Fire Forecasting

A forecast window (start date plus a 24, 48 or 72-hour horizon) drives four KPI cards, a map of sites coloured by risk level, a per-site risk chart with threshold and peak markers, and a site table, all recomputed from the bundled sample.

A Next.js dashboard prototype showing wildfire risk for Santa Barbara, Ventura and Los Angeles counties: KPI cards, a Leaflet forecast map, an ApexCharts risk chart and a site table, all driven by a small sample dataset built from a committed trihourly weather CSV. The monitoring sites and risk probabilities are fictional sample data, and the UI labels them that way. An earlier version included a Python ML pipeline (FIRMS, RAWS and FRAP data with ANN/LSTM models) and a FastAPI backend; that code now lives only in the git history while the project restarts from the front end.

A Next.js 14 App Router front end with Material UI renders four views: the dashboard, a full-width map, ML History and Settings. The dashboard reads a small JSON dataset that a Python script builds from the committed trihourly weather CSV. Leaflet draws the Tri-County box and risk-coloured sites, and ApexCharts plots the selected site's risk with threshold and peak annotations. The original ML pipeline and FastAPI service live in the git history.