a-roni...</p>]]></value>
+            <value><![CDATA[<p><b>Found it</b> by littlepod2005-07-03<br />Enjoyed the puzzle. We seemed to be about 50ft off though. TFTC.</p><p><b>Write note</b> by robertlipe2005-04-29<br />TB Drop to show he's hanging out in Nashville until we blast off for Geowoodstock in a few weeks.</p><p><b>Found it</b> by Big Bumblebee2005-04-18<br />Found it a while ago.  Thanks.</p><p><b>Write note</b> by robertlipe2005-03-27<br />I had to renew my permit with the CDC and in doing so, I trolled out here verified that the infectious ooze is fully within specification and industry accepted tolerance.   Ooze On!</p><p><b>Found it</b> by Virtual Babe2004-12-27<br />This was a great cache, however on this day I considered it a FIFM cache (Fun, Invigorating, Frustrating and Maddening), especially when the cache was not replaced in the proper spot by the previous cacher!  Thanks anyway!!</p><p><b>Write note</b> by robertlipe2004-01-12<br />I got a complaint from the CDC about oozy rat this weekend.     I went out tonight in the dark and verified that the infectious ooze is fully within specification and industry accepted tolerance. (Although I realize now I did misstate the cache container to the reporting officer when confronted.   It's, uuuuh, smaller than I said.)</p><p><b>Write note</b> by robertlipe2003-10-04<br />In the expectation that this cache will get some traffic in the next 48 hours, Ryan and I checked it earlier today.   The Rat is Oozing just as we planned it.</p><p><b>Write note</b> by robertlipe2003-07-03<br />It won't earn him a smiley face, but I've confirmed that rickrich would have indeed sunk the battleship!      Thanx for playing. You get a copy of the home game and some rice-a-roni...</p>]]></value>
           </Data>
         </ExtendedData>
         <Point>
diff --git a/reference/realtime.kml b/reference/realtime.kml
index bca161447..a55e7c2d8 100644
--- a/reference/realtime.kml
+++ b/reference/realtime.kml
@@ -9,7 +9,7 @@
       </gx:TimeSpan>
       <longitude>-32.056227</longitude>
       <latitude>30.474664</latitude>
-      <range>1728.780603</range>
+      <range>1726.846</range>
     </LookAt>
     <!-- Normal track style -->
     <Style id="track_n">
diff --git a/reference/track/bounds-test-track.kml b/reference/track/bounds-test-track.kml
index f0ec3641c..b005d828b 100644
--- a/reference/track/bounds-test-track.kml
+++ b/reference/track/bounds-test-track.kml
@@ -3,9 +3,9 @@
   <Document>
     <name>GPS device</name>
     <LookAt>
-      <longitude>-117.144015</longitude>
-      <latitude>36.438270</latitude>
-      <range>87257.203848</range>
+      <longitude>-117.155510</longitude>
+      <latitude>36.443045</latitude>
+      <range>68977.917</range>
     </LookAt>
     <!-- Normal track style -->
     <Style id="track_n">
diff --git a/reference/track/gpx_garmin_extensions-kml_track.kml b/reference/track/gpx_garmin_extensions-kml_track.kml
index 192d1d41f..f56a45a7d 100644
--- a/reference/track/gpx_garmin_extensions-kml_track.kml
+++ b/reference/track/gpx_garmin_extensions-kml_track.kml
@@ -9,7 +9,7 @@
       </gx:TimeSpan>
       <longitude>-0.035329</longitude>
       <latitude>51.506248</latitude>
-      <range>1300.000000</range>
+      <range>1300.000</range>
     </LookAt>
     <!-- Normal track style -->
     <Style id="track_n">
diff --git a/reference/track/gtrnctr_power-kml.kml b/reference/track/gtrnctr_power-kml.kml
index 10f614624..60d4df000 100644
--- a/reference/track/gtrnctr_power-kml.kml
+++ b/reference/track/gtrnctr_power-kml.kml
@@ -7,11 +7,11 @@
         <begin>2010-05-28T01:16:36Z</begin>
         <end>2010-05-28T02:41:44Z</end>
       </gx:TimeSpan>
-      <longitude>-122.139608</longitude>
-      <latitude>37.382794</latitude>
-      <range>19819.321245</range>
+      <longitude>-122.131432</longitude>
+      <latitude>37.382993</latitude>
+      <range>18292.113</range>
     </LookAt>
-<!-- Normal track style -->
+    <!-- Normal track style -->
