ArcGIS Tutorial: geostatistical Analyst Application Example

Geostatistical Analyst meets the needs of many different applications. Here are a few examples of geostatistical Analyst applications.

　　Analysis of exploratory spatial data

Geostatistical Analyst uses measured sampling points in the study area to create accurate predictions for other non-measured locations in the same region. The exploratory spatial data analysis tools included in Geostatistical Analyst are used to evaluate statistical attributes of data, such as spatial data variability, spatial data dependencies, and global trends.

The following example uses several exploratory spatial data analysis tools to study the properties of the ozone measurements obtained at the monitoring stations in the Carpathian Mountains.

　　

　　Semi-variant function modeling

The geostatistical analysis of data is carried out in the following two phases:

The model of the Semivariogram or covariance function is established to analyze the surface properties.

Kriging

In Geostatistical Analyst, there are several kriging methods available for surface creation, including ordinary kriging, simple kriging, pan-kriging, indicator kriging, probabilistic kriging, and disjunctive kriging.

The two phases of geostatistical analysis of data are described below. First, use the semi-variance/Covariance wizard to fit the model of the US winter temperature data. Then use this model to create a temperature distribution map.

　　

　　Surface Prediction and Error modeling

Use geostatistical Analyst to generate various types of map layers, including prediction plots, sub-bitmaps, probability plots, and prediction standard error graphs.

A prediction map showing the level of radioactive cesium soil contamination in Belarus was generated using geostatistical Analyst after a leak at the Chernobyl nuclear power plant.

　　

　　Threshold mapping

A probability plot can be generated to predict where the value exceeds the critical threshold value.

In the following example, the locations shown in dark orange and red indicate a probability greater than 62.5%, where radioactive cesium contamination exceeds the maximum allowable level (critical threshold) in the forest berries.

　　

　　Model Validation and diagnostics

You can split the input data into two subsets. The first subset of available data can be used to develop models for predictions. Then use the Validate tool to compare the predicted values with the known values from the rest of the locations.

The following shows the evaluation of the model using the Validate wizard, which is developed to predict organic matter in the Illinois farm.

　　

　　Surface prediction using the Synergistic Kriging method

The Synergistic kriging (an advanced surface modeling method included in Geostatistical Analyst) can be used to improve the surface prediction of main variables by considering two-level variables (assuming that the main variable is related to the two-level variable space).

In the following example, exploratory spatial data analysis tools are used to explore spatial correlations between California State ozone (primary variable) and nitrogen dioxide (level two variables). Because variables are spatially related, co-kriging uses nitrogen dioxide data to improve predictions when mapping an ozone map.

　　

ArcGIS Tutorial: geostatistical Analyst Application Example