From Ken Wed Jan 26 11:03:24 1994 From: Ken (Ken) Date: Wed, 26 Jan 1994 11:03:24 PST Subject: [IGSMAIL-0466] Northridge Earthquake Message-ID: ****************************************************************************** IGS Electronic Mail Wed Jan 26 11:03:24 PST 1994 Message Number 0466 ****************************************************************************** Author: Ken Hudnut and Mark Murray Subject: Northridge Earthquake January 25, 1994 To: Distribution on northridge at andreas.wr.usgs.gov and GPS colleagues by e-mail >From: Ken Hudnut Mark Murray U. S. Geological Survey U. S. Geological Survey 525 South Wilson Ave. 345 Middlefield Rd. MS #977 Pasadena, CA 91106 Menlo Park, CA 94025 (818)405-7232 (415)329-4850 FAX: 405-7827 FAX: 329-5163 hudnut at seismo.gps.caltech.edu mmurray at eratos.wr.usgs.gov ------------------------------------------------------------------------------------------------------- The GPS response to the Northridge earthquake has basically involved several components; 1) The PGGA continuously operating line between Palos Verdes (PVEP) and JPL (JPL1) - Y. Bock 2) The JPL pre- and post-earthquake response, headed by Andrea Donnellan 3) The USGS & SCEC pre- and post-earthquake response, headed by myself 4) Post-seismic studies that have been initiated, including; JPL efforts by Ken Hurst and Frank Webb SCEC efforts (UCLA and others) in cooperation with USGS on non-"Profile Array" points USGS efforts on establishing a N/S profile array ~60 km long, crossing the fault This is a first effort to distribute the initial results of a large group of people. See below for participants. The GPS field work and processing efforts of the first week have now been processed preliminarily, and both Mark Murray and I here present a couple of models for the slip in the event, based on different modeling approaches. If you have questions, please let us know. Also, we are going to be posting plots of the results onto the andreas.wr.usgs.gov anonymous ftp site in /pub/northridge directory. The files we post will be of the form "gps*.ps" for postscript files or "gps*.eps" for macintosh files. I know many of you will be interested in hearing the scientific results first, so here goes. Those in the GPS community will find additional notes on our plans for data availability below as well. PRELIMINARY DISPLACEMENT RESULTS: Summary of Preliminary GPS Pre- and Post-Earthquake GPS Solutions: ****************************************************************** NAME EDAT NDAT VDAT LATIT LONGIT ESIG NSIG VSIG Result **** **** **** **** ******* ********* *** *** *** ******** PVER 0.09 1.04 0.50 33.7438 -118.4035 0.4 0.3 0.6 YB/PGGA JPLM 0.65 -0.51 -0.20 34.2051 -118.1733 0.4 0.3 0.6 YB/PGGA CALA 2.10 8.48 -0.02 34.1400 -118.6457 1.0 1.0 2.0 MM/USGS PACO 7.98 -0.58 6.72 34.2635 -118.4083 1.0 1.0 2.0 MM/USGS GLEN 1.38 -0.59 4.92 34.1611 -118.2825 1.0 1.0 2.0 MM/USGS MULH 2.03 6.77 3.37 34.1300 -118.5600 1.0 1.0 2.0 MM/USGS NORT 2.79 -7.11 47.26 34.2167 -118.5547 1.0 1.0 2.0 MM/USGS RESE 18.36 0.07 21.46 34.2917 -118.4883 1.0 1.0 2.0 MM/USGS PICO -13.53 16.25 43.12 34.3295 -118.6007 1.0 1.0 2.0 MM/USGS SAFR -11.78 16.50 44.14 34.3295 -118.6007 1.0 1.0 2.0 MM/USGS 2131 -3.43 -1.54 2.08 34.3510 -118.4250 1.0 1.0 2.0 MM/USGS LOVE -10.80 -3.90 -9.00 34.4833 -118.6667 1.0 1.0 2.0 AD/JPL Notes: ****** In the above table is shown the summary of preliminary GPS displacement results that I now know of. The references for these results are given below This table has only rough estimates of the approximate repeatibility for errors. Data and errors are in centimeters. YB/PGGA = Yehuda Bock, results from GLOBK analysis of PGGA solutions, already distributed on IGS mail by that group. MM/USGS = Mark Murray, results from GLOBK analysis of solutions based on K. Hudnut and Y. Feng pre-earthquake solutions for the Inter-County '93 data and USGS & SCEC data collected since the earthquake AD/JPL = Andrea Donnellan, results from GIPSY/OASIS analysis (also, this group obtained similar displacement estimate to ours at Oat Mtn., aka PICO & SAFR) Results for stations SAFE and LOVE already distributed publicly by Andrea. ------------------------------------------------------------------------------------------------------- PRELIMINARY MODELING STUDIES: Mark Murray's Preliminary Model Solution for Rupture Plane: *********************************************************** I estimated a uniform-slip rectangular-fault dislocation model from the observed surface displacements given in the Section A table. I did not include LOVE and assumed JPLM was fixed. I used a random-cost optimization technique to search a broad range of model parameters to find a best-fitting fault geometry: Dip Depth(km) Width(km) Az Length(km) Lat Lon 39.22 7.77 12.19 109.13 7.11 34.40944 -118.56083 Slip(m) Rake Moment (N-m) 4.5213 89.1674 1.176E+19 where depth is to the top edge, azimuth is clockwise from north, and the origin is the projection of the west edge of the fault to the surface. The nrms of the residuals is about 2, with horizontal residuals generally < 2 cm, and vertical residuals generally < 5 cm. This model shows good agreement with seismological estimates in orientation, dip and moment. The fault