     <Style id="track_n">
       <IconStyle>
         <scale>.5</scale>
@@ -23,7 +23,7 @@
         <scale>0</scale>
       </LabelStyle>
     </Style>
-<!-- Highlighted track style -->
+    <!-- Highlighted track style -->
     <Style id="track_h">
       <IconStyle>
         <scale>1.2</scale>
@@ -42,7 +42,7 @@
         <styleUrl>#track_h</styleUrl>
       </Pair>
     </StyleMap>
-<!-- Normal multiTrack style -->
+    <!-- Normal multiTrack style -->
     <Style id="multiTrack_n">
       <IconStyle>
         <Icon>
@@ -54,7 +54,7 @@
         <width>6</width>
       </LineStyle>
     </Style>
-<!-- Highlighted multiTrack style -->
+    <!-- Highlighted multiTrack style -->
     <Style id="multiTrack_h">
       <IconStyle>
         <scale>1.2</scale>
@@ -77,7 +77,7 @@
         <styleUrl>#multiTrack_h</styleUrl>
       </Pair>
     </StyleMap>
-<!-- Normal waypoint style -->
+    <!-- Normal waypoint style -->
     <Style id="waypoint_n">
       <IconStyle>
         <Icon>
@@ -85,7 +85,7 @@
         </Icon>
       </IconStyle>
     </Style>
-<!-- Highlighted waypoint style -->
+    <!-- Highlighted waypoint style -->
     <Style id="waypoint_h">
       <IconStyle>
         <scale>1.2</scale>
@@ -112,13 +112,13 @@
     </Style>
     <Schema id="schema">
       <gx:SimpleArrayField name="heartrate" type="int">
-          <displayName>Heart Rate</displayName>
+        <displayName>Heart Rate</displayName>
       </gx:SimpleArrayField>
       <gx:SimpleArrayField name="cadence" type="int">
-          <displayName>Cadence</displayName>
+        <displayName>Cadence</displayName>
       </gx:SimpleArrayField>
       <gx:SimpleArrayField name="power" type="float">
-          <displayName>Power</displayName>
+        <displayName>Power</displayName>
       </gx:SimpleArrayField>
     </Schema>
     <Folder>
diff --git a/reference/track/segmented_tracks-track.kml b/reference/track/segmented_tracks-track.kml
index 9e3e70ee2..31d7c829d 100644
--- a/reference/track/segmented_tracks-track.kml
+++ b/reference/track/segmented_tracks-track.kml
@@ -7,9 +7,9 @@
         <begin>2007-07-27T05:24:05Z</begin>
         <end>2007-07-27T05:35:00Z</end>
       </gx:TimeSpan>
-      <longitude>-86.841461</longitude>
+      <longitude>-86.841519</longitude>
       <latitude>35.831217</latitude>
-      <range>1944.793167</range>
+      <range>1901.182</range>
     </LookAt>
     <!-- Normal track style -->
     <Style id="track_n">
diff --git a/reference/track/segmented_tracks.kml b/reference/track/segmented_tracks.kml
index 4427c47df..ad2faef0f 100644
--- a/reference/track/segmented_tracks.kml
+++ b/reference/track/segmented_tracks.kml
@@ -7,9 +7,9 @@
         <begin>2007-07-27T05:24:05Z</begin>
         <end>2007-07-27T05:35:00Z</end>
       </gx:TimeSpan>
-      <longitude>-86.841461</longitude>
+      <longitude>-86.841519</longitude>
       <latitude>35.831217</latitude>
-      <range>1944.793167</range>
+      <range>1901.182</range>
     </LookAt>
     <!-- Normal track style -->
     <Style id="track_n">
diff --git a/reference/track/tracks~gpx.kml b/reference/track/tracks~gpx.kml
index 0400a7e7e..f9593a684 100644
--- a/reference/track/tracks~gpx.kml
+++ b/reference/track/tracks~gpx.kml
@@ -7,9 +7,9 @@
         <begin>2002-05-25T17:06:21.250Z</begin>
         <end>2002-05-25T19:05:57.980Z</end>
       </gx:TimeSpan>
-      <longitude>-91.600158</longitude>
-      <latitude>30.054150</latitude>
-      <range>3611.461364</range>
+      <longitude>-91.600224</longitude>
+      <latitude>30.055900</latitude>
+      <range>3264.506</range>
     </LookAt>
     <!-- Normal track-none style -->
     <Style id="track-none_n">
diff --git a/src/core/welzl.cc b/src/core/welzl.cc
new file mode 100644
index 000000000..aed42afd4
--- /dev/null
+++ b/src/core/welzl.cc
@@ -0,0 +1,175 @@
+/*
+    Copyright (C) 2021 Robert Lipe, robertlipe+source@gpsbabel.org
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
+
+    This solves the smallest circle problem using Welzl's algorithm.
