Verification of the optimum tropospheric parameters setting for the kinematic PPP analysis

Abstract:

Kinematic GNSS analysis is useful for extraction of the crustal deformation phenomena between seconds to one day such as coseismic and postseismic deformation after a large earthquake. The kinematic GNSS analysis, however, have fundamental difficulties for the separation between unknown parameters such as the site coordinate and tropospheric parameters, caused by a strong correlation between each other. Thus, we focused on the improvement of the separation precision between coordinate time series of kinematic PPP and wet zenith tropospheric delay (WZTD) based on the comprehensive search of the parameter space.

We used GIPSY-OASIS II Ver. 6.3 software for kinematic PPP processing of whole GEONET sites in 10 March 2011. We applied the every 6 hours nominal WZTD value as a priori information based on the ECMWF global numerical climate model. For the coordinate time series and tropospheric parameters, we assumed white noise and random walk stochastic process, respectively. These unknown parameters are very sensitive to assumed process noise for each stochastic process. Thus, we searched for the optimum two variable parameters; wet zenith tropospheric parameter (named as TROP) and its gradient (named as GRAD). We defined the optimum parameters, which minimized the standard deviation of coordinate time series.

We firstly checked the spatial distribution of optimum pair of TROP and GRAD. Even though the optimum parameters showed the certain range (TROP: 2×10^-8 ~ 6×10^-7 (horizontal), 5.5×10^-9 ~ 2×10^-8 (vertical); GRAD: 2×10^-10 ~ 6×10^-9 (horizontal), 2×10^-10 ~ 1×10^-8 (vertical) (unit: km·s^-½)), we found they showed the large diversity. It suggests there are strong heterogeneity of atmospheric state. We also estimated temporal variations of optimum TROP and GRAD in specific site. We analyzed the data through 2010 at GEONET 940098 station located in the most southern part of Kyusyu, Japan. Obtained time series of optimum GRAD showed clear annual variation, and the time series of optimum TROP especially estimated from vertical component showed a constant value (~2×10^-8). The annual variation of optimum GRAD may represent that there is a strong variability of vapor distribution through the year. Taking these results into account, we can get the optimum parameters in advance at each station for routine GNSS analysis.