Experimental Survey of Geospatial Big Data Platforms

Recent advances in geospatial data acquisition techniques are instrumental in the generation of massive data that are being processed by geospatial big data platforms such as Spatial Hadoop and Geo-spark. While some of this data is stored in databases, much of the data is unstructured and temporal. In this paper, we survey alternatives available in geospatial big data frameworks. We present a comparative study of the different approaches and an experimental evaluation of the two most used platforms Geospark and Spatial Hadoop. We discuss our evaluation results in the context of various tasks in commonly used geospatial processing tasks, especially in the context of Volume, Value, Viscosity, Variability, Volatility, Viability, Validity and Variety.