1. Trivial applications of NeoKinema which illustrate its algorithm by Peter Bird UCLA 2002/2009/2015 Support from NSF & USGS are gratefully acknowledged.

2. Many grids used for testing overlap the pole and the international date line. Thus, they demonstrate nicely that there are no singularities, and that the spherical algebra was done correctly.

3. TEST01: No input data, zero velocity at two nodes → static solution.

4. TEST02: No input data, one fixed node, and one boundary node that rotates around it. Result is rigid-plate rotation, at rates on the order of 7×10 -16 {radian}/s.

5. Strain-rates in early versions of NK: Maximum strain-rate of 2  10 -20 /s = 0.3% in 4 Ga.

6. Now, using NKv4.1: Maximum strain-rate becomes 8  10 -23 /s [7 orders smaller than the rotation rate] = 0.001% in 4 Ga.

7. TEST03: No input data, uniform extension driven by boundary conditions.

8. Note: In absence of data, lithosphere behaves as a uniform viscous sheet. Therefore, in uniform stress field far from BCs, it undergoes equal vertical and horizontal shortening.

9. TEST04: Data on the azimuth of the most-compressive horizontal principal stress are given:

10. and NeoKinema interpolates these directions by the algorithm of Bird & Li [1996]: N.B. This interpolation was done with the independent-data variant of the method of Bird & Li [1996]. The alternate clustered-data method would infer larger uncertainties in the results.

11. When the same velocity boundary conditions are given (as in Test03), NeoKinema attempts to find a velocity solution that will honor these interpolated directions:

12. The new solution has nonuniform strain-rates (large where stress directions are compatible with the velocity BCs; small elsewhere):

13. and here is the (mis)match between the principal strain-rate azimuths of the solution and the target azimuths derived from the stress data: (Note that many of the red target azimuths are hidden by the yellow actual azimuths.)

14. TEST05: Three faults of unknown slip-rate make up a plate boundary system. Each dip-slip fault has assumed rake of 90°. (Same velocity BCs as in Test03; however, no stress- direction data as in Test04.)

15. NeoKinema finds a solution with most of the deformation assigned to slip on the faults:

16. However, the requirement of purely dip-slip faulting on the two normal faults requires some significant continuum strain-rates:

17. TEST06: The same 3 faults of unknown slip-rate make up a plate boundary system. But now, each dip-slip fault has assumed rake of 90°  20° (  ). (Same velocity BCs as in Tests03~05.)

18. NeoKinema finds a solution with virtually all of the deformation assigned to slip on the faults, and continuum strain-rates are smaller: Note oblique slip on these faults.

19. TEST08: A strike-slip fault whose trace is a small circle is entered with unknown slip rate. (Note that it is not necessary to outline fault zones with slender elements, although one may choose to do so.)

20. When the solution is driven indirectly by velocity at one node, the solution is Eulerian plate tectonics, with minimal strain-rates:

21. TEST09: Example of a convenience feature, the type-4 boundary condition, which allows boundary nodes to be assigned to a major plate by simply giving its abbreviation (e.g., “ NA ”, “ PA ”); the necessary velocity is calculated within NeoKinema by the Euler formula. (Note: Lacking any data, such as fault locations, the program finds a uniform- viscous-sheet solution to this problem.) PA NA

22. TEST10: Synthetic GPS velocities, which are consistent with uniform plate rotation, are input at many internal points. Model boundaries are free, except at 2 boundary nodes which are fixed:

23. TEST11: Same as Test10, except that now the velocity reference frame of the GPS data is treated as unknown, or free-floating. The result is that the velocity reference frame is determined by the 2 fixed boundary nodes, and all motion is reduced to less than 0.0004 mm/a.

24. TEST13: Conversion of input short-term interseismic GPS velocities (left) to long-term corrected velocities (right), along a strike-slip fault which is temporarily locked down to 100 km depth. (In this test, velocity BCs “enforce” the right plate-motion solution.) INPUT: OUTPUT:

25. TEST14: Conversion of input short-term interseismic GPS velocities (left) to long-term corrected velocities (right), along a strike-slip fault which is temporarily locked down to 100 km depth. (In this test, the southwestern plate is free, and GPS data determines its velocity.) INPUT: OUTPUT:

26. TEST15: Conversion of input short-term GPS velocities (left) to long-term corrected velocities (right), along the Cascadia subduction thrust, which is temporarily locked from 10 km to 40 km depths. (Note that long-term relative velocities within NA are less than short-term.) INPUT: OUTPUT:

27. (end)