Enhancing sustainability in agricultural practices through cover crop type mapping
11 Mar 2024
Cover crops ensuring optimal soil health and better yields
Cultivating crops is a complex process, as it involves the understanding of intricate interactions between plants, organisms, and the environment. Cover crops also known as ‘catch crops’ or ‘intercrops’ are cultivated during the off-season, after the harvest of ‘cash crops’, aimed at being sold on the market. The role of cover crops is primarily to protect and enhance soil health, and to increase yields of cash crops.
European Union’s Common Agricultural Policy (CAP) recognises it to be an important practice to efficiently prevent soil erosion, increase soil organic matter, suppress weeds, fix nitrogen, improve soil structure, enhance biodiversity, and attract pollinators. Different cover crop types exist, but the most common planted belong to the family of ryegrasses, mustards (see figure on the left), and clovers.
A new prototype candidate to help distinguish between the crops
In this context, the Copernicus Land Monitoring Service (CLMS) has invested significant resources to enhance understanding of the crops that shape our agricultural infrastructure.
EvoLand is currently working on bringing this understanding to the next level. The new CLMS prototype candidate, named “Cover Crop Type Mapping”, is an innovative Earth Observation product designed to map types of crops at a 10-metre resolution. The advanced spatial resolution of the CLMS prototype implies that the product will be able to identify and differentiate between various cover crop types with improved detail over relatively small areas.
EvoLand prototype candidate could become a nice accompaniment to the CLMS High Resolution Layer-Vegetated Land Cover Characteristics (HRL-VLCC) product still under development, which will map the seasonality of the cover crops at European scale.
Novel in-situ data, artificial intelligence and other methods deployed by the prototype
The prototype integrates novel methods developed within EvoLand such as i) novel in situ data, ii) weakly supervised learning, and iii) novel infrastructure and data access.
Field observations or reference data on cover crops is scarce due to the lack of mandatory reporting requirements for farmers on cover crops. To mitigate this issue, EvoLand compiles and harmonises existing databases of in situ data and develops a tool for automated training data generation through the novel in situ data task.
Additionally, EvoLand explores techniques such as weakly supervised learning to enhance the prediction of cover crops. Weakly supervised learning is a subset of machine learning which constructs predictive models by learning with weak supervision, i.e. without relying on a large number of labelled training examples. This technique might be particularly useful to enhance the prediction of cover crop types.
Creating knowledge to bring us one step closer to sustainability
The cover crop prototype represents a great set of fundamental data at high spatial resolution. By leveraging advanced technologies and methodologies, such as innovative in-situ data collection, weakly supervised learning, and sophisticated computational infrastructure, this tool can be a valuable input to several agricultural monitoring applications including soil conservation and protection, soil fertility, biodiversity (pollinators) and bare soil monitoring. Through this enhanced product, scientists will gain a deeper understanding of soil health, which will, in turn, lead to improved strategies for ensuring food security via such crops.
As healthy crops are more likely to meet food demands, this initiative moves society closer to achieving sustainable agriculture in the future.
Would you be interested to follow the results of EvoLand and the development of this candidate prototype? Sign up for our newsletter (here below or via this form), follow us on LinkedIn, Mastodon or X (Twitter).
Questions about the article or the project? E-mail us at contact@evo-land.eu.
==
This article is part of a series providing more details on all 11 EvoLand candidate prototypes.
Previous article in the series: Improving water body mapping by means of novel EO-data and super resolution
Inline image: An example of a Cover Crop (yellow mustard) in Belgium. Credit: VITO
Newsletter
Stay current! Subscribe to our EvoLand newsletter!