Placing moisture sensors in the correct location: a statistical method for saving water

Excellent articulation between quantitative falsifiable predictions and a sequential phased protocol (in silico → minimal → complete), with clear GO/NO-GO/PIVOT criteria that permit an objective decision at each stage, reducing the risk of resource wastage on a non-viable hypothesis.

The hypothesis proposes a rigorous methodological transfer from OED (D-optimality) to an applied problem of soil moisture sensor placement, which is conceptually sound and builds upon existing literature in geostatistics and Bayesian experimental design. The causal chain is logical and well-articulated.

The use of D-optimality to minimise the variance of spatial prediction is theoretically grounded within the framework of stationary Gaussian models, and the causal chain is clearly articulated.

Clear and quantifiable addressable market: precision agriculture (global market ~$12bn in 2025, CAGR 12%), with a specific segment for 'smart irrigation' (~$2bn) and 'soil sensing' (~$1.5bn). Early adopters are large agro-industrial groups (John Deere, Trimble, Netafim, Lindsay Corp) and high-tech agricultural cooperatives (e.g., CHS, Inc. in the US, or InVivo in France). These actors already pay for moisture sensors (TEROS, Decagon) and precision mapping services. A 15–25% reduction in prediction variance translates directly into water savings (10–20% according to the panel’s benchmarks) and input optimisation, which justifies a service premium.

A clear and testable hypothesis is presented, accompanied by a well-defined phased protocol (Go/No-Go), a structure that is highly valued by funding reviewers for risk management.