+
+    Welzl, Emo (1991), "Smallest enclosing disks (balls and ellipsoids)",
+    in Maurer, H. (ed.), New Results and New Trends in Computer Science,
+    Lecture Notes in Computer Science, 555, Springer-Verlag, pp. 359–370
+
+ */
+
+#include "src/core/welzl.h"
+
+#include <algorithm>            // for shuffle
+#include <cfloat>               // for DBL_EPSILON
+#include <random>               // for random_device, mt19937
+
+#include "defs.h"               // for fatal
+#include "src/core/nvector.h"   // for NVector
+#include "src/core/vector3d.h"  // for Vector3D
+
+
+namespace gpsbabel
+{
+
+Circle Welzl::b_md(const PointContainer& R)
+{
+  NVector center;
+  double radius;
+
+  switch (R.size()) {
+
+  case 0:
+    center = NVector(0.0, 0.0);
+    radius = 0.0;
+    break;
+
+  case 1:
+    center = R.at(0);
+    radius = 0.0;
+    break;
+
+  case 2:
+#if 0
+    double dp_a_b = NVector::dotProduct(R.at(0), R.at(1));
+    if (dp_a_b <= (-1.0+8.0*DBL_EPSILON)) {
+      // The points are so close to be exactly opposite each other that
+      // there isn't a unique great circle between them.
+      Vector3D result(nan(""), nan(""), nan(""));
+      return result;
+    }
+#endif
+    center = (R.at(0) + R.at(1)).normalize();
+    radius = NVector::distance(center, R.at(0));
+    //qDebug() << "center2 check" << NVector::distance(center, R.at(0)) << NVector::distance(center, R.at(1));
+    break;
+
+  case 3: {
+    auto n_EA_E = R.at(0);
+    auto n_EB_E = R.at(1);
+    auto n_EC_E = R.at(2);
+
+    // Form an orthonormal basis with two components being in the plane
+    // of n_EA_E, n_EB_E and n_EC_E, and the third being perpendicular
+    // to this plane.
+    // Call this the W frame.
+    // The columns of the rotation matrix from E to W are w1, w2 and w3.
+    Vector3D w1 = (n_EA_E - n_EB_E).normalize();
+    Vector3D w3 = NVector::crossProduct(w1, (n_EA_E - n_EC_E)).normalize();
+    Vector3D w2 = NVector::crossProduct(w1, w3);
+    // Rotate A to the W frame.
+    Vector3D n_EA_W = Vector3D(w1*n_EA_E, w2*n_EA_E, w3*n_EA_E);
+    // The z component of n_EA_W, n_EB_W, n_EC_W are all identical.
+    // The point n_EX_W that is equidistant from n_EA_W, n_EB_W, n_EC_W is
+    // either (0.0, 0.0, 1.0) or (0.0, 0.0, -1.0).
+    NVector n_EX_W;
+    if (n_EA_W.getz() >= 0) {
+      n_EX_W = Vector3D(0.0, 0.0, 1.0);
+    } else {
+      n_EX_W = Vector3D(0.0, 0.0, -1.0);
+    }
+    // Translate n_EX_W back to the E frame.
+    // We need to invert the matrix composed of the column vectors w1, w2, w3.
+    // Since this matrix is orthogonal it's inverse equals it's transpose.
