Hi,

I have a kriged GA layer of precipitation data from approximately 50 points (i.e. stations) and would like to assess the skill of the interpolated surface.

I would like to omit some of the stations prior to kriging, and calculate the mean absolute error between interpolated and actual data.Then repeat many times to give a distribution of errors.

Could anyone suggest an efficient method of carrying out the above task using the Geostatistical Wizard? For example, how can I remove some point data from the kriging without changing the source of the data?

Also, is the 'average standard error' equivalent to the 'mean absolute error'?

Thank you for your help, it is greatly appreciated.

I have a kriged GA layer of precipitation data from approximately 50 points (i.e. stations) and would like to assess the skill of the interpolated surface.

I would like to omit some of the stations prior to kriging, and calculate the mean absolute error between interpolated and actual data.Then repeat many times to give a distribution of errors.

Could anyone suggest an efficient method of carrying out the above task using the Geostatistical Wizard? For example, how can I remove some point data from the kriging without changing the source of the data?

Also, is the 'average standard error' equivalent to the 'mean absolute error'?

Thank you for your help, it is greatly appreciated.

Here's how you do it:

1. Use Subset Features to randomly partition your data into a training set and a testing set (in the tool, you specify how many points will in each subset).

2. Perform kriging in the Geostatistical Wizard on the training features.

3. Use the kriging surface as input to GA Layer to Points. Predict to the testing features, and specify the field to validate on (the filed you used to interpolate). This will create a new point feature class with validation statistics.

4. To calculate the mean absolute error, you'll use the "Error" field in the new feature class. Make a new field and calculate the absolute value of the error. Then take the average of these absolute values. To calculate the average standard error, take the average of the "Standard Error" field.