Let’s have a look first!

Whenever I sit behind the computer and start tinkering with data, I want to get familiar with it first. Phrased elaborately, I perform exploratory analysis before any attempt at interpretation or explanatory break down.

The same holds for satellite data. The amount of available imagery when inspecting an area from space means that I have to sift through a lot of it. Unfortunately, doing this blindly, trusting my knowledge of latitudes and longitudes and bounding boxes, downloading the data just to realize I’ve again swapped the coordinates or that my bbox is offset tens of metres (or kilometres) always crushes the tinkerer in me.

Using Sentinel-Hub tools on a daily basis we got to thinking how we could incorporate something similar to the Sentinel-Playground into the Jupyter notebook environment. Quick search revealed there is a Jupyter/Leaflet bridge enabling interactive maps in the Jupyter notebook — ipyleaflet. Unfortunately, at that time, it did not support WMS layer, but Rok Močnik quickly fixed that, created a merge request and voilà: it is available in the package since version 0.5.4 (\o/ for open source). Although it requires some fiddling with Jupyter, I can do now what I’ve wanted to in the first place: explore the imagery from within the notebook.

Screenshot of ipyleaflet in action with Sentinel-Hub WMS layer.

Now I can make sense of the colours and textures, discern what I am looking at, frame the window so that the feature is clearly shown, pick the best data (without clouds, unless I plan on using s2-cloudless), select the best band combination etc., and continue toying with EO data.

Have a go at it yourself! (example-notebooks)