Directed Edits/METEOR Digitizing Kathmandu

From OpenStreetMap Wiki
Jump to: navigation, search

This page describes the directed editing activities by Kathmandu Living Labs (KLL). KLL is working together with the Humanitarian OpenStreetMap Team (HOT) on the Modelling Exposure through Earth Observation Routines (METEOR) project. HOT is working together with the British Geological Survey, ImageCat, and the Global Earthquake Model on a UK Space Agency International Partnership Programme, focused on developing innovative applications of earth observation (EO) technologies to improve understanding of exposure to help minimise risks from natural hazards. KLL, is directly working with HOT on the digitization and then subsequent building exposure survey of Kathmandu Valley.

Kathmandu Living Labs (KLL), is a civic tech company that specializes in Open Mapping and creation of digital infrastructure that brings government, nonprofits and businesses together to leverage the latest technological tools to provide everyday citizens with better services. KLL has been working to improve open data of Nepal through the integration and use of OpenStreetMap (OSM) in all its projects.

Project Overview

  • Humanitarian OpenStreetMap Team is a 501(c)(3) not-for-profit organization and global community. It has been an integral/pioneering supporter of the use of OSM for humanitarian support.
  • Please click this link for the specific project HOT website
  • Learn more about the project through the official webpage

Goals and Deliverables

The escalating impacts of natural hazards are caused mostly by increasing exposure of populations and assets. It is estimated that the world will see the construction of 1 billion new dwellings by 2050 and this growth may lead to rapid increase in risk.

The project aims to build and strengthen local and global resilience through complete, up-to-date, accurate exposure data in order to better identify risk and enable more effective decision-making:

  • Deliver exposure data for 46 of the least developed Official Development Assistance (ODA) countries
  • Capacity-building of local decision makers to apply data and assess hazard exposure
  • Create open protocols to develop critical exposure information from EO data

The country-wide datasets developed by ImageCat will be tested for their suitability of purpose against OSM data surveyed on the ground. One such survey will be conducted in Kathmandu and is being led by KLL. This work will require the digitization of building footprints and a second phase will require the collection of building attributes from a sample of households in Kathmandu Valley.

Mapping Timeline

The first phase of this project, remote mapping of buildings lasted from 1st November 2018 - 30th November, 2018. The second phase of the project entails building attribute data collection and took place from 1st January 2019 till 1st March 2019. The final phase of the project, data field validation, verification, and upload took place from 15th February 2019 till 15th March 2019.

Area of Interest

The project is limited to Kathmandu Valley. Seven homogeneous zones were created by ImageCat. For specific areas within the valley see https://tasks.hotosm.org/project/5432

Homogeneous Zone 1


Remote Digitisation

  • For the remote digitsation process eight Geomatics Engineering students were recruited from Pokhara, Nepal. These students along with our expert mappers used satellite imagery to trace and correct OSM data of Kathmandu. For the remote digitsation process eight Geomatics Engineering students were recruited from Pokhara, Nepal. These students along with our expert mappers used satellite imagery to trace and correct OSM data of Kathmandu. Further detail on how this was done can be read here
  • For tracing buildings HOT Tasking Project 5432 was created.
  • Specific hashtag for this project are as follows: #hotosm-project-5432, #METEOR, and  #digitizingkathmandu
  • The imagery provided for the remote digitisation of building footprints within the homogeneous zones identified in Kathmandu was kindly provided by USAID in Nepal. It was provided through a Tile Map Service (TMS), which was hosted at their dedicated GeoCenter. The satellite imagery was captured by the WorldView-4 sensor in December 2017, making it a little under a year old. Before mapping, wherever possible the new imagery is aligned with the free Bing imagery (2015-2016 imagery). This is done so that the data created during the project is aligned to the rest of the data of Kathmandu and does not hinder navigation.
Satellite Imagery Details
Product WorldView-4
Acquisition Date 30-12-2017
Off-Nadir 4.4
Cloud Coverage 0%
Resolution 30 cm


