Sentinel Mosaic QGIS Plugin

This repository contains a QGIS python plugin for ordering low-resolution cloud-free mosaics from Sentinel Hub and loading them into QGIS. It is useful for determining which combinations of Sentinel orbits can be used for seamless mosaics and which years/months are necessary to include to get a high quality mosaic image.

Installation

To use the plugin you must first install the sentinelhub python package. This can be done in a virtual environment but there are additional steps required in that case.

No virtual environment

To install sentinelhub without a virtual environment you can simply run this:

pip install sentinelhub

Yes virtual environment

To install sentinelhub package in a virtual environment you need to run this:

workon {env}
pip install sentinelhub

Then in QGIS you need to add your virtual environment to the python path. You can add a few lines to a script called startup.py in your QGIS home directory:

1. Locate the QGIS home directory by navigating the menu bar to 'Settings' > 'User Profiles' > 'Open Active Profile Folder'. This directory is located within the QGIS home directory so now you need to navigate up to a folder called 'QGIS3'. Once you have found this directory make a file called startup.py and add these lines:

   import sys
   import os
   sys.path = [os.environ['WORKON_HOME'] + '{env}/lib/python3.8/site-packages'] + sys.path

   This will add the virtualenv to your QGIS python path and you should be able to use the sentinelhub package from within QGIS.

Once you have done that, you must authenticate your machine to access our Sentinel Hub account. You can authenticate your machine from the bash terminal:

sentinelhub.config \
  --instance_id '<instance_id>' \
  --sh_client_id '<sh_client_id>' \
  --sh_client_secret '<sh_client_secret>'

(Credentials here are in LastPass.)

If sentinelhub.config crashes with OSError: Could not find lib c or load any of its variants [], you may need to upgrade the shapely library to 1.7.1 or above. Run pip install -U shapely and then try again.

The plugin can be added to QGIS by downloading the zip file at the v0.1 release page and installing via the QGIS Plugin Manager (install from zip).

Gotcha: if you're getting a pyproj._network error, try upgrading your QGIS to 3.16 or above.

Usage

1. Open plugin dock widget
   • Click the plugin icon on your toolbar (it is the ST globe logo)
   • This will open a sidebar with the plugin toggles
2. Create a rectangular polygon feature that covers the extent of the area for which you want a mosaic preview.
   • From the menu at the top click Layer > Create Layer > Temporary Scratch Layer
     • Set it to polygon geometry with WGS84 projection
   • Digitize the feature using the digitization tools
     • the 'Shape Digitization Toolbar' is very helpful for drawing rectangles
3. Specify the years/months to include in the stack of images to be mosaiced
   • The mosaic process will take the median value of all valid pixels in the stack (i.e. no clouds/shadows/snow) so ideally we only include the most recent possible images.
   • There are places where it is necessary to take the stack all the way back to 2018 in order to get enough valid pixels in the stack (e.g. Maine).
   • The more dates that are
4. Set the maximum proportion of cloud cover for images included in the stack
   • default is 0.2 (20%)
   • we pay for each scene that is included in the processing stack so we can reduce processing cost by excluding scenes that are very cloudy
5. Specify the list of Sentinel 2 relative orbits that you want to include in this test
   • Sometimes running mosaics across multiple orbits leaves noticeable seams along the boundaries between orbits. We can check for this issue using this plugin so we don't spend a bunch of money making stripy mosaics!
   • enter a comma-separated list of relative orbits e.g. 112,69
     • see this GeoJSON
6. Select your temporary scratch layer from the dropdown menu for 'Layer for bounding box'
   • The plugin will order an image that covers the bounding box of this layer
7. Double check your settings then click the 'Order Mosaic Preview' button to order the image
   • This will order a low resolution (fixed 512 pixel width) false color composite image and load it into your QGIS session when it is done.
   • The process should take less than one minute
   • Review the image for stripes between orbits, or for hazy/cloud spots that were not adequately filled in by the cloud filling process.
   • If there are still clouds left behind you may need to include more years in the stack.

Sentinel Orbits

The sentinel orbit geospatial data is helpful to have in QGIS when you are experimenting with different mosaic settings. A GeoJSON of the relative orbits for Sentinel 2 is can be downloaded here.

Here is the R code used to scrape the data and create the GeoJSON:

library(magrittr)
url <- "https://sentinels.copernicus.eu/documents/247904/4584266/Sentinel-2A_MP_ACQ_KML_20210225T120000_20210315T150000.kml"

orbits_raw <- sf::st_read(url, layer = "NOMINAL")

orbits_dissolved <- orbits_raw %>%
  dplyr::group_by(OrbitRelative) %>%
  dplyr::filter(OrbitAbsolute == max(OrbitAbsolute)) %>%
  dplyr::summarize(
    .groups = "drop"
  )

sf::st_write(orbits_dissolved, "sentinel_orbits.geojson")

Development

If you clone the repo and work on it locally, you can deploy changes to QGIS using the system package plugin_build_tool (pbt).

http://g-sherman.github.io/plugin_build_tool/

You will need to modify the file SentinelMosaicTester/pb_tool.cfg to point to the correct QGIS plugin directory on your machine.

cd sentinel_mosaic_qgis_plugin/SentinelMosaicTester

pbt deploy

This will copy the source code into your QGIS plugin folder. In QGIS you can use the 'Plugin Reloader' plugin to refresh the plugin in your QGIS session.

Uploading an installable release

The easiest way to distribute a new release is to zip the contents of the SentinelMosaicTester directory and upload to GitHub.

1. Zip the package contents

cd sentinel_mosaic_qgis_plugin
zip -r SentinelMosaicTester.zip SentinelMosaicTester

2. Create a tag

git tag v{version} -m 'SentinelMosaicTester v{version}'

3. Publish a release via gh (GitHub command line interface). This will create a release on the specified tag and attach the .zip created in step 1.

gh release create v{version} SentinelMosaicTester.zip

You can also create the release from the GitHub web interface.