Blending Earth observations and agricultural knowledge through Artificial Intelligence towards a scalable monitoring and assessment scheme for agricultural ecosystems
The project
Climaca aims to establish data-driven and domain-aware developments to assess the impact of policy-relevant interventions on the overall sustainability and performance of agricultural ecosystems. In this sense, big space-borne EO, advanced ICT technologies, and Artificial Intelligence (AI) will be the key enablers in developing actionable recommendations for policymakers to achieve resilient agriculture.
Objectives
Strategic objective (StO1)
Revealing the role of CAP specific measures in supporting the sustainable management of agricultural systems while enhancing ecosystem services, in the context of the Farm to Fork and Biodiversity Strategies.
Scientific objectives (ScO1)
Develop transfer learning and domain adaptation techniques for addressing the lack of annotated data towards the development of a generic and dynamic EO-based CAP monitoring scheme.
Scientific objectives (ScO2)
Investigate the trajectories of agroecosystem conditions, in terms of agricultural productivity, ecosystem multifunctionality and resilience to climate change, through monitoring agricultural ecosystem services using Bayesian RNN.
Scientific objectives (ScO3)
Define the complex relationships between policy mechanisms, human interventions/practices, physical processes and agroecosystem performance through integrating causal discovery methods and domain knowledge.
Scientific objectives (ScO4)
Apply causal inference methods to evaluate the impact of policy measures drivers on agroecosystems and propose spatially personalized recommendations for policymakers.
Scientific objectives (ScO5)
Employ explainable AI to enhance trustworthiness and transparency among farmers, agricultural advisors, policy implementation bodies (i.e., paying agencies) and policymakers.
Technological & Data
objectives (TD1)
Establish the link between data acquisition/collection and knowledge production to identify good practices, bottlenecks and limitations on mobilising and harmonising monitoring data to publicly accessible infrastructures.
Technological & Data
objectives (TD2)
Development of AI models and benchmark datasets (freely offered to the AI community) using EO data for monitoring agricultural systems, including CAP and ES monitoring.
Technological & Data
objectives (TD3)
Set-up a Data Cube framework for efficiently storing, accessing and processing big multi-modal datasets (raw data, indices, knowledge, meteorological data).