ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Publications Copernicus
Download
Citation
Articles | Volume IV-4/W2
ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., IV-4/W2, 87–92, 2017
https://doi.org/10.5194/isprs-annals-IV-4-W2-87-2017
ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., IV-4/W2, 87–92, 2017
https://doi.org/10.5194/isprs-annals-IV-4-W2-87-2017

  19 Oct 2017

19 Oct 2017

EXTRACTING SPATIOTEMPORAL OBJECTS FROM RASTER DATA TO REPRESENT PHYSICAL FEATURES AND ANALYZE RELATED PROCESSES

J. A. Zollweg J. A. Zollweg
  • Dept. of the Earth Sciences, The College at Brockport, 350 New Campus Drive, Brockport, NY 14420, USA

Keywords: feature extraction, remote sensing, object-oriented modelling, storm elements

Abstract. Numerous ground-based, airborne, and orbiting platforms provide remotely-sensed data of remarkable spatial resolution at short time intervals. However, this spatiotemporal data is most valuable if it can be processed into information, thereby creating meaning. We live in a world of objects: cars, buildings, farms, etc. On a stormy day, we don’t see millions of cubes of atmosphere; we see a thunderstorm ‘object’. Temporally, we don’t see the properties of those individual cubes changing, we see the thunderstorm as a whole evolving and moving. There is a need to represent the bulky, raw spatiotemporal data from remote sensors as a small number of relevant spatiotemporal objects, thereby matching the human brain’s perception of the world. This presentation reveals an efficient algorithm and system to extract the objects/features from raster-formatted remotely-sensed data. The system makes use of the Python object-oriented programming language, SciPy/NumPy for matrix manipulation and scientific computation, and export/import to the GeoJSON standard geographic object data format. The example presented will show how thunderstorms can be identified and characterized in a spatiotemporal continuum using a Python program to process raster data from NOAA’s High-Resolution Rapid Refresh v2 (HRRRv2) data stream.