Conference Paper
Abstract:
Time-series analyses are an important task in earth observation for studying the change of the environment, not only for research but also for local governments and stakeholders. Existing earth observation time-series data (e.g., Landsat, MODIS) can be used in conjunction with existing time-series analysis software to extract information based on the temporal change of the vegetation.
Therefore, standardized and automated processing can foster the usage for time-series analyses, as the manual data processing and analysis can be very time-consuming and needs technical expert knowledge in specific data formats and analysis software. To overcome these barriers, individual steps for data processing and analysis were automated and published as OGC Web Processing Services (WPS) within the Earth Observation Monitor project [1]. Local stakeholders are now be able to access and analyze time-series data with just a few steps and without data processing.
PyWPS was used to provide geo processing services; geospatial results from analysis processes are published for visualization as OGC Web Map Service with MapServer. Several OGC WPS processes were developed for 1) data extraction, 2) breakpoint detection in seasonality and trend using BFAST software, 3) spatial distributed trend calculations using greenbrown software, and 4) the derivation of phenological parameters for vegetation data using TIMESAT software. The analyses processes are based on external software that is executed from Python code within the individual WPS process: The R-packages (BFAST, greenbrown) are executed within Python with the rpy2 module. TIMESAT is started as command line execution. The results are then further processed in Python.
Two clients were developed using these OGC WPS processes and were presented: A web portal and a mobile application. From the web portal the user can execute the individual processes one after another; from the mobile application an OGC WPS-based processing chain is being executed to provide data extraction and time-series analyses in one step. Furthermore our made experiences, drawbacks and achievements in the field of raster based time-series processing and analysis with OGC services were presented.
Presenter Biography:
Jonas Eberle is a Ph.D. candidate at the Department for Earth Observation, University of Jena. His research is focused on operational and automated time-series access and analysis to foster land monitoring based on earth observation. Another objective is to simplify the usage of earth observation data with easy-to-use web portals and mobile applications. Jonas Eberle did a Bachelor degree in Applied Informatics and a Master degree in Geoinformatics / Earth Observation.