Data Distribution

While the data from the WSM project is an exhaustive effort of compilation and quality control (11), there are the usual myriad factors opposed to any such large scale effort. One of the most apparent and immediately obvious is the unequal distribution of the data around the globe. Through no fault of anyone, there still is not a lick of data as to the orientation of the stress field, or state of stress (SoS) in the middle of Greenland , between mid-ocean ridges, or across some political borders. This has implications for this method as I have not implemented a normalization method for this computation.

As a further exercise in exploring this methodology, it would be useful to bin data into crustal segments that vary as a function of the angular distance away from the map point being considered. This will present some challenge as the map will manifest itself as a constantly updating variable grid problem that will be difficult to calculate and nearly impossible to code an effective adjoint.

 

hist_0
Figure 2 While azimuths are fairly well distributed around the world, sampling locations are woefully Eurocentric.

From the below variogram of the entire data set, it is clear that this method of 'normalization' should be effective and could be faked by simply calculating several maps discriminating data inputs over several distance criteria and then correlating (or rather hoping for a correlation of) each of these.

The variogram corroborates the notion that at small distance scales the stress azimuths are roughly parallel and get more evenly distributed at larger offsets. Local stress pattern investigation such as that undertaken by (6) in Europe or (3) in India, could well be enhanced by this data manipulation.

 

var_0
Figure 3 Degree of non-parallelness over angular distance (converted to km's). 20,000 km is half of earth circumference