Spatial Interpolation Comparison (SIC) exercises
In Spatial Interpolation Comparison (SIC) exercises, participants are invited to estimate values of a variable at given locations by using information provided by a subset of the total number of measurements of the variable (see Figure 1). The algorithms of the participants are then contrasted with each other by comparing the estimation errors of each spatial interpolation technique. Participants are further invited to present their techniques in a manuscript that is reviewed by an editorial board prior to the publication of the exercise.
Figure 1. Spatial Interpolation Comparison (SIC) exercise: participants are invited to minimize errors when using n observations (left) to estimate values located at N locations (right).
Submitted manuscripts are expected to describe the technique used and their results. Among other information, the following information is usually requested:
- Mean Square Errors
- Mean Absolute Errors
- Statistics of the estimated values: Minimum, maximum, mean, median and standard deviation,
- Statistics of the errors: Minimum, maximum, mean, median and standard deviation.
- A map with the estimated values
- A map with the levels of uncertainty
Calls for participations to SIC exercises are launched on the internet, mainly via related mailing lists and, obviously, on AI-GEOSTATS. The scope of the exercises is changing each time.