Samik Sil (1) and Jeffrey T. Freymueller(2) Poroelastic and Other Deformations Due To The November 2002 Denali Earthquake G21B-1283 Samik Sil (1) and Jeffrey T. Freymueller(2) University of Alaska, Fairbanks, Geophysical Institute, 903 Koyukuk Drive, Fairbanks, AK-99775 (1) ftss@uaf.edu (2) jeff@giseis.alaska.edu 3. Test of the poroelastic model: We found that with the relaxation time constants constrained by other information, the estimated afterslip and viscoelastic displacements were very similar whether the poroelastic component was removed or not. This allows us to test the poroelastic model. If we remove the afterslip and viscoelastic components, estimated without using the poroelastic model, from the observed GPS time series, the residuals are well explained by the calculated poroelastic deformation pattern. Introduction: Coseismic water level changes due to the 2002 Mw 7.9 Denali earthquake were observed in Alaska. These water level changes show a linear relationship either with seismically induced volumetric strain or with a combined effect of volumetric strain and ground shaking. Water level recovery after the earthquake has been modeled by least squares technique, using an error function and a maximum decay time of one week. GPS observations also show very rapid deformation in the vicinity of the rupture zone during the first week following the earthquake. These two observations lead us to evaluate the importance of poroelastic deformation immediately after the earthquake. Our search for poroelastic deformation signals from the GPS time series, lead us to find other postseismic deformation signals too. Results: Estimated postseismic components: The estimated coefficients A and B determine the magnitude of the afterslip and viscoelastic components, respectively. In general, the afterslip component is largest for sites located 10-30 km from the fault, and for sites far from the fault the viscoelastic component dominates. 2. Estimated afterslip model: Using the estimated afterslip displacements over a 2-year period as pseudo observations, we estimated an afterslip model on the fault plane allowing slip on the coseismic rupture plane and its deeper extension. Aftersilp and Viscoelastic Deformations: We estimate the contribution of afterslip and viscoelastic relaxation to the total postseismic deformation by fitting appropriate time-dependent relaxation functions to the observed time series. We then try to explain the spatial pattern of afterslip. We have assumed that time and space components of afterslip are separable, that is, that we can describe the distribution of afterslip as S(x,t) = X(x)T(t) where the time-dependence has a simple form. We make a similar assumption about the viscoelastic component. We first fit the residual time series with the combination of two relaxation functions: A*log(1+t/ґ) +B*(1-exp(-t/ґ1) +C. (1) Where ґ is the afterslip decay time and ґ1 is the viscoelastic relaxation decay time and C is the total deformations at the date of installation of the GPS site. For fitting the timeseries we used the value of ґ1 from other studies ( Politz et. al, 2004, Freed et al., 2005) and we determined ґ from the aftershock decay rate (Perfettini et al, 2004). We also performed a separate grid search to determine the optimal values of ґ and ґ1 based on the time series alone. The best fitting values of ґ are very similar based on the time series and the aftershock decay rate. The value of ґ1 is weakly constrained based on the time series alone, which currently provide only a lower limit on ґ1. We adopted the values of 0.07 years and 2.5 years based on the aftershock decay rate and the viscoelastic postseismic models respectively. Residuals (observed timeseries- equation 1) and poroelastic deformation model for the sites MEN and ATT. These two campaign sites started a few days after the earthquake, and captured the poroelastic deformation pattern. Cumulative number of events with M>=3 (AEIC catalog) following the 2002 Denali earthquake. Afterslip decay time is determined using the method described by Perfettini et al., 2001 Poroelastic Deformation: We estimated the magnitude of poroelastic deformation of the 2002 Denali earthquake at each GPS station using a coseismic slip model (Hreinsdóttir et al., 2005) and a halfspace dislocation model (Okada, 1992). We assumed 0.25 and 0.22 for the undrained and drained Poisson’s ratio, respectively. Predicted horizontal poroelastic deformation is small. The maximum magnitude of horizontal deformation was predicted at the site MEN (North component 22 mm). We estimated the characteristic decay time of the