First AgMIP Workshop on Machine Learning for Agricultural Modelling (AgML)

Organised by Laboratory of Geo-information Science and Remote Sensing

Mon 22 January 2024 09:00 until Wed 24 January 2024 17:00

Venue Impulse, building number 115
Stippeneng 2
6708 WE Wageningen
+31 (0) 317 - 482828

Why this workshop?

The Machine Learning team of the Agricultural Model Intercomparison and Improvement Project (AgMIP) meets in Wageningen on January 22-24.

AgML is the AgMIP transdisciplinary community of agricultural and machine learning modellers that aims to coordinate and advance global research activities on the interface of artificial intelligence and agricultural modelling. A key focus is how machine learning can be used to help improve agricultural modelling and contribute in addressing global challenges related to food security.

Since 2010, the Agricultural Model Intercomparison and Improvement Project (AgMIP) community of experts has been advancing methods for improving predictions on the future performance of agricultural and food systems. AgMIP has advanced widely used tools and protocols for harmonized analyses of agricultural systems using the best available models. The AgML team of AgMIP aspires to investigate not only how machine learning can help improve agricultural modelling, but also to what extend agriculture poses new challenges for the machine learning community. It builds upon the AgMIP experience, and the growing availability of agricultural data across spatial scales, alongside the increasing ease of use and popularity of machine learning methods and tools.

The workshop will bring together researchers at the intersection of agricultural modelling and machine learning, to enhance collaboration and engagement among them, and conduct protocol-based studies towards the first set of AgML intercomparison studies and benchmarks.

The specific aims are to advance the first two protocol-based studies of AgML, that aim to:
(1) evaluate the ability of machine learning models to emulate existing process-based crop models under climate change scenarios and extrapolate to future climate scenarios, and
(2) intercompare machine learning methods for in-season crop yield forecasting at sub-national scales in different environments and for different crops.

By bringing together data, models and expertise across disciplines, workshop participants will design and develop benchmarks to assess the skill of machine learning models for reproducible, intercomparable, generalizable and interpretable modelling of agricultural and food systems.

What will we do?

The workshop will focus on the two protocol-based studies, and involve four main activities, related to:

  1. Datasets: onboard new datasets or harmonize existing ones.
  2. Experimental setup: define data splits and cross-validation strategies.
  3. Evaluation: define domain-specific metrics to evaluate and intercompare models.
  4. Open science: document the data benchmarks, make them public and prepare for a research publication.

Depending on the progress made in preparation before the workshop, hands-on sessions will focus on data processing, model building or model intercomparison. In-depth discussions and lightning presentations will complement the programme.


The workshop is open, but registration is mandatory. Registration is free of charge and closes on January 8.
Workshop participants are expect to arrange their own transport and cover their accommodation costs. We will share travel instructions to registered participants.
A limited partially- and fully- funded places will also be available, which will be allocated according to need, giving preference to early career researchers, and non-Annex I parties.
Further logistics information will be announced here soon.

If you wish to participate, please get in touch with Ioannis Athanasiadis, or preferably join AgML on