The analysis of vector and raster-based modelling proves the existence of significant differences when comparing the final results. Vector modelling shows higher yields at higher prices than raster modelling. Evidence of this can be seen visually in Figure 11 and 12. The significant differences were also reconfirmed in Table 4, displaying the results of Paired T-tests, which were ran at the 95% confidence level, and confirmed that there is indeed a significant difference between raster and vector modelling. A significant difference is also found individually as each type of biofuel residues showed significant differences. Network analysis recognises and updates the topology of the landscape more accurately than raster. By implementing geometry calculations the assessment of area is more refined when using network analysis. Raster modelling does not take into consideration road networks and compensates for this inaccuracy by using a tortuosity factor. All these trade-offs are at play in determining why network analysis resulted in higher yields and prices when compared to raster modelling.
While the process of determining the accuracy of our results is sound, research limitations and problems became apparent that may have affected our results. A tortuosity factor of 1.35 was used in order to account for the lack of road networks and topography found within raster datasets. This was deemed an average tortuosity factor by Calvert and Mabee (2014). Our results may have been affected by this as an average is used rather than an exact value. Also, this can further explain the difference between supply-cost curves derived from vector and raster modelling. Another research limitation that became apparent was converting raster to polygon vector files. Some polygons may have been simplified, affecting the accuracy of the area of specific residues. Again, this further explains the difference between supply-costs curves derived from vector and raster modelling.
Although research limitations and problems arose, the supply-cost curves are believed to still be accurate and thorough in assessing any significant differences between raster and vector-based modelling. Thus, the findings of this project prove that there is a significant difference between vector and raster-based modelling when deriving supply-cost curves for bio-energy production in southeastern Ontario.