How it works
UTM is one transverse Mercator cylinder: lay it on its side so it wraps round the globe along a
meridian, and use only the strip 3 degrees either side of that line, where the map is nearly true.
Turn the cylinder 6 degrees and do it again: 60 times round the globe gives the 60 zones.
"Unroll" moves the zone's strip onto its cylinder and opens it flat. The cylinder is a little
smaller than the globe (scale 0.9996), so it cuts the surface along two lines about 180 km either
side of the central meridian, where the map is exactly true.
MGRS cuts each zone into 8-degree latitude bands, C to X without I and O, to name a grid zone
such as 32V, then into 100 km squares named by two letters. A reference such as
32V KN 12345 67890 adds easting and northing within the square, to the metre.
Two places break the pattern: 32V is widened over south-west Norway, and over Svalbard zones
31X, 33X, 35X and 37X are widened to cover 32X, 34X and 36X, which do not exist.
A Sentinel-2 tile is one of those 100 km squares, 109.8 km on a side, so tiles overlap by
9.8 km on the east and south: drawn translucent, overlaps glow brighter. Each zone's tiles keep its
own UTM projection, and the squares at a zone's edge run on past it, so near the edge the same
ground is in tiles from two zones, in two different coordinate systems.
Beyond 84° N and 80° S MGRS switches to UPS, the polar stereographic projection: a
light at one pole casts the other polar cap onto a plane across it. The plane is lowered to scale
0.994 so it cuts the globe near 81.1°, like the UTM cylinder. Each cap is split at Greenwich into
two grid zones, A and B in the south and Y and Z in the north, with 100 km squares of their own.
Sentinel-2 does not use UPS: its UTM tiles in band C run on south of 80° S.
Click the globe or a zone for the point's UTM coordinates, its MGRS reference and the tiles that
contain it, or paste a longitude and latitude, an MGRS reference or a tile name.