Using “raster.transform” function of “rasterio” in Python

Question

When I execute the "raster.transform" function in Python it returns me a matrix.

What I can understand is that it tells me my data has a 10 m pixel resolution, It also tells me that my left bounding box is at 728749.5 and my top bounding box is at 4373200.5. What I don't understand is the rest of the elements like "-10", "1", and "0s". I also noticed that it can be a 2x3 matrix. Why is this dimension different?

I would like to know what does this matrix conveys and how can I use this information, especially when I have this info from two different rasters.

Answer 1

The .transform property returns an instance of an Affine object. This is essentially an augmented affine transformation matrix, where the 3x3 matrix always has a bottom row [0 0 1].

| x' |   | a  b  c | | x |
| y' | = | d  e  f | | y |
| 1  |   | 0  0  1 | | 1 |

The 2x3 matrix representation only shows the six coefficients (a, b, c, d, e, f) that represent this information. These coefficients are very similar to a 6-line world file, where the only difference is that a world file registers the offset to the centre of the origin cell, whereas GDAL (and rasterio) register the corner of the origin cell.

A few general remarks:

• Coefficient a is generally the cell width along the x dimension. This is only true if b and d are zero.
• Coefficient e is generally negative, since the y dimension is oriented in the opposite direction (upwards) to the raster row direction (downwards). Again, this is only true if b and d are zero.
• Coefficients b and d are usually zero, unless this is an affine transform for a rotated or sheared raster. If they are not zero, then a and e don't represent cell dimensions.
• Offsets c and f are usually to the corner of the upper-left cell.

Further reading:

• https://www.perrygeo.com/python-affine-transforms.html