DENETHOR
DENETHOR is a test dataset with high geometric resolution and is used for daily crop monitoring from space. The dataset contains daily, analysis-ready planetary fusion data together with Sentinel-1 radar and Sentinel-2 optical time series for crop classification in northern Germany. In addition, information on field boundaries and crop types from two test areas in northern Germany is included. The latter are collected within the framework of the Common Agricultural Policy of the European Union. Farmers self-declare which crops they grow on their fields in order to receive subsidies. The data are not only geographically accurate, but also of high quality, as a large number of checks through in-situ measurements or EO data can potentially detect misreporting.
DENETHOR was compiled by the Technical University of Munich (TUM), DLR and Planet Labs as part of Dynamic Earth Net and is made available on EO-Lab as a copy of the original dataset without modifications.
The dataset contains two tiles. Both tiles are identical in size, 24 km × 24 km. One tile is used for training and validation, the other for testing. For the training tile, field masks and crop types from 2018 are available together with the corresponding satellite images. The test evaluations are based on 2019 data. The raw crop data contains the fields in vector format with crop ID coded between 1-999. Fields with an area of less than 1000 m² are not included, as they often have a broken shape and cannot be easily included. This affects about 1 % of all fields, with cropping forms grouped into a limited number of higher-level classes.
The nine classes with the respective number of fields in the training set in brackets are:
- wheat (305),
- rye (276),
- barley (137),
- oats (45),
- maize (251),
- oilseeds (201),
- root vegetables (23),
- meadows (954),
- forage crops (339).
The uneven frequency of the classes is a challenge for machine learning algorithms. However, it is representative of the geographical region and occurs frequently in real mapping tasks for crops.
Other crops that do not fit into one of the categories are rare in the reference data and were removed instead of grouping them into a tenth class "Other".
Usage requirements
| Required? | Remark | |
| Registration on EO-Lab | Yes | --- |
| Cloud resources on EO-Lab | Yes | --- |
| Individual request | Yes | --- |
Data access
These data are only available on individual request. Please create a support ticket to request the access information.
More information
Further information on this product can also be found on the following external pages:
References
The following list of references does not claim to be exhaustive, but it does provide an initial introduction to more advanced specialised literature:
Terms of Use
Use of the product is subject to the following terms of use:
