Our proposed innovation harmonizes diverse data through spatiotemporal co-alignment in a POSIX-compliant data store to enable scalable, performant parallel processing required by in-depth event-based analysis for supporting risk-informed decision making in wildfire and water management.
Both the spatiotemporal co-alignment and event-based analysis innovations build upon the same technological foundation, i.e., the SpatioTemporal Adaptive-Resolution Encoding, STARE, a geo-spatiotemporal encoding methodology developed to support combining diverse datasets in their native states for integrative analysis. STARE encodes spatiotemporal coordinate locations, along with neighborhoods (or intervals, or resolutions), of the data elements using two (2) 64-bit integers in a hierarchical manner.
The spatial scheme of STARE encodes geolocation and spatial neighborhood hierarchically in 8 branches of quadtrees, whereas its temporal scheme encodes International Atomic Time and temporal intervals hierarchically also in a tree but with branching following calendrical units, such as day, week, month, etc. Mapping space-time intervals onto tree hierarchies then encoded into integers not only 1) provides an outstanding way to uniformly index and thus organize geo-data of different, irregular layouts with spatiotemporal co-alignment for scalable processing but also 2) establishes a solid foundation to facilitate efficient event-based analysis.
We plan to use the POSIX-compliant flexFS of Paradigm4 to implement a directory (folder) hierarchy mirroring that of STARE for Cloud web objects. Such a POSIX-compliant data store not only realizes the spatiotemporal co-alignment of diverse, unaltered data for easy, performant retrieval and utilization but also present the Cloud web object store in a “view” compatible with the all-familiar filesystem, to which most users are accustomed, e.g., in on-premises high-performance computing (HPC) environments and on individuals’ desktops or laptops.
Our technology possesses the unique capability of harmonizing geo-spatiotemporal data varieties for fusional analysis in their native resolutions and layout, including the vast barely-tapped resource of NASA Level 1 and 2 data currently in HDF files in Distributed Active Archive Centers. This data-variety harmonization facilitates spatiotemporal data placement alignment first in storage for effortless search-and-filter and second in memory for performant and scalable distributed parallel processing, including Cloud and minimizing duplication.
Our technology will enhance analysis productivity while reducing resource demand and cost for all geospatial analytics practitioners in academics, government agencies, and industrial-commercial organizations. The industries relying on geospatial analytics include, but are not limited to, transportation (air, sea, and land), logistics, tourism and travel, risk management, insurance, etc.