Building added-value services on top of Sentinel Hub

NDVI maps

There were several science and industry events related to Copernicus programme in last few weeks — from Living Planet Symposium, European Space Solutions and ESA’s industry workshop on EOP-5. The message is clear — everybody is working on integrating Sentinel-2 imagery in their applications.

However, there are still not many application out there. Speaking with various people, it seems that many are struggling to implement an operational S-2 data feed. Amazon hosted public data set that we have established, helps quite a bit as it eliminates SciHub-related connection problems and limitations. But the vast volume of data and processing seems to be too complex for many. It does not have to be this way.

By using Sentinel Hub’s services, it is more easy than ever to integrate Sentinel-2 data. There are several options:

1. Use of OGC standard WMS and WMTS services

This is the easiest way as almost all WebGIS solutions support addition of WMS and WMTS data sources (e.g. CartoDB, OpenLayers, MapBox, etc.). If one favours Tile Map Service (TMS), this is possible as well by using WMTS with service’s URL instead of a WMS one.

Developers can choose amongst many Sentinel-2 renderings and control these by setting parameters — time range, cloud cover filtering/replacement, colour stretch correction and others. If one wants to explore a temporal aspect of the data — e.g. implementing a time-lapse feature, there is a GetIndex request, which lists all available scene dates at the chosen location.

One such example is our Farm management tools software for farmers, where we added a new group of Sentinel layers. For each one, user can choose the time range for data preview. Date selector has been implemented in a way to show to the user, for which dates scenes are available — these are highlighted.

Time selector with greyed-out available scenes.

NDVI map helps differentiate between crops and crop stages.

Similarly it is possible to integrate data source in smartphone application. An example is from our Sentinel-2 NDVI Maps Android application, which shows last available scene at any location on the world.

2. Data processing based on Sentinel Hub outputs

For advanced developers, who would like to add image processing, there is a Web Coverage Service (WCS) available. Powered with same parametrisation as above described WMS, developers can retrieve data values in 16-bit GeoTiff format, without loss of quality. It is therefore possible to implement a process, which downloads a stack of data values from Sentinel Hub (e.g. weekly values for a chosen area) and then implement a change detection mechanism on top of it. Or crop classification. Or anything else. Possibilities are limitless.

3. Additional processing options

We have added most standard EO products, which can be selected within the WMS Configurator (and we are adding more). But some of our users would like to add their own data processing products to the service. This is possible as well by programming the process within the “Custom script” — this process will be then automatically applied, when WMS/WMTS/WCS service will be used. It is possible to combine all available bands in many different options. In the near future, we will add also temporal aspect to the scripting, as well as object based processing.

4. Raw band data

For hard core remote sensing specialists we have added a possibility to simply download raw band data — either one by one or several bands together in one GeoTiff. These are disseminated in 16-bits reflectance values.

Benefits of building applications using Sentinel Hub services

There are several:

• No need to worry about storage, processing power — all of this is handled by Sentinel Hub.
• No need to handle data downloads, JPEG 2000 decompression, mosaicking, compositing, cloud replacement, atmospheric correction, reprojection and other processes, which take a lot of time to be done right.
• No need to worry about updating the data — we take care for this; data are usually available in less than half an hour after being disseminated on SciHub.
• Last, but far from least, applications can be developed much faster. End-users are not really interested in observing Sentinel-2 data, even though these are beautiful. They would like to get some relevant information — this is where the highest added value is. This is why developers should focus to final applications, not to raw data processing.

We will be happy to work with developers and help them integrating Sentinel data in their applications. If there is some important feature missing in the services, let us know.

Similarly, inform us if you would like to get a test account so that you can try the service out.

Originally published at sentinel-hub.com.