AI-Powered Soil Quality Monitoring: A Success Story from Abu Dhabi

This case study was developed in partnership with Jahiz, an upskilling initiative equipping UAE federal government employees with future-ready skills.

Healthy soil is essential for sustaining our planet. It filters and purifies water, supports biodiversity, and reduces erosion and helps mitigate climate change by sequestering CO₂ and reducing greenhouse gas emissions. As National Geographic puts it in Why Soil Matters, “Without soil, we cannot survive.” Healthy soil underpins the productivity of crops and livestock, forming the foundation of the global food supply. It influences crop yields, enhances nutrient availability and helps suppress diseases that could otherwise impact plants. In simple terms, maintaining soil health is critical for long-term food security, ecosystem preservation and sustainable land management.

Yet to keep soil healthy, we must understand the dynamic processes occurring beneath the surface. Soil health is influenced by temporal and spatial variations, even within a single field, making it a continuously changing parameter rather than a static attribute. These complexities necessitate innovative, technology-driven approaches for accurate, long-term assessment and management of soil quality.

This case study delves into a project from the Abu Dhabi Environment Agency that explored using a combination of drones, satellites, field surveys and AI to increase the efficiency and efficacy of its soil monitoring efforts.

How has soil quality been measured in the past, and why is an innovative approach needed?

Soil monitoring in Abu Dhabi has traditionally involved field teams visiting various sites to collect soil samples for lab analysis. This approach is time and resource-intensive. The costs involved mean that researchers have been unable to collect enough data points for a comprehensive assessment. To address this, field data collection needs to be supplemented with additional sources. The first step involves exploring new methods for gathering soil data, followed by finding faster, more efficient ways to analyse the increasing volume of information.

What were the first changes made to the way data is collected?

Satellites and drones now allow data collection across large areas, which can be verified through targeted soil samples processed by research teams in the laboratory. The areas that can be surveyed are far larger, and it is also now possible to focus on making changes to specific areas, making the whole process of improving soil quality far more efficient and precise.

The Environment Agency adopted a “remote sensing methodology” to enhance soil data collection, comprising four key components:

1. Collecting spectral data using a handheld spectrometer — this makes it possible to gather information without disrupting the ground.
2. Deploying custom-built drones to fly over target areas at various altitudes to capture highly detailed data.
3. Using satellites and drone sensors to collect data on soil health indicators such as hydrocarbon and heavy metal content.
4. Continuing to deploy field teams to collect soil samples in strategic areas identified during remote data collection. These results are used to validate findings from the other data sources.

Where does AI actually come into play?

AI plays a crucial role when it comes to processing the vast and complex data collected from multiple sources across large geographic areas. Analysing such a volume of data is challenging, but with AI and machine learning, we can quickly assess whether an area’s soil is contaminated. AI is integrated to predict contaminant levels—particularly heavy metals—using data from advanced sensors.

In this approach, images taken by a drone are processed with AI models, which are trained and correlated with lab-analysed soil data, providing highly accurate predictions. Specialised companies and high-spec equipment ensure the precision required for industrial applications. This approach allows for the detection of subtle changes in soil quality, and by continuously refining algorithms based on the data, AI models can become even more effective at identifying contamination.

Overall, the approach leverages AI to turn complex data into actionable insights, providing highly accurate predictions of soil health and contaminant levels.

What were the results? Was the project successful?

The integration of AI in soil quality monitoring is estimated to have a 95% increase in field coverage compared to traditional methods, as the approach can now cover 1.5 hectares in a single 15-minute flight. The system also achieved 88% accuracy in predicting contaminants like arsenic and selenium, supporting proactive environmental protection decisions.

Overall, this approach has significantly accelerated the speed of obtaining actionable results.

What’s next?

Soil quality monitoring represents just the initial application of this technology. Its principles could be adapted to monitor marine environments, water quality, and air quality, broadening its scope to comprehensive environmental monitoring. The availability of more accurate and high-resolution data can significantly enhance the development of evidence-based policies. Future advancements include the utilization of hyperspectral satellite data, which provides a much larger spatial coverage compared to drones. The recent availability of satellites capable of delivering such imagery opens new opportunities for large-scale, efficient, and precise environmental assessments.

This case study highlights how AI can enable the development of systems that gather and analyse large-scale environmental data, overcoming previous limitations in data processing and predictive capability.