Assessing the Quality of Spatial Prediction

The accuracy of spatial interpolation or predictive models is critical as it determines the quality of the interpolated values. It is less easy to develop spatial prediction models than accuracy assessment of spatial prediction which, however, remains unknown in general.

Diagnostic measures of quality of spatial prediction are: mean error (ME), mean absolute error (MAE), root mean square error (RMSE) and Mean Squared Deviation Ratio (MSDR) of residuals with kriging variance can be calculated as:

In this Lesson, we will evaluate quality of ordinary kriging prediction using following three methods:

• Cross-validation

• Validation with an Independent Data set

• Conditional Simulation for Spatial Uncertainty