This use case applies AI-based downscaling techniques to convert coarse-resolution climate projections into high-resolution datasets useful for agriculture and insurance. It aims to enable more localised climate risk assessments and unlock new exploitation pathways through Data-as-a-Service models.
High-resolution downscaling of climate scenarios and risk trend analysis in agriculture
RI: ENES
Challenge
Existing climate datasets from CMIP6, EURO-CORDEX, and DestinE are too coarse for sector-specific applications in agriculture and insurance. This use case aims to downscale these to a few hundred metres resolution using AI (Convolutional Neural Networks) and statistical techniques, enabling more precise environmental risk analysis and supporting new business models.
Target
Create and validate high-resolution climate datasets (O(100)m) using AI downscaling, and assess their usability and value in real-world agriculture and insurance scenarios.
Development Steps
-
Data preparation for training (merging, formatting, cleaning datasets)
-
Initial model setup to accept pre-processed data
-
Model tuning to optimise performance and accuracy
-
Run inference to generate high-resolution climate datasets
-
Stakeholder workshops to validate usability and explore business models
Target Stakeholders
-
Research Infrastructure (RI) users
-
Stakeholders in agriculture, insurance, and climate impact modelling
-
Potential downstream users of high-resolution climate data
Impact
The project enhances the usability of large-scale climate projections by creating application-ready high-resolution maps. Benefits include:
-
Improved workflows for environmental users
-
Enhanced agricultural and insurance planning
-
Enabling Data-as-a-Service business models
-
Promoting downstream exploitation of RI climate data