Homogeneous Zones Building Count
Before After Increase
1 Residential 8,909 16,564 46%
2 Dense Residential 71,310 104,945 32%
3 Urban 8,829 10,671 17%
4 Industrial 1,274 2,259 43%
5 Informal 175 752 77%
6 High Urban 595 749 21%
7 New Industrial 510 675 24%
Total 91,602 136,615 33%

Kathmandu Buildings Count Pre and Post Digitisation

  • Before and after digitization

The outcome of the digitization process in before and after map form can be viewed through this link: https://kathmandulivinglabs.github.io/ktm-buildings-before-after/

  • The following mappers contributed to mapping remotely

Manoj Thapa; Ro Sun; asmitasummer; Shrestha Suman; Pradeep Rana; Tejendra kandel; Aman KC; Rabin Ojha; Anil Basnet; Berzerk505; sharmanubhav; Roger Anish; Gaurav Thapa; OSKUDA; Parassrest; Sazal(Solaris)

Field Data Collection

  • Ascertain sample households

There are seven development patterns (homogeneous zones). We are keeping the error margin at 5%. This means that 40 randomized points per development pattern were identified and then 10 households per point was sampled. This gave us a total number of 2,800 households. However, development pattern 5 only has 20 randomized points. This means that 2600 buildings were to be surveyed in total. However, in reality several points fell under military/government complexes where ground survey is prohibited. In these cases where possible alternative points were chosen and when not possible these points had to be ignored. Unfortunately, one particular development pattern had  

Randomized points were generated from a shapefile created by ImageCat containing 53,000 randomized points. The points were then clipped to each development pattern. Then the points were sorted based on their UID and the top 44 points were selected. The 4 extra points per development pattern are there as a backup in case it is impossible to map around any of the first forty.

  • Recruitment of surveyors

Recruitment of surveyors began from the month of January. For this project surveyors were primarily civil engineering students with a sound understanding of building structures. A local Engineering consultancy firm, was brought on-board by Kathmandu Living Labs to identify and recruit these surveyors. Group interviews were conducted on January 13 and 14 to ensure a match in terms of technical knowhow and motivation. After the recruitment process, 12 surveyors were selected for the project.

  • Surveyor training

Training of surveyors took place on January 16 and 17 at Kathmandu Living Labs Office. A training session was conducted by two senior civil engineers and provided details on building structures, materials commonly used in Nepal and challenges of visual inspection of buildings. Second day of the training session focused on the use of data collection tool, OpenMapKit (OMK) and the process of collecting and uploading the collected data. For this the surveyors were also taken out to the field where they collected real world data. This data was subsequently evaluated and feedback were given to the surveyors. Before surveyors were sent out to the field a map with all the randomized sample points was created. From local understanding of Kathmandu and its roads six clusters were created. The clusters were then drawn on GIS. We then counted how many points each cluster contained and adjusted the clusters so that all six clusters contained between 45-55 points. Based on these clusters six groups of two surveyors each was created. Groups that had longer travel times were assigned fewer points. Maps of the field areas were created and given to each group. The maps ensure that that there are no overlapping field areas and thus prevents duplication of collected data. Throughout the data collection period a weekly meeting was held. These in person meetings were used as a place for surveyors to discuss the challenges they faced, ask specific technical questions to experts, learn more about OSM/OMK and get feedback on the quality of the data being collected. Desk validation by our experts were conducted two times. Once when 50% of the data was collected and second time when 90% of the data was collected. To do this .csv file of all survey groups were downloaded. These were then divided into six groups based on the individual name/device id through which the data was uploaded.

Once sorted into groups each tenth row was validated by experts based on the images gathered by the surveyors. This is done so that 10% of all data brought in is validated through desk research.  

Spot checking occurred by taking the 10% data that had been validated through desk research out of which latitude and longitude of every tenth row was noted. These locations were then visited to ensure that data on the ground matched the images and data collected by field surveyors.

