Using GIS Raster Data Analysis with Reconstructability Analysis Tools
Reconstructability Analysis (RA) has been successfully used to analyze a myriad of different data types. At its core it relies on a row-based approach to data where each row represents an observation. RA has been used successfully on time-series data, as well, using a masking technique.
This talk demonstrates using RA on spatial data, specifically GIS landuse raster data sets. This approach looks at the neighborhood around a target cell and linearizes it into a row of data that can be processed using RA techniques. This algorithm is sequentially applied to all cells in the raster data set, similarly to how one would analyze imagery data.
Different neighborhoods are examined, including the Moore, Von Neumann, and 5x5, with evaluation of the results of each. In the end the Von Neumann neighborhood yielded the best results, and the RA output proved to be more illustrative of patterns of influence than other techniques in the GIS arsenal such as Moran’s I. The next phase of development is to combine the time-series and spatial masks into a space-time cube that can be linearized for analysis.
Percy (David Percy) has been tweaking scientific databases since 1984 starting in medical research at Legacy Health Systems and Oregon Health Sciences University. During the mid-90s he retrained as a geologist, joined PSU Geology in 1998 as a data manager, and began teaching GIS classes in 1999. He has worked on national and international standards for geologic databases and helped construct comprehensive databases of glaciers and coastal deposits. Always interested in Systems Science, he began casually taking classes during the early 90s, eventually completing an MS in Systems Science in 2016, and continues to collaborate with Marty Zwick on space-time data cubes in RA.
Reconstructability analysis, System design, System analysis, Raster data, Geographic information systems, Geospatial data
Computer Sciences | Geographic Information Sciences
Percy, David, "Using GIS Raster Data Analysis with Reconstructability Analysis Tools" (2020). Systems Science Friday Noon Seminar Series. 7.