1. Can GPS horizontals provide useful information about surface loading? Case studies in California and Greenland. John Wahr (U of Colorado) Abbas Khan (DTU Space), Tonie van Dam (U of Luxembourg), Lin Liu (Stanford U), Michiel van den Broeke (U of Utrecht), Chuck Meertens (UNAVCO) The main issue addressed in this talk: Most published GPS-based studies of surface loading use only the GPS verticals. Can GPS horizontals also provide useful information?

2. GPS

3. Add loadGPS

4. Add loadGPS GPS receiver moves downward and towards the load Displacements caused by adding a load

5. remove loadGPS GPS receiver moves upward and away from the load Displacements caused by removing a load

6. Uplift caused by the removal of a 1-m thick, 20-km radius, disc load. Uplift sensitivity kernel falls off with distance, but has a long tail. Vertical displacements can help determine the total nearby mass loss (or gain).

7. Horizontal displacements can help determine the location of the mass loss. For example: If a load to the north is removed, the site goes up and moves southward. If a load to the south is removed, the site goes up and moves northward

8. Sensitivity to the removal of a 20-km disc load. Vertical displacements are 2.0-2.5 times larger than horizontal displacements

9. What happens to the vertical-to-horizontal ratio if there are loads at different azimuths?

10. One thinning glacier Two thinning glaciers 2.2 × Horizontal = Up 2.2 × Horizontal < Up The Up values add. The Horizontal vectors partially cancel. So: GPS

11. But, it could also work the other way around:

12. Thinning glacier 2.2 × Horizontal = Up 2.2 × Horizontal > Up The Up values cancel (one is positive, the other negative). The Horizontal vectors largely add. So: GPS Thinning glacier Thickening glacier

13. Rest of talk: apply these ideas to two examples of localized load. Lake Shasta, a reservoir in northern California Kulusk, a GPS site in southeast Greenland

14. Lake Shasta 20 km P349P341 P338 P060

19. P349 lies just south of the lake, and is the site closest to it. Daily station positions obtained from UNAVCO’s PBO website. SNARF 1.0 reference frame. P349

20. Data have been detrended. Black line = smoothed values. The horizontals are dominated by north- south displacements The lake is due north of this site.

21. Remove loading signal caused by regional soil moisture variability. Computed using water storage output from Goddard’s GLDAS/Noah land surface model (Rodell, et al 2004).

22. GPS minus GLDAS/Noah soil moisture load Horizontals are mostly in the north-south direction, which is consistent with lake loading.

23. Rotate the horizontals through the angle expected from lake loading. For lake loading, the expected ratio of vertical to horizontal at P349 is 4.1

26. Each site’s vertical displacements have been scaled by the expected (lake level)-to-(vertical) ratio for that site. Generally good agreement with the observed lake level from tide gauges. The agreement is best for P349, which is closest to the lake.

27. Greenland

28. GNET permanent GPS stations (from Mike Bevis, Ohio State) Kulusuk (KULU), 1996 GPS receiver at Kulusuk, on bedrock along the southeast Greenland coast All GPS analyses, below, were done by Abbas Khan, using GIPSY.

29. Helheim Glacier Catchment Kangerdlugssuaq glacier Other glaciers KULU

30. GRACE time series of surface mass, averaged over a disc of radius 250 km.

31. Mass loss began accelerating in 2003/04.

32. GRACE time series of surface mass, averaged over a disc of radius 250 km. Mass loss began accelerating in 2003/04. Deceleration in 2007/08.

33. KULU GPS data. IGS05 reference frame. Trends have not been removed.

34. Trends have been fit and removed over the entire time span. There’s a clear change in slope in late 2003/early 2004. GRACE and other results also suggest a 2004 acceleration in nearby mass loss.

35. Use only the pre-2004 data to determine the trend.

36. Remove the loading effects of snow accumulation (precip- minus–evap-minus-melt). Use RACMO surface mass balance (SMB) model (van den Broeke et al, 2011).

37. GPS minus SMBThe direction of the horizontal motion is south-southeast. Which suggests the post-2004 mass loss was located north- northwest of the site.

38. The mean horizontal direction is southward along this green line. Displacements along mean direction Displacements perpendicular to mean direction.

39. The direction away from Helheim Time series of the horizontal direction These angles are away from points lying east of Helheim The direction changed in 2007. So did the GRACE mass loss rate.

40. Helheim Glacier Kangerdlugssuaq glacier glaciers KULU This is consistent with the following sequence of events: (1)Helheim, and glaciers east of Helheim, underwent an acceleration in mass loss rate around 2004. (2)The Helheim rate decreased around 2007, but the rates of the other glaciers did not.

41. Assume mass loss occurs in two places. One is Helheim Glacier. The other lies somewhere along the green line. Use the GPS verticals and horizontals to determine a time series for the horizontal loading signal for Helheim and for the second glacier. Helheim KULU

42. Helheim flattens out between mid-2006 and mid-2008, but then starts to lose mass again. The second glaciers(s) keeps losing mass at a reasonably steady rate. The second glacier is unlikely to be Kangerdlugssuaq, because it’s so far from KULU. Its mass loss rate would have to be 4-5 times the Helheim mass loss rate to explain these numbers. Kangerdlugssuaq Helheim

43. Implications of the KULU results 1.Mass loss from Helheim Glacier accelerated around 2004, and then decelerated around 2007. 2.Mass loss rates from glaciers east of Helheim, remained relatively constant from 2004 onwards. 1.The actual locations of the other thinning glaciers (both their azimuth and distance from KULU) cannot be unambiguously determined using data from a single GPS site. 2.Multiple GPS sites would allow triangulation, to help uniquely determine which glaciers are losing mass, and by how much. 3.GNET will help.

44. GENERAL SUMMARY 1) Vertical displacements help determine the total nearby mass loss. Horizontal displacements help determine the location of the mass loss. The horizontal displacement caused by the removal of a localized load is away from the load. 2) The vertical displacement is 2.0-2.5 times the horizontal displacement. But the ratio is likely to be larger than this if the loading is spread over a range of azimuths. (Though it could also be smaller than this, if the load is positive on one side of the site and negative on the other.)

45. The End