STEM3003 Modelling in Space and Time, Geostatistics and GIS

2019

4.5

1 x 1-hour lecture weekly
1 x 3-hour computer lab weekly

1 of STEM1002, GEOG1003, GEOG2700

1 of GEOG3731, GEOG3751, GEOG7711, GEOG8731, GEOG8751, STEM8004 has been successfully completed

Assignment(s), Laboratory exercise(s), Test(s)

Computer models designed to analyse processes operating space and time are useful tools for gaining an understanding of human, plant and animal responses to their environment. These help professionals understand underlying drivers, predict/map outcomes, prioritise management strategies and quantifying risks, particularly when they can be visualised in a GIS. However, spatial (and separately temporal) autocorrelation needs to be dealt with explicitly, otherwise assumptions relying on data independence are violated.

Geostatistics is also a popular technique in the sciences for measuring the level of spatial dependence in data. An important component of Geostatistics is surface interpolation. This results in a GIS surface depicting changes in the intensity of environmental phenomena including climate data, air/water pollution, underground minerals and animal/human population distributions. This topic provides students with an opportunity to apply these techniques in a variety of application areas. Students are taught a variety of regression, simulation and population modelling techniques, together with a summary of their advantages and disadvantages. Students are also introduced to spatial statistics, density mapping and Dynamic Brownian Bridges (for modelling animal movement).

The aim of this topic is to provide students with a good understanding of spatially-explicit and temporal modelling approaches, together with associated spatial autocorrelation and non-stationarity issues, as they apply to human, ecological and environmental problems. Students are also provided with a good understanding of Geostatistcs, density surface creation and movement models.

On completion of this topic, students will be expected to be able to:

1. Describe a range of parametric and non-parametric regression modelling and data mining methods

2. Identify the level of spatial autocorrelation in data and describe solutions to managing spatial autocorrelation

3. Demonstrate two predictive spatial modelling techniques and test their predictive power using independent data

4. Describe and demonstrate the use of Geostatistics

5. Identify the Kernel Density Estimator and Dynamic Brownian Bridges methods, their application, utility and limitations

6. Successfully communicate the methods, results and discuss the above computer analysis.