Description: <p style="margin-top:0px; margin-bottom:0.75rem; font-family:&quot;Avenir Next W01&quot;, &quot;Avenir Next W00&quot;, &quot;Avenir Next&quot;, Avenir, &quot;Helvetica Neue&quot;, sans-serif; font-size:16px; max-width:100%;"><span style="max-width:100%; display:inherit;">ARTC data is ingested into an enterprise data warehouse and processed monthly by the NFDH using the following methodology:</span></p><ul><li>Snap train GPS pings to the nearest point on the nearest rail segment in a foundation rail map. Prioritise standard/dual-gauge.</li><li>The default precision tolerance is 10km. This is lowered algorithmically to drop more fuzzy pings in areas where the rail network is more complex and more pings are generated by slower trains.</li><li>Median correction distance is 2.3 metres in urban centres and 2.6 metres in regional areas. 84% snaps to a track within 10 metres.</li><li>After correction, GPS pings can be aggregated by the rail segment they are assigned to. They are also merged with train attributes in the schedule data and fitted into various boundaries to derive origin and destination for each train service.</li><li>A single origin (destination) is selected for each train service, this is often the first (last) ping where the train stopped in an intermodal terminal that is also in or assigned to an urban area. This has to be within 50km of the first (last) ping recorded by the transponder.</li><li>The is then summarised to calculate the average speed between pings (where the speed is not zero) along each rail segment, grouped by commodity type and combinations of origin-destination pairs. This is then combined with the foundation rail map to produce this dataset.</li></ul>