Best Practice recommendations for Data

If you are finding that map render times are not acceptable please consider whether any of these recommendations may help.

Zoom layering data with many records

A map image that contains many 10,000s of geometries will render more slowly. Displaying large amounts of data is computationally more intensive as it requires Spectrum Spatial to retrieve a larger number of geometries from the table or database and render them on the map. Also when a map containing this number of geometries is shown the base map is obscured and the map can look cluttered. In order to mitigate this issue we recommend applying appropriate zoom layer settings to the layer in MapInfo Professional before the Named Map is uploaded to Spectrum Spatial. Open the data in MapInfo Professional and make sure that the map Window is the same size as a map tile in Spectrum (approximately 512 x 512 pixels note that it does not need to be exact). Zoom in until there are no more than around 10,000 (for polygons/lines) or 25,000 (for points) records for a layer at its densest location. Zoom layer settings should be added to ensure the layer is visible only from this level down.

Providing alternative ways to show information

Sometimes it is useful to see the pattern of distribution of density of data. For example there may be 100,000s of customer records and showing them all allows their distribution to be seen. However rendering a map which shows 100,000s of records is not performant. An alternative approach would be to create a different layer to show density (perhaps a thematic map by county or other region, or a grid map). These approaches allow the data density to be shown much more performantly. The original customer layer can still be enabled when the user zooms in allowing them to get information on individual customer records if needed.

Complex Geometries

Some data has geometries with many 10,000s nodes. Flood boundaries are typical of this kind of data set and some datasets can have polygons which have over 400,000 nodes. When rendering a map even if the user is zoomed in, if the complex feature is within the map view then all of the nodes will be returned to Spectrum Spatial when rendering the map. There are two mitigation’s for this issue; thinning the geometries or splitting them.

Thinning Complex Geometries

Where the level of detail is not needed we recommend that complex geometries are thinned using MapInfo Pro prior to upload. We have seen cases where customers have used very detailed polygon boundaries for regions. One example was a map of French departments which were drawn to high detail, each department containing 100,000’s of nodes. Thinning the data to contain around 1,000 nodes per department did not affect the map detail which was intended to be seen for the whole of France as a thematic map but improved render time significantly.

Splitting Complex geometries

Where the level of detail is needed (such as flood boundaries needing to be shown at property level) we recommend that the geometries are split into multiple smaller geometries. There is no hard and fast rule, but if records contain more than around 30,000 to 50,000 nodes then some consideration should be given to splitting these geometries. To split a large number of records in MapInfo Professional you could create a grid and split the data using the grid.

Data in different projections

Many Spectrum customers use Bing and OSM mapping for the base map which are in the Spherical Mercator projection (also called Popular Visualization or EPSG:3857). However the overlay maps added to Spectrum are often in a different underlying projection. In these cases, when Analyst requests map images or performs feature queries, Spectrum performs a co-ordinate transformation on the geometries to re-project them into the required projection prior to rendering the map or returning the query results.

This process happens automatically and ensures that the overlay maps align correctly with the base map. For small datasets this process is quick but for larger data sets or data with complex geometries it can take time and the more complex or large the dataset the longer it can take. For example a query inside the map view on a TAB file with 200,000 records can take about 1 or 2 seconds to return. If the geometry has to be transformed the same request can take well over a minute. To mitigate this issue we recommend that larger data sets are saved in MapInfo Professional to the same projection used for the base map in Spectrum, and that these are used with Spectrum instead of the original data. This can be done using the File > Save Table As menu option and clicking the projection button to choose a new projection. For use with Bing and OSM pick Projections of the World > Popular Visualization.