Iceland Dike Model

USING ARCGIS PRO TO MODEL THE DIKE UNDERLYING THE VOLCANIC EARTHQUAKE SWARMS NEAR GRINDAVIK, ICELAND

For this project, I utilized various geospatial analysis tools to approximate a model of the Icelandic magma intrusion that has caused recent seismic activity near Grindavik, Iceland.

Data was obtained from the Interactive Earthquake Explorer (which is sourced from USGS) on November 15th, 2023, so there is more recent data that is not included in my model. I used this data to interpolate data across the Reykjavic peninsula using the Empirical Bayesian Kriging 3D (EBK 3D) tool.

Empirical Bayesian Kriging 3D (EBK 3D)

The EBK 3D tool has certain data requirements that must be met in order to produce a useable interpolation:

It must have x, y, and z data, as well as a value to be interpolated, and,
The data must be in a projected coordinate system.

The earthquake data is given in decimal degrees, with a depth that I used for the z-value, and I interpolated magnitude based on the relationship between dike (intrusion) size and magnitude of volcanic earthquake swarms. I also projected the data from its original (geographic) coordinate system of WGS 1984 to the WGS 1984 web mercator (auxiliary sphere).

The tool also works best if the data is normally distributed, so I performed some analysis of my data before running the tool.

Histogram of earthquake data, showing normal distribution of count of earthquakes and magnitude. The locations that experience the most earthquakes usually have earthquakes of between 4.5 to 4.6 magnitude.

QQ plot of magnitude compared to a normal distribution. This shows that this data will be good to work with for the prediction of values between points (interpolation).

After my analysis of the data, I ran the EBK 3D tool with the Power semivariogram model, which weights measured data closest to an interpolated point by a power and decreases the weight by a power the further from that point any measured data is.

QQ Plot (above) for interpolated data shows the interpolated surface (shown on right) fits the trend line of a normal distribution fairly well.

Summary of statistics (above) for interpolated model. These give data on how much confidence you can have in the predictive model. Both the Mean and Root-Mean-Square values should be as close to zero as possible; and the Root-Mean-Square Standardized should be close to one. My data fits this model well.

Predictive model created using Empirical Bayesian Kriging 3D. Note the North-West trending orange polygon in the center of the map.

With the EBK 3D tool complete, I converted the data into a NetCDF file, which can be used to create a 3D visualization using "voxels."

Visualizing Three Dimensional Data with Voxels

A voxel is a three-dimensional pixel, representing a value or attribute in a volumetric space. It is commonly used in computer graphics and scientific visualization to represent data in a three-dimensional grid. Each voxel contains information about a specific point in space, such as color, density, or material properties.

In the 3D environment hosted by ArcGIS Pro (a "local scene"), I imported the NetCDF file I had created as a multi-dimensional voxel layer. By adjusting the data shown and how it was displayed, I was able to create a three-dimensional approximation of the magmatic intrusion under Grindavik, Iceland.

I haven't been able to locate any examples of other (more authoritative) models of the dike, although they have been referenced in several articles I've read. The north-south trend fits the plate boundary in the region, and the analysis of the data prediction shows that the model fits the data.

Page updated

Report abuse