A Road in Auvers After the Rain by Vincent Van Gogh
April 08, 2021
Goal: Make everyone feel more comfortable using spatial stats when analyzing field experimental data.
(you don’t have to be a geospatial statistics expert)Where to Find This Information
This Presentation:
https://github.com/IdahoAgStats/lattice-spatial-analysis-talk
A longer tutorial:
https://idahoagstats.github.io/guide-to-field-trial-spatial-analysis
What Are Barriers to Using Spatial Stats?
- Perceived lack of need
- Unsure of benefits
- No training in the topic/intimidated by the statistical methodology
- Limited time to devote to statistical analysis
- Unclear what would happen to blocking if spatial stats are used
- very few resources for easy implementation
Spatial Variation in Agricultural Fields
Univeristy of Idaho’s Parker Farm (Moscow, Idaho)
Spatial Variation in Agricultural Fields
Blocking in Agricultural Fields
Blocking versus Spatial Analysis
This is not how this works. Blocking is compatible with spatial analysis and recommended for most (all?) field trials.
There Are Many Spatial Methods Available
| areal data | correlated error models |
|---|---|
| row and column trend | exponential |
| nearest neighbor | spherical |
| separable ARxAR models | Gaussian |
| spatial error model | Matern |
| spatial lag model | Cauchy |
| ARIMA | power |
| splines | linear |
| GAMs | many more… |