Hybrid GNSS-Rainfall Decision Support for Staged Early Warning of Rainfall-Induced Slope Instability in Agricultural Landscapes: A Yarra Valley Prototype Supporting SDG 2 and SDG 11

Madlazim Madlazim (Universitas Negeri Surabaya)
Muhammad Nurul Fahmi (Universitas Negeri Surabaya)
Arie Realita (Universitas Negeri Surabaya)
Baba Musta (University of Malaysia Sabah)
Dyah Permata Sari (University of Surabaya)
Khaista Rehman (University of Peshawar)

AMPLITUDO: Journal of Science & Technology Innovation
Vol. 5 No. 2 (2026): August

Early detection of rainfall-induced slope instability in agricultural terrain remains difficult because hydrological loading commonly precedes measurable ground movement. This study develops an explainable decision-support framework that integrates GNSS-derived detrended displacement, antecedent rainfall indicators, and a machine-learning anomaly score for staged warning in the Yarra Valley, Australia. A regional target-control GNSS architecture was implemented with four slope stations and two control stations, and a retrospective rainfall-driven event window in April 2025 with complete processing products was analysed. The workflow combines 24 h and 72 h rainfall accumulation, target-control displacement metrics, deformation gradients, and a weighted fusion index to distinguish background variability from physically plausible slope response. Cumulative rainfall increased before localized deformation emerged, and the lower slope sector showed stronger response than the upper sector. The fusion layer therefore supports escalation from normal to watch and, when rainfall and deformation thresholds are jointly exceeded, to warning. This prototype strengthens operational decision-making for agricultural slope management and provides a basis for multi-event validation and wider deployment