In the last article, we got introduced to the concept and use of spatial analysis. Also, we briefly touched the error and solutions possible. Here we see the various types of Spatial Analysis.
Since Spatial Analysis covers a wide variety of areas, it is difficult to categorise the processes into definite boundaries. But the broad types of spatial analysis are:
• Spatial Data Analysis: Countries who deal with census report data and results of survey need this kind of analysis. The large amount of data comes with a large number of variables which are usually inter-related to each other. The use of Multivariate analysis or Factor analysis reduces them to a few independent variables. It becomes easier to spot smaller processes this way by the third factor analysis, which would have remained latent in the large amount of data.
Use of multivariate process started in the 1950’s but gained speed during 1970’s, after the use of computing devices for a large number of data gained popularity. A data matrix is used to determine the independent variables; it is not possible to compare data from two censuses. However, this can be solved by uniquely tabulating the different data matrices, which can be put to further use.
• Spatial Autocorrelation: This method estimates the extent of dependency among the observations in a space. A spatial weight matrix is created and measured, which shows the extent of relationship between observations in a given area for e.g.-length of a common border, distance between neighbours etc. If the matrix returns more positive values than expected, it indicates similar arrangement across the geographical region while a more negative result shows that the observations are dissimilar.
• Spatial Interpolation: This method is used to predict the variables at an unobserved location in an area using the values obtained from the observed locations. Method of inverse distance weighting is put to use.
• Spatial Regression: This method of spatial analysis sidetracks the problems of unreliable significance tests and unstable parameters, and at the same time provides information about the dependency of the involved variables. A variation of the same method, known as GWR which stands for geographically weighted regression, allows estimation of heterogeneity in the relationship between the dependent and independent variable.
• Spatial Interaction: Also known as “Gravity Model”, it measures the flow of material, information, people between a geographical location. Proximity Relationships are measured in terms travel time or driving distance. Factors include number of travellers in residential ares and attractiveness of destination variables. Also, the topological relationships must be identified.
• Simulation and modelling: Spatial interactions are aggregate models. The same characteristics are shared by urban models, based on mathematical programming. An alternate persepective has been presented which offers that the system should be presented at the peak point of disaggregation and then study the emergence of more complicated pattern from the bottom to the top.
• MPS: This abbreviation stands for Multple Point Geostatics. The algorithm serves the main purpose of spatial analysis of a conceptual model of geology. The process analyses the model and then creates realisations. There are recent developments of this process. Honarkhah-a process based method and Tahmasebi-unbiased towards any type of data.