Calculation of a “remoteness index” for selected households
The three objectives of this project are to experiment with digitizing from satellite imagery, to experiment with the creation and modification of network transport models and to be able to compute measures related to remoteness for selected households in the area, with the help of GIS network analysis methods. The area of interest is the Mekong Delta, Vietnam. A survey was conducted where 638 households in the area were identified, in search of “pockets of poverty”. Both the locations of each household and a number of points marking the route taken to reach them were documented using GPS. Thus, the requirement for this exercise is to build (rather, extend the existing) network to connect the survey points and to measure the remoteness of each one of them.
The data for this project consists of the following:
|Existing road network (not very precise)|
|Location of survey points and group number of each point|
|Interviewers track-logs (points) with speed data|
|Interviewers track-logs (lines)|
|Centroid points for large urban areas (incl. Ho Chi Minh)|
|Centroid points for mid-size urban areas (communes with a population density >1000 people/km2)|
|Population data (polygons)|
The given network included only major roads, most of which could not “reach” the survey points. The first step for this exercise was to extend the road network. In ArcMap, a World Imagery layer was loaded behind the other layers. The World Imagery layer provide Earth Observation imagery in a concise way, similar to the way Google Earth does. In combination with the interviewers’ track-logs (points), the base road network was digitized and connected to the not-very-precise existing road network. While digitizing, each network line segment was given a default speed ID value, according to the table:
|ID||Description||Default speed (km/h)|
In occasions where the EO layer was unclear about the type of road, the ID was attributed according to the speed reading of the local tracking points. After the digitizing had been completed, the network was “cleaned” using the Feature to Line command, which ensures the user that all lines are connected. The network dataset was exported to new, complete Network, which included the existing, not very precise, road data. Prior to executing ArcMap’s New Closest Facility function, the newly produced network had to be tested for consistency. This was done by running a few simple A to B routes. The final step was to actually find the urban areas (major and mid-sized) close enough for each one of the survey points. The new layer was created using the New Closest Facility tool from Network Analyst toolbox. This function finds the cost of traveling between two given groups of locations (incidents and facilities) and determines which is nearest to one another. The tool ranks the results by least impedance.