Quantitative error evaluation of GNSS-A observation using new analysis tool "GARPOS": high-rate GNSS and ocean field disturbance

Abstract:

GNSS-A seafloor geodesy has monitored time variation of cm-order crustal deformation such as postseismic phase following the 2011 Tohoku-oki earthquake, interseismic coupling condition and shallow slow slip events along the Nankai Trough [e.g., Yokota et al., 2016; Yokota & Ishikawa, 2020]. Although the frequency and accuracy are being improved, one of the major issues is that quantitative error evaluation has not been performed. The expression of the error is indispensable for implementing the time series analysis such as Kalman filter. It is also desirable to express the errors clearly when statistically processing the final time series and performing SSE detection, time variation of the coupling condition, and analysis of postseismic effects. The main error factors of GNSS-A are high-rate GNSS error and error due to ocean disturbance. We are currently aiming to develop these quantitative analytical methods. In this presentation, we discuss the initial results.

The effects of GNSS errors on GNSS-A observations can be estimated from empirical GNSS errors. The effects of bias during observation cannot be estimated by analysis. (For example, if there was a 5 cm shift during observation, it cannot be corrected.) This quantity depends on the observation time and is estimated to be 2-3 cm (1σ) vertically and 0.5-1 cm (1σ) horizontally.

The effects of variability during observation are often confused with ocean disturbance errors. Therefore, it is thought that there are many cases where it was mistakenly taken out as an ocean disturbance during the analysis. Since the development of new GNSS-A analysis software “GARPOS” [Watanabe et al., this AGU] has made it possible to perform a large amount of GNSS-A analysis under various conditions at high speed, we prepared a large amount of assumed high-rate GNSS errors and estimated the GNSS error amount of each epoch. As a result, it was suggested that in the estimation of the ocean disturbance, the amount for the up-down terms could be extracted appropriately, but the amount for the horizontal term was not sufficiently estimated. Such analysis may be able to estimate the error amount of the extracted ocean disturbance. In the near future, it is necessary to confirm characters of GNSS errors amount on the disturbed sea surface. We also plan to analyze the ocean disturbance for each frequency.