Effects of the past- and present-day ice melting on observed uplift rates in Southeast Alaska

A joint Japanese-US observation project called ISEA (International geodetic project in South-Eastern Alaska) was initiated in 2005 to follow up the work of the University of Alaska Fairbanks (UAF) by adding new geodetic data sets (Miura et al., 2007). Based on 91 improved GPS velocities and the proposed ice loss models by Larsen et al. (2004, 2005, 2007) and by Tushingham and Peltier (1991), we estimated three effects: present-day ice melting (PDIM), post-little ice age (LIA) melting, and deglaciation following the last glacier maximum (LGM). We also have reevaluated the thickness of lithosphere and the viscosities of the mantle including that of asthenosphere in the region. The effects of LGM deglaciation were computed based on the ICE3G model (Tushingham and Peltier, 1991). Although the effect of LGM ice is small in SE-AK (i.e. 2 mm/yr at the maximum), we have confirmed that, by comparing with the misfits obtained by correcting for only two effects of PDIM and LIA, correcting for the LGM effect can improve the chi^2 of misfits by about 0.5 %. We estimated the effect of post-LIA ice melting based on the melting history proposed by Larsen et al. (2004, 2005). For the viscoelastic structure, we have tested two different kinds of mantle viscosity structure: a two-layer model similar to the model used by Larsen et al. (2005) and a 4-layer model. To search the best combination of the lithosphere thickness and the mantle viscosities, the combinations of the rheological parameters tested here were 161 in total number. To test the sensitivity of the estimated rheological parameters to the elastic PDIM effects, we have compared two PDIM models; one is EPSL2005 model (Larsen et al., 2005) and the other is JGR2007 model (Larsen et al., 2007). Statistical examinations of the misfits between observations and model predictions, which were based on the magnitudes of the minimum chi^2 and the mean of residuals fitting to a Gaussian function, indicate that the JGR2007 model gives much consistent results than the EPSL2005 model. From the comparison between the observations and the model predictions, we have obtained an asthenosphere viscosity value of the range of 3 to 13 in unit of E18 Pa s in SE-AK. At the 95% confidence interval, our best fit values are 5.6E18 Pa s for the 2-layer model and 1.0E19 Pa s for the 4-layer model. Although it is still within overlapping error ranges, they are 1.5 to 3 times larger than the previous value by Larsen et al. (2005). The most likely reason for the change is the different magnitude of the estimated PDIM effects in the respective studies.