HR: 1330h
AN: G52B-0047    [PDF]
TI: Towards a Realistic Uncertainty Budget for GPS Heights
AU: * Sch\"on, S
EM: schoen@dgfi.badw.de
AF: Deutsches Geod\"atisches Forschungsinstitut (DGFI), Marstallplatz 8, Munich, 80539 Germany
AU: Kutterer, H
EM: kutterer@dgfi.badw.de
AF: Deutsches Geod\"atisches Forschungsinstitut (DGFI), Marstallplatz 8, Munich, 80539 Germany
AB: Precise GPS-determined heights play a key role in many geodetic and geophysical applications, in particular as input for studies on loading or rebound effects. For a meaningful interpretation of the heights and height changes their effective uncertainty has to be considered. It consists of stochasticity (reflecting the purely random component) and imprecision due to unknown systematic deviations between observations and model. Its assessment is indispensable for the distinction between signal and noise. In case of local GPS networks the entire uncertainty budget can sufficiently be described by normally distributed measurement errors since many disturbing effects are cancelled out by measurements methods or analysis strategies. In case of regional (or global) networks imprecision becomes the dominant component. In the common approach, both components are modeled by stochastic means. However, there are two main shortcomings yielding too optimistic uncertainty measures of the results. First, this strategy pretends a reduction of systematics. Second, since GPS permanent networks provide a sufficiently large number of observations, the total uncertainty of the results will decrease below any limit. As both effects massively contradict any practical experience, stochasticity and imprecision should be treated in a more adequate way. For this purpose, this contribution focusses on the description of imprecision by means of deterministic intervals without any stochastic properties. In a first part, the imprecision of GPS phase observations is quantified by intervals using a forward modelling approach based on a set of basic parameters. The different processi ng and correction steps which are applied to the original GPS observations in GPS data analysis (such as, e.g., due to tropospheric or ionospheric refraction, or antenna phase cen ter variation) are studied in detail. As a result, typical imprecision intervals are composed which show the impact of standard correction models as well as of different baseline lengths and satellite constellations. In a second part, these intervals are transferred to the estimated point coordinates. It will be shown that the imprecision is maximum for the vertical coordinate component (height). This is similar to the random errors which are adequately represented by the variance-covariance matrix. The extended uncertainty is finally derived by superposing stochasticity and imprecision. Numerical examples are computed for a network of GPS permanent stations in Europe (EUREF) in order to illustrate the theoretical results.
DE: 1204 Control surveys
DE: 1206 Crustal movements--interplate (8155)
DE: 1243 Space geodetic surveys
DE: 1294 Instruments and techniques
DE: 9820 Techniques applicable in three or more fields
SC: Geodesy [G]
MN: 2003 Fall Meeting