Mercator vs Natural Earth
Mercator and Natural Earth, measured item by item. Every figure below is labelled with the projection whose equator it is normalised to.
What each one is
| Property | Mercator | Natural Earth |
|---|---|---|
| Family | Conformal cylindrical | Compromise pseudocylindrical |
| Axis held at 1.00 | Shape | Neither |
| Leads with | Area distortion | Area distortion |
| True-shape parallel | Everywhere | 36.2° |
| Aspect ratio | 1.288 : 1 | 1.923 : 1 |
| Pole drawn as a line | None | 55% of the equator |
The same parallel, under each
Six parallels, two axes. The area columns are normalised to that projection's own equator; the shape columns are the ratio of the Tissot indicatrix's axes at the same point, where 1.00 means the shape there is unchanged. Each projection reads against its own baseline, so the two area figures on a row are not two readings on one axis.
| Latitude | Mercator area | Natural Earth area | Mercator shape | Natural Earth shape |
|---|---|---|---|---|
| 0° | 1.00 | 1.00 | 1.00 | 1.16 |
| 30° | 1.33 | 1.11 | 1.00 | 1.05 |
| 45° | 2.00 | 1.26 | 1.00 | 1.12 |
| 60° | 4.00 | 1.49 | 1.00 | 1.55 |
| 72° | 10.5 | 1.97 | 1.00 | 2.51 |
| 84° | 91.5 | 3.47 | 1.00 | 9.24 |
What they disagree about most
The objects Mercator and Natural Earth treat most differently. The first two columns stand alone: each is the area that object is painted at under that projection, against that projection's own equator. The third is the ratio between them.
Once more, because it matters: the Mercator multiples are normalised to Mercator's equator and the Natural Earth multiples to Natural Earth's. The two do not share an axis. What the third column exactly means is this - print both maps with equators the same length, and this shape comes out that many times larger on the first than on the second. It is a checkable statement about two pictures, not a physical quantity.
| Object | Mercator area | Natural Earth area | Mercator ÷ Natural Earth |
|---|---|---|---|
| Greenland | 16.5 | 2.13 | 7.74 |
| Norway | 9.38 | 1.79 | 5.23 |
| Iceland | 5.64 | 1.64 | 3.44 |
| Canada | 5.21 | 1.53 | 3.40 |
| Finland | 5.46 | 1.62 | 3.37 |
| North America | 4.64 | 1.43 | 3.25 |
| Russia | 4.90 | 1.55 | 3.16 |
| Sweden | 4.88 | 1.57 | 3.11 |
Area multiples are integrated over the whole polygon as drawn, not estimated from a centroid latitude. Shape ratios are taken at the object's anchor.
Which to use
Neither is equal-area, and they fail to be in completely different ways. The conformal one is badly wrong about area and exactly right about angle. The compromise one is a little wrong about both and badly wrong about neither.
If what you need is direction and local shape strictly right, use the conformal one. If what you need is a world map that looks most like the Earth overall, use the compromise one.
If sizes are going to be read off it, use neither. There is no equal-area projection in this pairing, so on the area question neither of them is the answer.
See it in the tool
Switching projections in the tool morphs the map between the two - it interpolates projected pixel coordinates, so which places shrink and which swell is visible at a glance. It is the fastest way on this site to understand what separates a pair.