|
Abstract |
Operational predictive and risk modelling of landscape-scale hazards such as floods and fires requires rapid processing of geospatial data, fast model execution and efficient data delivery. However, geospatial data sets required for hazard prediction are usually large, in a variety of different formats and usually require a complex pre-processing toolchain. In this paper we present an overview of the Geostack framework, which has been specifically designed for this task using a newly developed software library. The platform aims to provide a unified interface for spatial and temporal data sets, deliver rapid processing through OpenCL and integrate with web APIs or external graphical user interface systems to display and deliver results. We provide examples of hazard and risk use cases, particularly Spark, a Geostack based system for predicting the spread of wildfires. The framework is open-source and freely available to end users and practitioners in the hazard and geospatial space. |
|