STILF - A spatiotemporal interval logic formalism for reasoning about events in remote sensing data

Favoritar este trabalho

Although several studies perform time series analysis using remote sensing data provided by Earth observation satellites, few have been explored concerning the reasoning about land use change using these data. Besides, exists the challenge of make the best use of big Earth observation data sets to represent change. In this context, this work presents a new formalism - STILF (Spatiotemporal Interval Logic Formalism), and shows how to use it for reasoning about land use change using big Earth observation data. Extending the ideas from Allen’s interval temporal logic, we introduce predicates holds(o, p, t) and occur(o, p, Te) to build a general framework to reason about events. Events can be defined as complete entities on their respective time intervals and their lifetime is limited while objects persist in time, with a defined begin and end. Since events are intrinsically related to the objects they modify, a geospatial event formalism should specify not only what happens, but also which objects are affected by such changes. The formalism proposed and predicates extended from Allen''''''''s ideas can model and capture changes using big Earth observation data, and also allows reasoning about land use trajectories in regional or global areas. Examples for tropical forest area application is presented to better understand our proposal using STILF. For the future, the proposed formalism will be include other temporal analysis tools to thinking about events related the land use and cover change.