+    Vector3D wt1 = Vector3D(w1.getx(), w2.getx(), w3.getx());
+    Vector3D wt2 = Vector3D(w1.gety(), w2.gety(), w3.gety());
+    Vector3D wt3 = Vector3D(w1.getz(), w2.getz(), w3.getz());
+    Vector3D n_EX_E = Vector3D(wt1*n_EX_W, wt2*n_EX_W, wt3*n_EX_W);
+
+    center = n_EX_E;
+    radius = NVector::distance(center, R.at(0));
+
+    //qDebug() << "center3 check" << NVector::distance(center, R.at(0)) << NVector::distance(center, R.at(1)) << NVector::distance(center, R.at(2));
+    break;
+  }
+
+  default:
+    fatal("welzl should not occur");
+  }
+  return Circle(center, radius);
+};
+
+int containment_count{0};
+bool Welzl::outside(const Circle& D, const NVector& n)
+{
+  ++containment_count;
+  return NVector::distance(D.center(), n) > D.radius();
+}
+
+PointContainer Welzl::unionof(const PointContainer& R, const NVector& p)
+{
+  bool contains = false;
+  for (const auto& e : R) {
+    double dp_e_p = NVector::dotProduct(e, p);
+    if (dp_e_p >= (1.0-8.0*DBL_EPSILON)) {
+      contains = true;
+      break;
+    }
+  }
+  auto Rprime(R);
+  if (!contains) {
+    Rprime.push_back(p);
+  }
+  return Rprime;
+}
+
+/* This is Welzl's algorithm */
+Circle Welzl::b_mindisk(const PointContainer& P, const PointContainer& R)
+{
+  Circle D;
+
+  if (P.empty() || (R.size() == 3)) {
+    D = b_md(R);
+  } else {
+    auto p = P.back();
+    auto Pprime(P);
+    Pprime.pop_back();
+    D = b_mindisk(Pprime, R);
+    if (outside(D, p)) {
+      PointContainer Rprime = unionof(R, p);
+      D = b_mindisk(Pprime, Rprime);
+    }
+  }
+  return D;
+}
+
+// Return the center and radius of the smallest circle containing the points.
+// This works across the antimeridian and at the poles.
+Circle Welzl::welzl(PointContainer Points)
+{
+  // randomize order of Points
+  std::random_device rd;
+  std::mt19937 generator(rd());
+  std::shuffle(Points.begin(), Points.end(), generator);
+  PointContainer BoundaryPoints;
+  Circle bound = b_mindisk(Points, BoundaryPoints);
+  //qDebug() << "containment count per point" << containment_count / Points.size();
+  return bound;
+}
+
+} // namespace gpsbabel
diff --git a/src/core/welzl.h b/src/core/welzl.h
new file mode 100644
index 000000000..6395b8219
--- /dev/null
+++ b/src/core/welzl.h
@@ -0,0 +1,66 @@
+/*
+    Copyright (C) 2021 Robert Lipe, robertlipe+source@gpsbabel.org
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with this program; if not, write to the Free Software
+    Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
+
+    This solves the smallest circle problem using Welzl's algorithm.
+
+    Welzl, Emo (1991), "Smallest enclosing disks (balls and ellipsoids)",
+    in Maurer, H. (ed.), New Results and New Trends in Computer Science,
+    Lecture Notes in Computer Science, 555, Springer-Verlag, pp. 359–370
+    
+ */
+
+#ifndef SRC_CORE_WELZL_H
+#define SRC_CORE_WELZL_H
+
+#include <vector>              // for vector
+#include "src/core/nvector.h"  // for NVector
+
+
+namespace gpsbabel
+{
+
+class Circle
+{
+public:
+  Circle() = default;
+  Circle(NVector center, double radius) : center_(center), radius_(radius) {}
+
+  NVector center() const {return center_;}
+  double radius() const {return radius_;}
+
+private:
+  NVector center_;
+  double radius_;
+};
+
+//using PointContainer = QVector<NVector>;
+using PointContainer = std::vector<NVector>;
+class Welzl
+{
+public:
+  static Circle welzl(PointContainer);
+
+private:
+  static Circle b_md(const PointContainer& R);
+  static bool outside(const Circle& D, const NVector& n);
+  static PointContainer unionof(const PointContainer& R, const NVector& p);
+  static Circle b_mindisk(const PointContainer& P, const PointContainer& R);
+};
+
+} // namespace gpsbabel
+
+#endif  // SRC_CORE_WELZL_H