A Study of finding Sites for Ecological Villages, Using Multi Criteria Evaluation
Data and method
The study area is located in southwest of Stockholm and it includes the towns of Sodertalje and Tumba. The data used for this project are summarised in table 1 below
Table 1: A summary of Data used
| Serial No. | Data type | Data description |
| 1 | Land use | Data of different ways the land has been put to use. It ranges from uncovered to covered areas and low exploited to high exploited areas. |
| 2 | Soils | Information on various categories of soils within the study area. Examples include clay, silt, sand, gravel till etc. |
| 3 | Elevation | Information on the topography of the area. |
| 4 | Bedrock | Data on the types of underlying or parent rock in the area |
To find suitable spots for the ecological villages certain factors and constraints were seriously taken into consideration in the evaluation. Based on the evaluation of suggested criteria, four important factors that facilitate ecological villages were identified. These factors include, close proximity to agricultural areas (food), proximity to aquifer (water), proximity to wetland (waste and sewage) and proximity to service centres (jobs and services). Other factors that restrict the usage of an area (called constraints) were also noted. They were technical accessibility (which prevents residential areas to be built on areas regarded as inaccessible), radon-gas risk areas, higher grounds, areas close to beaches and natural reserves.
All that is needed for the constraints is the creation of Boolean image -an image containing only zeros (excluded areas) and ones (permitted areas). These images were created by dividing the maps containing the constraints into two, one for unacceptable area (zeros) and one for the acceptable ones
Since we are interested in finding 5 areas, each being able to house 50 to 100 families, areas required should be about 0.25km2 to 0.5km2 in size. This means that small areas need to be excluded. This was achieved by using appropriate series of methods in Idrisi to get the size of the individual areas in hectares. Areas below a certain threshold were removed to get maps of extensive areas. For instance, thresholds set for factors like agricultural areas; aquifers, wetlands etc. were areas larger than 20, 5, and 5 hectares respectively. Distances on these maps were calculated based on the scores according to proximity to these desired factors (resources). The factor maps were then stretched to get the interval score of 0-255. These stretched maps with low values for close proximity were inverted to obtain high scores for close vicinity.
After checking all the criteria maps, a set of weights were developed for the factors based on relative importance. These weights were real numbers ranging from 1/9 to 9/1 that sum to 1.0. In the matrix below, agric was assigned as the most important, followed by aquifer, job and wetland as the least important. The exercise produced respective eigenvector (values) of weights that show the relative importance of the factors with the most important one having higher score and likewise for the least.
To test the robustness of this decision rule, the weights assigned to the factors were slightly changed (as shown in fig.2 below). This means that importance attached to the factors by way of values were changed. Here aquifer was the highly important one, followed by agriculture, wetland and job in that order.
Matrices for the weights have been summarised below. Figure 1 is the original weights whiles figure 2 represents the slightly altered weights.
| | Job | Agric | Aquifer | Wetland |
| Job | 1 | | | |
| Agric | 5 | 1 | | |
| Aquifer | 3 | 0.33 | 1 | |
| Wetland | 0.33 | 0.142 | 0.20 | 1 |
Fig 1: matrix for relative importance for pairwise factors
| | Aquifer | Agric | Job | Wetland |
| Aquifer | 1 | | | |
| Agric | 0.2 | 1 | | |
| Job | 0.142 | 0.33 | 1 | |
| Wetland | 0.33 | 3 | 5 | 1 |
Fig 2: matrix for relative importance for pairwise factors (2nd iteration)
Multi criteria evaluation (mce) was then conducted on the factors and the constraints to get the statistics for the chosen areas. The statistics comprises size (hectares), average, standard deviation, mean, variance etc. Best areas were selected based on their average score, the area (size), mean and variance values. This was done manually by assessing the summary statistics of the different sites. Five areas that are quite extensive in size and have higher average values were selected as the suitable sites.
This methodology has been summarised and may be visually comprehended from the macro modeler below: