Trustable AI Mapping

Unlocking the Potential of
Nature Tech with Trustworthy Data

AI technologies offer significant benefits to the environmental sector by automating routine tasks and eliminating the need for manual mapping.

Nature Tech: Bridging the Data Gap

Nature tech encompasses technologies that enhance and scale nature-based solutions (NBS) for environmental sustainability. These include remote sensing satellites, drones, AI, machine learning, IoT, blockchain, and bioacoustics. These tools help close the nature-data gap, providing critical information for businesses to make informed decisions.

The Challenge of Trust in AI Mapping

While AI mapping methods offer cost-effective solutions, they often lack trust due to the closed-source nature of many algorithms. Different interpretations of the same data can lead to varying visualisations of natural assets, making it difficult to rely on these maps.

Introducing the TAiM Project

The Trustable AI Mapping (TAiM) project,funded by Innovate UK,addresses this challenge by developing standards and frameworks to measure and compare the quality of AI algorithms in digital mapping. By using precise ground-based mapping of baseline sites and tester AI algorithms, nature tech organisations can now evaluate their own algorithms against verified data.

Collaborative Efforts for Reliable Solutions

The TAiM consortium, including EOLAS Insight, Omanos Analytics, Agrimetrics, Highlands Rewilding, The James Hutton Institute, AECOM, and Scottish Government departments, promotes the adoption of AI technologies in the environmental sector. By sharing open, transparent frameworks, the TAiM project ensures reliable and innovative mapping solutions that benefit the wider natural economy.

The Framework

We have developed a standardised framework for evaluating the end-to-end accuracy of AI derived maps, with a focus on nature markets. The framework allows both users and producers of AI maps to evaluate the quality of outputs against high accuracy field validation data.

To apply the framework to your mapping data, you can use the TAiM Portal to search for field sites that match your needs. The portal will walk you through the steps required to make a project and submit it for validation using the framework. You will receive back an accuracy report and some maps showing the results over one of our field sites.

Validation Sites

Field validation data was collected for eight sites across Scotland that were selected to represent a diverse range of natural assets and ecosystems:

The Centre for Sustainable Cropping

Glensaugh Farm

Tayvallich Estate

Bunloit Estate

Beldorney Estate

Annan Farm

Caerlaverock Estate

Caerlaverock Merse

Across these sites a diverse range of ecological surveys were conducted to collect data for map validation:

•Whole site habitat survey (UKHAB or NVC).
•Detailed inventory for specific ecosystems such as woodland, grassland, peatland and saltmarsh.
•Drone Lidar survey to collect surface topography, woodland canopy height and plant area index.

The Knowledge Base

The TAiM project brought together an interdisciplinary team across natural capital, AI and mapping. One of the main lessons we learned was how important communication is between these disciplines as there is so much we can learn from each other. To that end a Knowledge Base was created that contains articles written by experts across the consortium and provides insights into different aspects of the project such as field survey, AI trustability, map accuracy assessment, AI algorithms and much more.