RookCase: Explore Spatial Autocorrelation with a Microsoft Excel Add-on
|January 16, 2012||Posted by karmadsen under blog, Geostatistics, GIS, Groundwater equations, Groundwater Modeling, Groundwater Modeling Software, Modeling Software, Visualization|
Anyone dealing with environmental data must take great care to test for spatial autocorrelation. Many statistical tests, include regression analysis, are based on an assumption that errors are not autocorrelated. Errors should be random because they hypothetically represent “noise” around some real mathematical relationship. If the errors are autocorrelated with each other, you might not really understand the system.
Things get even more complicated in environmental science where models are being constructed in 3-dimensional space. Before you use ordinary least squares on your environmental model, you better consider spatial autocorrelation.
In the 1990s, Professor Michael C. Sawada of the University of Ottawa created RookCase, a Microsoft Excel Add-On that calculates spatial autocorrelation for data-sets in Euclidean geometry (i.e. not latitude and longitude). The VBA program will calculated two statistics of spatial autocorrelation, Moran’s I and Geary’s C. Both were invented in the 1950s, but they are still widely used.
The RookCase add-on can be downloaded at the University of Ottawa’s Laboratory for Paleoclimatology and Climatology.
Sawada, M. 1999. ROOKCASE: An Excel 97/2000 Visual Basic (VB) Add-in for Exploring Global and Local Spatial Autocorrelation. Bulletin of the Ecological Society of America, 80(4):231-234.