Server-side clustering for mapping in Drupal 7 based on Geohash: Geocluster increases performance of data-heavy map visualizations by clustering results on the server-side.
Read my Geocluster master thesis.
1. Install Geocluster and update existing content
Re-save any previously created content using node_save or Views Bulk Operations, so that the Geocluster index is updated.
2. Create a clustered GeoJSON Feed
- Create a display based on Views GeoJSON
- Add the Geofield and exlude it from display
- Enable Geocluster in the Format Settings of your View
- Add Geocluster fields for aggregates of latitude, longitude, count
- Use Geocluster lat, lon as data source in the Views GeoJSON format settings
- Make sure that no additional fields are used for aggregation, you can still use them with aggregate functions like GROUP_CONCAT
- You maybe want to save this step until everything else works: set the item limit to 0 so that all your content will get clustered
Note that Views GeoJSON is the obvious example, you can also use other display formats for representing clustered data.
3. Create a map that shows the clustered data
Use Leaflet GeoJSON Bean to create a map based on the clustered GeoJSON feed at
For Apache Solr: enable geocluster_solr, create a Geocluster Solr server + index where you index the Geofield, latitude & longitude values and Geocluster index fields and create a Search API View based on that index.
- The PHP based clustering algorithm is for testing purposes only and scales up to 1,000 items.
- The MySQL Aggregation based clustering algorithm scales up to 100,000 items.
- The Search API Solr based clustering algorithm scales beyond 1,000,000 items using Search API Solr.
See a visual comparison of the different algorithms.
How geocluster works
Spatial index based on Geohash
Geocluster adds a hierarchical, spatial index to geofields based on Geohash. Every geofield geographic point coordinate gets encoded into a string identifier. Every character of the geohash divides space into sub-quadrants. For each geohash identifier, all its substring-prefixes from 1 to n length are stored separately. This creates an index, upon which the clustering algorithm can then be performed.
Cluster points on request
The clustering algorithm
- determine the geohash index level to cluster on, based on the bounding box size, current zoom level and configured cluster distance
- dynamically cluster (aggregate) all points within cells that share a common prefix at the calculated index level
- check for overlaps at the edges of each cluster cell and merge clusters if appropriate
Return clustered results
The clustered results are returned. The provided Leaflet integration visualizes clusters and provides appropriate interactions: a click on clustered results will zoom-in whereas normal results are attached the default popup behavior.
for a visual explanation on what's going on.
- Geoclustering is a server-side Drupal 6 solution
- Openlayers Quick Query - related Drupal 7 sandbox project
- Blog post on Leaflet maps with Drupal based on a (more efficient) custom database query