During this sprint of my capstone project, I worked on doing a final run-through of the process I had written for PHS’ LandCare updates and updated the documentation. Over the course of the project I had run each section individually, but had not run all together. Because I had failed to review the way in which many of the shapefiles interact with each other in further sections, I had not considered that some changes may affect other processes.
With my update test, I found that a number of the geometry functions could not work when tested all together. Some functions that failed include: copying or inserting geometries, exporting and updating geometries. Because of this, the first major issue to handle was the type of geometry the project intended to use. Originally, each shapefile had been imported with ArcMap ST geometry, a form that would not be compatible for export from SpatiaLite. Eventually after some level of significant research, I found that the SpatiaLite geometry with the SRID specified in the PHS selected projection (2272, Pennsylvania State Plane South) would work for this project. Eventually I worked out with some troubleshooting that an R-Tree Index could not be specified upon upload for further updates in QGIS.
Next, I found that without the ST geometry format, it would not be possible to use the ST_Union function that was relied upon heavily in the update process. This was remedied using QGIS. Once corrected, it would be possible to connect to the database in QGIS then utilize a plugin called ‘Geometry Updater’. This plugin allowed for us to manipulate geometries when moved from one table to another (like every single one of our update procedures). Once run, geometries were updated in the shapefiles based on the field specified using a reference shapefile.
Finally, when exporting, I found that the ‘CREATE TABLE AS’ option would not function with the changed geometry type. Because of our need to manipulate columns and change their names for the City of Philadelphia data management system, we needed a new way to export. Again, a QGIS plugin was used to change column names, delete columns, and reorganize them in the shapefile. The completed shapefile can then be exported and opened in ArcMap where the ArcMap specific metadata can be added. Because of the unique ESRI format, it was not possible to streamline this through SpatiaLite or QGIS.
When all is said and done, the final product takes about 30 minutes to run. This does not include the download of software and major excel preparation. As our new data changes come from one large audit excel, it is important to make sure we adjust the data for what we want before copying to the metacoding spreadsheets. An example of this could be the address updates and their ambiguity. Because some address updates will read as “Update address to 1501 – 1507 N 15TH ST” we will have to pick through and determine what needs to be added. While this can still take some time, the overall method is much faster than what was performed previously in ArcMap.