Timeline for Minimum number of samples for kriging interpolation
Current License: CC BY-SA 3.0
7 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Feb 12, 2013 at 10:02 | comment | added | jareks | The rule of thumb that I've been tought: 30 points for omnidirectional Kriging and 100 for bidirectional. | |
| Feb 12, 2013 at 9:25 | vote | accept | Marcin | ||
| Feb 12, 2013 at 3:35 | history | edited | Fezter | CC BY-SA 3.0 |
added amazon links to references
|
| Feb 11, 2013 at 20:59 | comment | added | Mike T | I'm convinced that anyone using kriging should need either a good geostatistics course or have a solid GIS/statistics background. Leaning how to properly model a semivariogram requires some skill. | |
| Feb 11, 2013 at 19:58 | comment | added | whuber | @Jeffrey That's partly the case. The amount of data bears on two things: the kriging prediction variance (which varies across the spatial domain) and the accuracy with which the variogram itself can be estimated. The latter is often overlooked, especially in traditional treatments of kriging: it's kind of an elephant in the room. If you know the correct variogram and it has a small nugget/sill ratio and large range relative to the extent of the spatial domain, you can krige with remarkably few data, especially if they adequately sample the full range of data values. | |
| Feb 11, 2013 at 18:35 | comment | added | Jeffrey Evans | Great answer @whuber. However, doesn't the minimum number of points also partially depend on the extent of the spatial domain and the desired grain of the prediction? If you distill it down to a sampling issue then it becomes a question of capturing the population and spatial variation in the sample. | |
| Feb 11, 2013 at 17:33 | history | answered | whuber | CC BY-SA 3.0 |