Near-real time forecasting and change detection for an open ecosystem
Domaine:
environment and energygeospatialclimateType de record:
software
Créateur:
Éditeur:
fig
Hôte:
Presentation given for the GEO BON Open Science Meeting, July 2020 conf2020.geobon.org - complete with narration!
We present a hierarchical Bayesian modelling framework that allows us to forecast remotely sensed vegetation indices in a fire-dependent and seasonally fluctuating ecosystem, the Fynbos of South Africa. This framework allows several applications including: 1) detecting near real-time changes in the state of the ecosystem by comparing observed vegetation signal with the model forecasts; 2) determining the influence of plant traits on vegetation productivity and seasonality; 3) forecasting changes in vegetation productivity and seasonality under altered climate or community composition; and 4) estimating ecosystem properties like leaf area index (LAI) or above ground biomass. As such, it provides the means to draw linkages across and/or monitor several EBV classes.
We present a hierarchical Bayesian modelling framework that allows us to forecast remotely sensed vegetation indices in a fire-dependent and seasonally fluctuating ecosystem, the Fynbos of South Africa. This framework allows several applications including: 1) detecting near real-time changes in the state of the ecosystem by comparing observed vegetation signal with the model forecasts; 2) determining the influence of plant traits on vegetation productivity and seasonality; 3) forecasting changes in vegetation productivity and seasonality under altered climate or community composition; and 4) estimating ecosystem properties like leaf area index (LAI) or above ground biomass. As such, it provides the means to draw linkages across and/or monitor several EBV classes.