Assessing the impact of uncertainties and errors on multi-criteria evaluation procedures when using remote sensing derived land cover input data

Date of Completion

January 2006


Physical Geography|Remote Sensing




Spatial decision-making procedures, such as multi-criteria evaluation (MCE), often rely upon input data derived from remote sensing in the form of land cover maps. However, these data can often contain inherent errors and misclassifications that can influence the accuracy of the final output associated with spatial decision-making procedures. The research presented here explores what influence errors and misclassifications in remote sensing derived land cover data can have on the output from spatial decision-making procedures. Four land cover data layers, each covering the same spatial area and easily obtained from scholarly or governmental institutions, are used throughout this research. These comparative land cover data layers are compared with an author-created land cover data layer known as the BASE layer. The BASE layer represents the nearest approximation of the land cover condition in the study region. Overall land cover classification accuracy is assessed for each of the land cover data layers used in this research. In addition to overall land cover classification accuracy, classification accuracy is evaluated along the edge and interior of land cover regions in order to determine if the spatial location of errors and misclassifications, rather than the quantity of errors and misclassifications alone, can have an impact on the output form spatial decision making procedures. This is followed by an assessment of the influence of errors and misclassifications on site and situational factors, two of the primary components of spatial decision-making procedures. Direct comparisons are made between the BASE layer and the comparative land cover data layers. A series of multi-criteria evaluation procedures using the BASE layer and the four comparative data layers as input concludes this research. The output from the spatial decision-making procedure using the BASE layer is directly compared with the same spatial decision-making procedure using each of the comparative land cover data layers as input. Differences between the BASE layer and the comparative land cover data layers are then assessed to determine if errors and misclassifications have had an impact on the output of the spatial decision-making procedures.^