(2) Weighted Cost Distance Analysis
Before conducting the cost distance analysis, the friction surface was first prepared. The existing roads in Singapore were given a friction value of 1, while the rest of the land area was given a friction value of 100. Thereafter, the cost surface [Spatial Analyst > Distance > Cost Distance] was created, together with the backlink raster. This was done for the two regions identified in Part 1.
Centroids were then created for the 53 planning areas that are part of our study area. In essence, this next step that we tried to do was to find the least cost path from all planning areas to the two regions, which were generalised as point features. As this would in theory require us to perform 106 cost distance iterations (53 planning areas multiplied by two regions), we used the Model Builder in ArcGIS to help us quicken the process. The model that we created is shown in the diagram to the right. It consists of four functions: (1) to select a single centroid through Make Feature Layer; (2) performing the Cost Path function to find the least cost path in raster format; (3) performing a Raster to Polyline function to convert the raster line to a vector line; and (4) to Merge all individual polylines created using this model into a single layer.
This process took a total of more than three hours and the final two maps that we obtained are shown below. For the bottom map, an additional route is added between the two regions as the waste ash produced in Region 2 has to be transported to Region 1, where the Tuas Marine Transfer Station is located, and where the ash will subsequently leave the main island to be ferried to the offshore Semakau Landfill (click here for more information).
(1) Waste Distribution across Planning Areas
Data for annual waste incinerated was obtained from the National Environmental Agency (NEA) website. The data for waste dumped at incineration plants from 2005 to 2012 were used and extrapolated to give the amount of waste that will be generated in 2018 which is when the new incineration plant would be ready.
Thereafter, the total amount of waste was divided up into planning areas. The waste statistics is classified into Trade & Domestic (T&D) - which includes waste from homes, schools, trade premises and military camps - and Industrial & Commerce (I&C). The proportion of waste generated in each sector is 56% and 44% respectively [1]. The waste amounts are then distributed among the planning areas according to residential population (for T&D) and area size (for I&C). The map for the final distribution of waste used in our analysis is shown on the left.
Two planning areas - one in the north west, and one in central Singapore - were assigned a waste generation amount of zero. These two areas are the Western Water Catchment and the Central Catchment Nature Reserve respectively. These areas have few urban structures (see Figure 3 and Figure 4 for satellite imagery) and they are mostly used for recreation, light agriculture and military exercises.
[1] Ministry of the Environment and Water Resources. (2013). Solid waste management. In Key Environmental Statistics 2013. Retrieved from http://app.mewr.gov.sg/data/ImgCont/52/MEWR%20KES%20booklet%20updated%20Aug%202013.pdf
Thereafter, the Solver tool available in Microsoft Excel was used to find the minimum cost of transporting waste generated in all planning areas to the two regions. This step will establish the planning areas whose waste should be sent to the respective regions. Solver allows us to set the (1) objective: to minimise transportation costs; (2) constraints: that the total amount of waste sent to each region should not exceed the region’s daily capacity by too much; and (3) the changing cells: the binary cells that would be repeatedly changed by Solver to determine the region that the waste should be sent to. Due to the computational limitations in solving such a large number of possibilities, Solver was left to run for around two hours for each scenario before an intermediate trial answer was used for our final maps.
(Authors' note: In the interest of brevity on this webpage, the explanation for this step provided here is not comprehensive. If you wish to find out more about this Solver model, please click here to read more or contact us should you need any clarification.)
part 2: cost distance analysis
methods
Click image to enlarge
Click image to enlarge
Click image to enlarge (PDF, 1.4mb)
(PDF, 1.4mb)
(PDF, 1.5mb)
Figure 3 - Western Water Catchment
Figure 4 - Central Catchment Nature Reserve