extends from 7.8 to 15.5 km depth, and lies close to the main shock hypocenter. The small (7x12 km) fault plane suggests that the slip was concentrated at depth. The location of the plane corresponds to a region with relatively few aftershocks. A figure showing the observed and model-estimated horizontal displacements and the fault plane location with respect to the aftershocks is in /pub/northridge/gps_murray_model1.ps Ken Hudnut's Preliminary Model Incorporating Variable Slip: ********************************************************** Used data and errors as shown in above table as input (GINV code developed by Shawn Larsen, LLNL) Assumed fault geometry based partly on Mark's result above and on aftershock plane: TYPE SLIP LENGTH TOP WIDTH DIP LONGITUDE LATITUDE AZIMUTH FAULT NX NZ ---- ---- ------ --- ----- --- --------- -------- ------- ----- -- -- t -33.0 30.00 0.00 30.00 40.0 34.3917 -118.5583 160.0 NT_T 3 3 r 0.0 30.00 0.00 30.00 40.0 34.3917 -118.5583 160.0 NT_S 3 3 (UpDip Ctr of Flt) (Azim. c-cl from east) NRMS Residual = 3.13 : abs. val. (Obs. - Model)/Sigma (for all components & all data modeled) : some (Obs.-Model) residuals are up to ~22 cm in vertical and ~11 cm horiz. Total Moment = 1.88 x 10^26 dyne-cm (assumes 3 x 10^11 for rigidity) top NW_____________(surface)____________SE | | | | Information on the diagram and model: dip>| 14 | -23 | 10 | | | | | Projection viewed from a point above the strike>| 4 | 53 | 1 | fault looking down orthogonal to the plane. | | ....|... | The fault plane is 30 x 30 km, based on |__________|______.___|__._______| the aftershock distribution. | | . | . | dip>| -108 | -142 . | . -60 | Each sub-fault is 10 x 10 km and both the | | . | . | dip-slip and strike-slip component of slip strike>| -2 | -11 . | . -28 | is solved for in the inversion. These slip | | . | . | estimates are given in each sub-fault box. |__________|______.___|__._______| Thrust and left-lateral are negative, and | | ....|... | normal and right-lateral are positive. dip>| -94 | -117 | -61 | Dip-slip # and strike-slip # indicated at | | | | left side of subfaults. strike>| -9 | -26 | -29 | | | | @<----------- Approximate location of mainshock hypocenter. |__________|__bottom__|__________| (depth of bottom of model plane @19.3 km) The dotted box in this diagram roughly represents the location of Mark's model slip patch where he obtains ~450 cm of slip, for comparison. The model residuals are still substantial (should be ~1.0), and moment seems too high, but remember these results are preliminary. Also, remember that these data will contain the slip contribution of the aftershocks that were occurring as the data were collected. We will have additional GPS displacements for an additional 20 GPS stations spanning the area affected by displacements within the next week or so, probably. These additional results will help greatly in constraining the model. I am solving for the dip-slip and strike-slip component on each subfault, employing smoothing and truncating at 5 singular values. I am not constraining slip sense or amount in the inversion parameterization. So, there are 18 parameters solved for and 12 GPS displacements, each comprising 3 components for 36 'data'. In this model, the fault plane is meant to be approximately co-planar with Mark Murray's initial result, but solving for slip on the entire plane that has been defined by the aftershock relocations of Jim Mori and Egill Hauksson and others. ******************************************************************************************************** The following people have been working on this effort to my knowledge right now varying from field work to data processing and logistics. USGS NEHRP funds, the NSF via the Southern Calif. Eq. Center (SCEC), NASA and others support these groups SIO/UCSD Y. Bock, E. Calais, H. Johnson, P. Fang, K. Stark UCLA X. Ge, Z-K. Shen, D. Potter, Y. Feng Caltech M. Cline, B. Zajac, Louie, Mark MIT B. Hager, T. Herring, Bob King, and 6? students USGS K. Hudnut, M. Murray, J. Savage, K. Gross, J. Sutton, C. Stiffler, G. Hamilton, K. Wendt, several others as well JPL A. Donnellan, K. Hurst, F. Webb, G. Franklin, others LLNL S. Larsen ******************************************************************************************************** DATA AVAILABILITY: Starting very soon, we will have data access available by ftp. Our method of doing so is that we will establish two main data repositories. First, all of the immediate response data for the coseismic will be placed on ftp at the SCEC facility at UCLA, including our first observations of the USGS "Profile Array." We will place RINEX files into that location. This should be set up very soon, but is not yet ready. In the meantime until this is set up, we can make data available but it is more work for us. The second site will be on line at Menlo Park, and will be our main repository for the "Profile Array" data set. For our convenience, we prefer to maintain that data set separate from the SCEC archive at UCLA. Again, however, people may call us to discuss their interests in the data and ask for that password. ********************************************************************************************************