The following statistics describe the accuracy of the geostatistical layer used to create the final grid:
Root mean square error: 19 991 mg/L
Mean error: -347.99 mg/L
Mean standardized error: -0.02375
Root mean square standardized: 0.9352
Average standard error: 20903 mg/L
The root mean square error of the final smoothed grid is 12 828 mg/l
See AGS DIG 2022-0028 (data points) for details on the horizontal accuracy of the input total dissolved solids point data, which is affected by the locational accuracy of the wells.
The grid was smoothed using focal statistics which calculates for each input cell location a statistic of the values within a specified neighbourhood around it, in this case a 3-cell radius. Smoothing the surface used the average cell value.
Selection criteria and the final data set used for interpolation of this grid are documented in the metadata for AGS DIG 2022-0028.
Step 1 (Modelling the surface): An empirical Bayesian kriging algorithm was used in the ArcMap Geostatistical Analyst extension to interpolate the total dissolved solids values and create a grid of the total dissolved solids distribution of the Leduc HSU.
Method: Kriging
Type: Empirical Bayesian
Subset size: 100
Overlap factor: 1
Number of simulations: 100
Output surface type: Prediction
Transformation: None
Semivariogram model type: Power
Neighbourhood type: Standard circular
Maximum neighbours: 8
Minimum neighbours: 8
Sector type: 8 sectors
Angle: 0
Radius: 5000
Step 2 (Grid Smoothing): The grid was smoothed using focal statistics which calculates for each input cell location a statistic of the values within a specified neighbourhood around it, in this case a 3-cell radius, and then smooths the surface using the average cell value.
Step 3 (Grid Alterations): The final grid was clipped based on the extent of the Leduc HSU, and spatial distribution of data.