Building Blocks of Spatial Analysis > Directional Operations

Directional Operations

<< Click to Display Table of Contents >>

Navigation:  Building Blocks of Spatial Analysis >

Directional Operations

Directional analysis is perhaps less well known and well used than other forms of spatial analysis. It primarily deals with the analysis of lines, points sets and surface orientation, sometimes from different time periods, examining patterns in order to identify and utilize information on specific directional trends. Researchers In the fields of geology, sedimentology and geomorphology utilize directional analysis to examine fault lines and fracture systems, glacial striations, particulate distributions and lithography patterns; in ecology it is used to study patterns of wildlife movement and plant dispersal; in hydrology and other forms of flow analysis directional considerations are paramount; in wildfire management directional analysis is used to predict local (30metre-100metre) wind flows as a function of local topography; and directional analysis can be used when studying patterns of disease incidence or the distribution of particular crime events.

If a particular dataset exhibits no discernible variation in direction across a study region it is described as isotropic. Whilst such situations may be regarded as unusual in the real world, it is frequently the starting assumption for many forms of analysis. Variations that have a particular directional bias are known as anisotropic. In some areas of GIS analysis, for example geostatistics, examination of anisotropy is a fundamental part of the modeling and prediction process (see further, Section 6.7). For an introduction that includes a discussion of hypothesis testing for directional data, and a review of applications of these techniques to spatial datasets, see Gaile and Burt (1980) CATMOG 25. It should be noted that when comparing and analyzing directional datasets computed with projected (plane) coordinate systems, there may be introduced directional bias. Comparisons should be made between datasets that have been generated using the same projections and approximate scale, or by utilizing spherical coordinates where necessary. For small areas (under 100kmx100km) the problems for most applications should be minor.