Harnessing a vital tool for global food, water security

Workers inspect maize on a farm in Gauteng. Accurate estimates on crop-type distributions and irrigation practices help to guide policy decisions, allowing governments and farmers to allocate resources efficiently, say the writers.

Workers inspect maize on a farm in Gauteng. Accurate estimates on crop-type distributions and irrigation practices help to guide policy decisions, allowing governments and farmers to allocate resources efficiently, say the writers.

Published Jan 29, 2024


Jeroen Degerickx, Kristof Van Tricht and Steven Krekels

The recently held Group on Earth Observations (GEO) Week and Ministerial Summit from November 6 to 10 in Cape Town shone a spotlight on the importance of Earth observation and its benefit to humankind.

Defined by the GEO as the gathering of information about planet Earth’s physical, chemical and biological systems, Earth observation (EO) involves monitoring and assessing the status of, and changes in, the natural and man-made environment.

It can be used for exploiting new opportunities, such as the sustainable management of natural resources. GEO is a global network of over 100 national governments and institutions. Its aim is to improve the availability and use of EO for sustainable development and sound environmental management.

In the context of ensuring global food and water security, accurate estimates on cropped areas, crop type distributions and irrigation practices play an important role in guiding policy decisions, allowing governments and individual farmers to efficiently allocate resources and optimise their practices.

They may also help in identifying regions at risk of food production shortages. Real-time crop monitoring has become increasingly important, particularly for addressing climate change-induced losses and damages, as discussed during last year’s annual international climate change conference, COP27.

In June 2020, the European Space Agency-WorldCereal consortium started a scientific initiative to demonstrate the feasibility of global crop mapping at field scale. WorldCereal is a global research and development challenge aiming to engage the global agricultural community to achieve optimal results in global cropland monitoring.

The initiative was aimed at creating seasonally updated cropland, crop type and irrigation maps by using state-of-the-art classification algorithms, open EO datasets gathered through satellite imaging (mainly Copernicus S1 and S2 and Landsat 8) and reference datasets collected on the ground.

A key component of the World-Cereal solution includes a community-based, open and harmonised reference database on land cover, crop type and irrigation. Given the importance of high-quality reference data for model calibration and validation purposes, the user community is invited to contribute to this growing public database.

The WorldCereal product catalogue currently consists of six products. The first product is a base layer of temporary crops that shows all annual cropland over a one-year period.

Next to this annual product, WorldCereal provides several season-specific binary classification products including maize, winter cereals, spring cereals, active cropland and active irrigation.

The active cropland layer indicates whether in a certain growing season a temporary crop has been cultivated.

To allow for these seasonal updates, the WorldCereal team has produced global crop calendars for maize and wheat.

Based on these, the world has been split into agro-ecological zones, or zones with similar start and end of the growing seasons for maize and cereals.

In each zone, up to three growing seasons have been defined, and the following products are generated in these respective seasons:

Main cereals season: active cropland, active irrigation and winter cereals.

Main maize season: active cropland, active irrigation, maize and spring cereals (the latter typically only in the northern hemisphere).

Optional second maize season (typically only in tropical areas): active cropland, active irrigation and maize.

Although we can create a global view of these products, the map is actually a patchwork of products that change per agro-ecological zone.

For now, the WorldCereal products have only been generated for the 2021 growing seasons. Our classification algorithms have been trained on several years’ worth of data to create more robust models, enabling them to be transferred across space and time.

Details on classification model inputs, architecture and training have been recently published in a study, WorldCereal: A dynamic open-source system for global-scale, seasonal, and reproducible crop and irrigation mapping.

WorldCereal has demonstrated the power of integrating satellite remote sensing, machine learning and shared reference data, offering critical agricultural information for water and food security.

The project is now entering a new three-year phase, during which the product suite will be extended with new crops and the production system will be improved and offered as an operational and customisable crop location and crop type mapping service.

Along with a strong focus on capacity building and outreach activities, this new phase is aiming to considerably boost user uptake by the broad agricultural monitoring community.

Degerickx, Van Tricht and Krekels work for Vito Remote Sensing, the prime contractor of the European Space Agency-WorldCereal consortium.

Related Topics: