| Popis: |
To find the optimal by-pass road alignment in the Tlokweng Planning Area in Botswana, a multicriteria spatial-based model is developed using the GIS-MCE approach. In respect to the environmental impact assessment (EIA) approach, in this research a set of criteria is classified under 3 themes – economic, environmental and social. To determine the criterion and theme weights and perform their aggregation, the Analytical Hierarchy Process (AHP) and Weighted Sum (WS) are utilised. The least-cost path analysis is used to produce road alignments. The entire model is developed using ModeBuilder in the ArcGIS 10.5 environment. Depending on the theme’s preference, 4 road alignments are produced: economic, environmental, social and equal. Comparing GIS-produced alignments and the planned route proposed in the “Tlokweng Development Plan 2025” is carried out by applying an independent validation matrix using the DEFINTE software package. The investigation into the robustness of the model is completed by examining the model output to identify criteria sensitive to weight changes. For this purpose, One-At-a-Time (OAT) sensitivity analysis and the statistical test for zero proportion are used. Sensitivity analysis results for criteria identified as sensitive are also presented spatially. To carry out the sensitivity analysis, a standalone Python script has been created which communicates with ArcGIS 10.5 through the ArcPy module. This study has successfully investigated, developed and applied the MCE method for optimal planning of highway and road alignments together with a sensitivity analysis for the MCE method. The results show that the social alignment is the best of the 4 road alignments – economic, environmental, social and equal. The results further confirm that the planned alignment, not produced by applying the multi-criteria approach, substantially differs from the 4 mentioned routes produced in GIS. The results of the sensitivity analysis and statistical test for zero proportion reveal 6 criteria as sensitive. The criterion referring to land use/land cover displays the greatest difference in the results – the model becomes sensitive at -5% of this criterion weight change. Further research is recommended to increase the robustness of the model. Some recommendations are further analyses regarding the assignment of criteria weights, standardisation of criteria and applying a “global” sensitivity approach in which more than one criterion weight is changed at a time. |
