HMAS ingests high-dimensional hyperspectral data from airborne, satellite, or ground-based sensors. Built-in preprocessing handles calibration, noise reduction, band selection, and normalization — so the platform works with data from different sensors without custom setup.

The core of HMAS is an ML model that estimates heavy metal concentrations directly from spectral signatures. It's designed for high-dimensional input and handles spectrally complex or heterogeneous soils where simple index-based methods break down.

Spectral predictions are converted into continuous contamination maps. You can view distribution patterns at multiple scales, identify hotspots, and track changes over time — all without running a new field campaign.

Physical sampling is slow, expensive, and spatially limited. HMAS replaces or complements it with remote spectral analysis — covering the same area in far less time. Particularly valuable for agricultural land, post-industrial sites, and mining-affected regions.

Cross-validation workflows, uncertainty estimates, and feature importance tools are built in. You can see which spectral bands drive a prediction and why — which matters both for reproducibility and for building trust in the outputs.

Outputs are compatible with standard GIS and statistical analysis tools. Export formats support downstream modeling, publication workflows, and data sharing — so HMAS fits into existing research pipelines rather than replacing them.