Creating a Mosaic in Python

Build a single-image composite of multiple rasters: a hands-on Jupyter Notebook tutorial using Planet data

Creating a composite image from multiple PlanetScope scenes

In this guide, you'll learn how to create a composite image (or mosaic) from multiple PlanetScope scenes that cover an area of interest (AOI). You'll need GDAL (Geospatial Data Abstraction Library) and its python bindings installed to run the commands below.

First, let's use Planet Explorer to travel to stunning Yosemite National Park. You can see below that I've drawn an area of interest around Mount Dana on the eastern border of Yosemite. I want an image that depicts the mountain on a clear summer day, so I've narrowed my data search in Planet Explorer to scenes with less than 5% cloud cover, captured in July and August 2016.

Mount Dana in Planet Explorer

As you can see in the animated gif above, my search yielded a set of three PlanetScope scenes, all taken on August 20, 2016. Together these scenes cover 100% of my area of interest. As I roll over each item in Planet Explorer, I can see that the scenes' rectangular footprints extend far beyond Mount Dana. All three scenes overlap slightly, and one scene touches only a small section at the bottom of my AOI. Still, they look good to me, so I'm going to submit an order for the visual assets.

After downloading, moving, and wrangling the data, I'm ready to create a composite image from the three scenes. First, though, I'll use gdalinfo to inspect the spatial metadata of the scenes.

In [ ]:
!gdalinfo data/175322.tif 
!gdalinfo data/175323.tif
!gdalinfo data/175325.tif

The three scenes have the same coordinate systems and the same number of bands, so we can go ahead and use the gdal_merge.py utility to stitch them together. In areas of overlap, the utility will copy over parts of the previous image in the list of input files with data from the next image. The -v flag in the command below allows us to see the output of the mosaicing operations as they are done.

In [ ]:
!gdal_merge.py -v data/175322.tif data/175323.tif data/175325.tif -o output/mtdana-merged.tif

We can see in the verbose output above that the mosaicing operation is fairly simple: the utility script is basically copying a range of pixels for each band in each scene over to the designated output file. We can use gdalinfo to inspect the metadata of the merged raster file we created.

In [ ]:
!gdalinfo output/mtdana-merged.tif

The merged raster is a large GeoTiff file, so we're going use gdal_translate, another GDAL utility, to convert it to a PNG and set the output image to a percentage of the original. That will make it easier for us to view in this notebook.

In [ ]:
!gdal_translate -of "PNG" -outsize 10% 0% output/mtdana-merged.tif output/mtdana-merged.png

Now let's view the merged image.

In [1]:
from IPython.display import Image
Image(filename="output/mtdana-merged.png")
Out[1]: