HYDRA-EO: GRS leads ESA project on hybrid modelling and AI for multi-stressor crop monitoring

HYDRA-EO addresses one of the key scientific and operational challenges in agricultural Earth Observation (EO): how to robustly detect, disentangle, and scale multiple biotic and abiotic stressors in crops using heterogeneous EO data streams. The project is firmly embedded in the GRS research line on hyperspectral and thermal remote sensing, radiative transfer modelling (RTM), and hybrid machine-learning approaches for plant trait retrieval.

The project initially focuses on two high-priority European crop–disease systems: potato fields in the Netherlands, affected by Potato Virus Y (PVY) and associated bacterial infections, and grapevine systems in Italy, threatened by Flavescence Dorée, a regulated disease of major concern for European viticulture. These systems provide contrasting canopy structures and stress dynamics, enabling controlled evaluation of early disease signals. Beyond these priority cases, HYDRA-EO explicitly addresses water stress (drought) and fungal and insect pressures in additional cropping systems, including olive, pistachio, and alfalfa in Mediterranean environments. A central research objective is to understand how combined stressors modify plant physiological responses and how these interactions propagate into spectral, thermal, and fluorescence signals observed from UAVs, airborne sensors, and satellites.

At the core of HYDRA-EO is a hybrid modelling framework that couples physically based plant and radiative transfer models with machine-learning algorithms to robustly detect and interpret multiple crop stressors. Radiative transfer models establish mechanistic links between plant traits, such as chlorophyll content, leaf and canopy water status, LAI, canopy temperature, and photosynthetic activity, and EO signals, while machine learning exploits these model-constrained features to enhance robustness, interpretability, and generalisation across sensors, crops, and environments.

The project integrates high-resolution hyperspectral and thermal observations acquired from UAV and airborne platforms, complemented by leaf-level measurements, biochemical analyses, and genetic assessments using genome-wide association studies (GWAS), together with satellite data from operational missions including Sentinel-2, PRISMA, and EnMAP. This integrated dataset is explicitly designed to detect, monitor, and disentangle plant infections and disease progression, as well as their interaction with abiotic stressors such as water stress.

A central scientific objective is to quantify the effects of spectral and spatial degradation and to develop sensor-independent plant trait retrieval strategies, enabling reliable upscaling from field and airborne observations to spaceborne EO data for operational disease monitoring. This scaling framework directly supports upcoming next-generation ESA missions, specifically FLEX and CHIME, and is developed in synergy with ongoing European projects, including STELLA and CERBERUS, to ensure methodological consistency and operational relevance across initiatives.

HYDRA-EO is coordinated by Wageningen University & Research, with GRS leading the scientific integration of EO data, modelling, and scaling strategies. The project is carried out in collaboration with Wageningen Environmental Research, the Consiglio Nazionale delle Ricerche – Institute of BioEconomy, and the Chaparrillo Agro-Environmental Research Center.

Scientific and technical coordination is led by Dr. Carlos Luis Camino González together with Professor Dr. Lammert Kooistra, reinforcing GRS’s role in advancing trait-based EO methodologies, hybrid RTM–ML frameworks, and preparatory science for future ESA missions.