Comparison of the military maps of trafficability developed by different methods

Abstract

This paper presents the basis of the system for automation of the military trafficability maps development. The system is based on determining the index of trafficability (IOP) for square primary fields of 1 km by 1 km size. It is defined with use of data on land cover elements, obtained from both military and civilian spatial databases. To determine the trafficability, various methods have been used: artificial neural networks (multilayer perceptron and Self Organizing Maps) and GIS multi-criteria spatial analyses. Tests were performed for both military (Vector Map Level 2) and civilian spatial databases (Corine Land Cover and Open Street Map). The article presents a comparison of used methods. In addition to the basic statistical analyses (average value of IOPs, Pearson's correlation matrices, etc.), spatial statistics such as spatial autocorrelation coefficients (Moran I) were calculated. The key experiment was also to compare generated maps to the map of trafficability made by the methods commonly used in the Armed Forces. The result of conducted experiments is therefore choosing the optimal map of trafficability and the answer to the question: which method and data should be used for the best trafficability map elaboration? It was found that the most useful and most faithful map showing the conditions of trafficability is the study elaborated using the Vegetation Roughness Factor method and using Corine Land Cover data.