Last week, as a part of improving the quality of road network data in Taiwan in OpenStreetMap and correct data misalignment issues in Taiwan, the Mapbox data team along with the Taiwan OpenStreetMap community has completed aligning major roads in 6 cities in Taiwan.

For this task, the team used updated Mapbox satellite imagery and Strava heat map for aligning the major highway network.

We have presently managed to align and improve more than 4000 kilometers of major highways in Taiwan in just 3 days by working alongside the Taiwan community. The 6 cities where the aligning effort is concentrated are: Taipei, New Taipei, Hsinchu, Taichung, Yilan and Tainan.

New updated Mapbox satellite imagery in Taipei

… See full entry

Guides to improve navigation data on OSM

As a part of our recent push to improve navigation data on OpenStreetMap, the team at Mapbox started mapping turn lanes with the help of local mappers in San Francisco, Washington DC and Los Angeles areas. The initial documentation on turn lanes that was referred during the mapping project was the OpenStreetMap wiki on turn:lanes. However as the mapping started progressing, our team started to discover edge cases which have not been covered adequately on the wiki.

As a result, we started sharing our observations and discussed mapping strategies with the OSRM team, which we compiled into a guide for Mapping to improve navigation data in OpenStreetMap. The guide is designed to help anyone who wants to improve OpenStreetMap data from the perspective of navigation, to understand routing and guidance, and different scenarios that can help during data validation.

… See full entry

This month local mappers, in collaboration with our data team at Mapbox, completed the task of adding missing buildings in the SF area. Our sprint focused on the whole of SF Peninsula covering the areas of Sunnyvale, Mountain View, Palo Alto and Menlo Park areas covering ~1290 sq.km. This is a continuation of our earlier efforts from 2014 when we addded ~114,000 buildings in San Francisco.

Based on the existing building density, we estimated a total of ~180,000 missing buildings in the project area, and finally traced ~173,447 buildings over the last two months.

Methodology and Observations

• An initial trial run was conducted to determine a rough estimate of the amount of buildings to be added. It was crucial to gather an idea to timebox the task to a particular timeframe.
• Regular reviews were conducted to determine any gaps in coverage and accordingly create new tasks.
• There were outdated imports in the area which needed to be cleaned up. Accordingly clean up tasks were created for these areas. Since these misaligned footprints aren’t due to any mapping efforts a separate ticket was created for the clean up tasks in mapping repository.
• Imagery is outdated in some cases, where new buildings have cropped up. One such instance was the new Apple spaceship campus in OpenStreetMap which was absent from Bing/Mapbox imagery.

… See full entry

Mapping my hometown: Jorhat

[Jorhat] (https://en.wikipedia.org/wiki/Jorhat) is a small town in Upper Assam and is known as the cultural capital of Assam. A small town at it’s heart but one of the major centers of education and commerce in the state of Assam. Jorhat also has historical significance in that it was one of the erst-while capitals of the Ahom dynasty, which ruled Assam for close to six centuries before the colonial rule. Surrounded by lush and verdant tea gardens and numerous wetlands and rivers, it is the only place which tells me I am home. Logically this was the place from which I would like to start my mapping.

Here is what the city looked like before I started mapping:

Though some initial mapping had been done in the city, the city was still missing a majority of the roads, streets and the numerous institutions that make Jorhat such an integral part of Assamese culture. The major streets in the town were still to be mapped, the iconic institutions in the town were not yet mapped and also the most of the road network were still missing.

… See full entry