poroelastic deformation from the postseismic decay of groundwater level in wells around Alaska (7 days). The low magnitude and short decay time of poroelastic deformation makes it difficult to identify this component directly from the GPS timeseries, except in near-fault sites that have time series starting immediately after the earthquake. The other components of postseismic deformation dominate the GPS time series. To explain the whole time series, we removed the predicted poroelastic deformation for each station, and tried to model the residual with a combination of viscoelastic and afterslip postseismic deformation. Conclusions: Multiple postseismic mechanisms are active in Alaska after the earthquake (Freed et al., 2005). Poroelastic deformation has a small magnitude and rapid decay in time. We can fit the observed GPS time series well using a combination of a a logarithmic decay function and an exponential decay function. For a few sites, the poroelastic contribution is important to fit the data in the first few weeks after the earthquake. The viscoelastic relaxation time is not well constrained because the time series are still too short. With time and more data, the viscoelastic model can be improved. The improvement may also improve the afterslip model. The present afterslip model predicts the maximum slip at the depth of 40 to 50 km. Contour plot of misfit for the afterslip and viscoelastic decay time based on fitting the timeseries data. The rectangle represents the decay times from independent studies (see text). The plus sign represent the values we adopted for this study. GPS Timeseries (with models): Top: Far from the fault sites, Middle: Near from the fault sites, Bottom: Typical Campaign sites from both sides of the fault. Left: Predicted horizontal deformation at the studied GPS sites using the described method. Top: Water level modelled using a combination of linear function, offset and erf(t/ґ), to determine characteristic decay time of poroelastic deformation Estimated afterslip model. Unit of slip is mm. A patch of maximum slip is obtained at 45-54 km depth. A 60 km depth model does not change the misfit very much. References: Freed, A. M., R. Bürgmann, E. Calais, J. Freymueller, and S. Hreinsdóttir (in press), Implications of Deformation Following the 2002 Denali, Alaska Earthquake for Postseismic Relaxation Processes and Lithospheric Rheology, J. Geophys. Res., doi:10.1029/2005JB003894. Okada, y., Internal deformation due to shear and tensile faults in a half-space, Bull. Seismol. Soc. Am., 82, 1992. Perfettini, H., and J., -P. Avouac, Stress transfer and strain rate variations during seismic cycle, J. Geophys. Res., doi:10.1029/2003JB002971. Pollitz, F., Transient rheology of the upper mantle beneath central Alaska identified from the crustal velocity field following the 2002 Denali earthquake (in press), J. Geophys. Res.,. Hreinsdottir, S., J. T. Freymueller, R. Bürgmann, and J. Mitchell, Coseismic Deformation of the 2002 Denali Fault Earthquake: Insights from GPS Measurements (in press), J. Geophys. Res.,doi:10.1029/2005JB003676. Pre-Earthquake Velocities: Estimated Afterslip Components: 2 years afterslip deformation map derived from the modelling of the GPS timeseries. The above slip model is obtained from these pseudo observation. Residuals based on the slip model are also ploted. Stn. Name Ve(m/yr) Vn(m/yr) SigVe(m) SigVn(m) TAZL -0.0145 -0.0117 0.0006 0.0007 GNAA -0.0152 -0.012 0.0002 0.0003 CLGO -0.0095 -0.0214 0.0001 FAIR -0.0091 -0.021 CENA -0.0097 -0.0196 MEN -0.0136 -0.0156 0.0075 0.0085 MENT PAXC -0.0134 -0.0182 0.0009 FCRK -0.0141 -0.0183 ATT -0.0119 -0.0189 0.001 0.0012 DNLC -0.0122 -0.0199 0.0008 BSB4 -0.02 DH97 -0.0166 -0.0194 M110 -0.0178 -0.0212 0.0019 0.0022 SSWB -0.0125 -0.0205 HIWC -0.0216 PANA -0.0127 -0.0207 Stn. Name Ve mm Vn mm Sig Ve (mm) Sig Vn (mm) TAZL -0.26057 9.461047 8.966 4.58217 GNAA -11.594 10.5458 7.2756 4.84634 CLGO 8.065 -0.27229 4.32429 3.370994 FAIR 21.7119 -1.9201 4.529066 3.1365 CENA 6.756502 -6.33888 5.9923 5.036 MEN -18.67123 1.39E+02 5.085189 4.697 MENT -42.2981 1.12E+02 4.5508 3.87606 PAXC -1.10E+02 41.4596 3.51722 2.99049 FCRK -1.04E+02 83.1934 3.75401 3.579 ATT 1.29E+02 -77.2045 6.41298 4.687694 DNLC 54.2648 -33.6953 4.30457 3.79126 BSB4 63.32723 -56.044 3.58092 2.42557 DH97 -51.83872 -14.5782 2.3527 1.90943 M110 -18.0784 -65.1708 4.913491 3.3297 SSWB -4.1231 11.8236 2.52802 2.405 HIWC -15.43725 13.7219 4.0258 3.1476 PANA 33.37414 -17.7527 6.4818 3.6772 Acknowledgements: We thank Dr. Roland Burgmann and Dr. Sigrun Hreinsdóttir for allowing us to use their codes and for their valuable advices.