PhD Student : DeepSoil: Integrated Soil Hydrological Assessment CPT–ML Flo

Overview

Farming in Scotland faces increasing pressure from extreme weather events, with floods and droughts threatening productivity, soil health, and water security. Yet, most monitoring remains confined to the topsoils, overlooking subsoil layers (0–2 m) that control infiltration, storage, and runoff generation. Conventional statistical approaches and pedotransfer estimates cannot capture the vertical heterogeneity of soil processes that regulate water movement. Deep Soil addresses this evidence gap by integrating Cone Penetration Testing (CPT) with machine learning (ML) and geostatistics to map soil infiltration and water storage functions.

We ope rationalise two project‑defined, CPT‑derived indices, Infiltration capacity (I*) and Storage potential (S*) will quantify how water moves and is retained in the soil profile. Using CPT profiles calibrated with intact soil cores, the project will create 10 m (field) and 25 m (catchment) resolution maps of soil hydraulic functioning across the soil profile, enabling early identification of flood- and drought-prone zones.

Soil degradation already costs Scottish agriculture an estimated £25–75 million annually through compaction alone, and each 1 % increase in runoff can raise flood losses by £57–76 k per affected property, underscoring the urgency of improved hydrological risk screening.

The project aims to develop an integrated CPT–ML–geostatistical framework for deriving and mapping the I
* and S
* indices to assess soil resilience under climatic and land‑use pressures. Its objectives are to: (i) calibrate CPT data against soil cores and hydraulic tests; (ii) upscale point measurements using ML and kriging to produce uncertainty‑aware maps; and (iii) combine static capacity with dynamic environmental data (rainfall, soil moisture, PET) to identify flood and drought hotspots. Outputs will include validated maps, uncertainty layers, and dashboards to inform sustainable land and water management.

Representative sample locations will be statistically determined to obtain with CPT soundings and co‑located soil cores across two long‑term experimental platforms: the Centre for Sustainable Cropping (CSC), in Balruddery Farm, offering over a decade of soil health data under regenerative and conventional management, and the Glensaugh Climate‑Positive Farming Initiative (CPFI), representing hill farming and upland soil contexts. CPT variables (qₚ, fₛ, u₂) will be calibrated to measured hydraulic properties, producing local I
* and S
* values. ML models (e.g. Random Forest) will predict I
* from covariates such as terrain indices, geology, land cover, and Sentinel indices, while S
* will be interpolated using kriging with uncertainty propagation. These static maps will be fused with CHESS‑SCAPE rainfall, COSMOS‑UK and river flow from SEPA gauges to generate dynamic flood/drought indicators validated against observed events. Technical considerations include corrections for peat and stony tills, CPT normalisation, and explicit treatment of measurement and model uncertainty.

This 4yr PhD project is a competition jointly funded by The James Hutton Institute and Abertay University. This opportunity is open to UK students and will provide funding to cover a stipend and UK level tuition. International students may apply, but must fund the difference in fee levels between UK level tuition and international tuition fees. Students must meet the eligibility criteria as outlined in the UKRI guidance on UK and international candidates.

Applicants will have a first‑class honours degree in a relevant subject or a 2.1 honours degree plus Masters (or equivalent).