Data Model

  • The attributes to be collected are built upon the GED4ALL taxonomy developed as part of the GFDRR Challenge Fund. The OpenDataKit questionnaire contained the following building features and their attributes listed below. Two civil engineering experts were brought in by Kathmandu Living Labs to localise the values. They were also able to tell us what building materials were more likely to be found within Kathmandu Valley and needed to be added as options.
General Title OSM Key OSM Description
Material of the Lateral Load Resisting System building:lateral:material=* Proposed lateral load resisting material tag
  • concrete reinforced
  • steel
  • steel-concrete composite
  • brick
  • stone
  • adobe (earth)
  • timber/wood
  • bamboo
  • light gauge steel/cold formed steel
  • others
Lateral Load Resisting System building:lateral:system=* Identify structural system of buildings
  • moment resisting frame
  • shear wall
  • braced frame
  • dual (frame wall) system
  • masonry wall
  • confined masonry
  • hybrid
  • others
Height building:levels=* Number of above-ground levels of a building
building:levels:underground=* Proposed tag for number of below-ground levels of a building
Date of Construction or Retrofit building:age=* Proposed age tag, associated with buildings
  • pre_2000
  • post_2000
  • unknown
building:condition=* Describe the condition of the building
  • good
  • average
  • poor
  • unknown
Occupancy building=* Describe the building purpose
  • residential
  • commercial
  • public
  • mixed_use
  • Industrial
  • agriculture
  • assembly
  • government
  • educational
  • health
  • unknown
capacity:persons=* Describe the number of people a building can support
Roof roof:shape=* Well known roof shapes
  • flat
  • pitched
  • monopitch
  • double_pitch
  • sawtooth
  • curved
  • complex_regular
  • complex_irregular
  • unknown
roof:material=* Outer material for the building roof
  • cgi
  • bamboo
  • thatch
  • mixed
  • masonry
  • earth
  • concrete
  • metal
  • wood
  • fabric
  • slate
  • stone
  • clay
  • unknown
Neighbouring condition building:adjacency=* Describe the neighbouring condition of the building
  • attached
  • free_standing
Geological site condition building:geological_site=* Describe the geological site the building is built upon

flat_land; river_bank, slopy_land; landslide_prone_area (multi select option)

  • Project feature source tag

For all data uploaded through the field survey a source tag ‘METEOR Kathmandu Field Survey 2019’ has been applied.

MapCampaigner


MapCampaigner

Further details on KLL's approach to field data collection can be found here.

Results

At the end of the field work a total of 2,701 buildings were surveyed. Some of these buildings had pre-existing attribute information following the tags identified from the data model such as building, building:adjacency, building:levels, roof:material and roof:shape, while the other tags were not present at all. The below table shows how the information uploaded through this project has added the tags outlined in the data model to this pre-existing information.

Tags Before After % change
building 2589 2701 4.1
building:adjacency 118 2694 95.6
building:age 0 2689 n/a
building:condition 0 2701 n/a
building:geological_site 0 2682 n/a
building:lateral:material 0 2701 n/a
building:lateral:system 0 2701 n/a
building:lateral:system 0 2701 n/a
building:levels 146 2701 94.6
building:levels:underground 0 2694 n/a
capacity:persons 0 2701 n/a
roof:material 65 2701 n/a
roof:shape 685 2701 n/a
source 119 2701 95.6
fixme 0 366 n/a


Sustainability Plan

  • Kathmandu Living Labs is a local organization and leads the OSM Nepal community. They continue to monitor and improve data in Nepal beyond the project period.
  • All surveyors were also trained on the basics of mapping and in addition at the end of each survey  form a ‘fixme=*’ tag was kept. This tag was used to describe any mapping errors discovered by the surveyors on the ground. Some examples could be “north side of the building is attached to a neighbouring wall” or “the building complex is actually 4 separate buildings evenly divided”. This tag has been specifically included to maximize the improvement of OSM. It has also made it easier for surveyors to continue improving OSM after the project period.

Contact Details

  • You can contact the team through User: Gaurav Thapa. Technical Supervisor and Local Project Manager.
  • Other mappers active in this project are User: Manoj Thapa and User: Ro Sun
  • Project Manager and Mapper active from HOT's side User: Mhairi

See also

OpenStreetMap's Directed_Editing_Policy