Working with large datasets in Geostatistical Analyst
In general, the interpolation methods of Geostatistical Analyst can process large amounts of input data. However, some of the methods—for example, kriging and radial basis functions—may fail to complete, or the process may take a very long time to complete.
The reading of data, especially large datasets, often accounts for most of the processing time. To ensure optimal performance when using Geostatistical Analyst, the input data is rewritten to scratch files, using an indexing system. These scratch files are utilized when the same data is used with different interpolation techniques or when different interpolation parameters are used with the same input data.
The scratch files are written to the operating system's temporary file location, and the file names are usually preceded by ga. These files usually have a file extension of .ds01, .ds02, or .ds03. The location of the files can be changed by creating an operating system environment variable called GA_TMP and pointing it to a folder.