Oral presentation at
CGPS Workshop, Toulouse 2002
(extended Abstract)
Correction of a jump in tide gauge time-series using altimetry data
1Institute of Physical Geodesy;
Darmstadt University of Technology
In long-term sea level change studies the stability of the reference point of the tide gauge station is very important. Altimetry data of the Topex/Poseidon and ERS-2 missions are used to correct for a jump in the time-series of six tide gauge stations in the Ionian Sea, caused by a change in the zero-level of the tide gauge in Summer 1998. T/P is used here as reference. ERS-2 altimetry data are corrected for a relative drift with respect to T/P and the long-term sea level changes computed from T/P and from ERS-2 altimetry data during 1995-2001 agree after this correction. Also the long-term sea level change computed from the corrected tide gauge series agrees with the altimetric long-term sea level : a sea level drop occurs in the Ionian Sea in the interval between 1995.5 and 2001.5, as already observed from the altimetry data.
1.Introduction
Hourly data of the six tide gauge stations of Catania, Messina, Crotone, Otranto, Bari, Taranto located in Southern Italy have been made available from the Servizio Idrografico e Mareografico Nazionale (SIMN, Italy) from 1991 to 2002. Since July 1998 SIMN is taking care of the quality of the data. Before that date gross errors can be present in the data, in particular caused by the use of different zero points. The last change in the reference point occurred for all stations in Summer 1998, but it was not carefully documented in order to allow a reduction of the measurements to the same zero level. Due to this possible bias within the data the tide gauge stations are not suitable for estimating long-time sea level change. The difference in the reference levels before and after 1998 is estimated here using altimetry data of the Topex/Poseidon (T/P) and ERS-2 missions.
2. Data Analysis
Monthly mean are computed from the tide gauge data. The change in the reference point made in 1998 is evident in all the time-series. Comparison between altimetry data and tide gauge stations of higher accuracy and without known jumps have shown that T/P gives the most accurate and stable altimetric data set (Fenoglio-Marc, 2002). Monthly mean grids of sea level heights are computed from T/P altimetric data, montly grids from ERS-2 altimetric data after application of an empirical correction for the relative drift between ERS-2 and T/P derived from dual crossover analysis (Fenoglio-Marc 2002). The inverse barometric correction is not applied to both altimetry and tide gauge data. The nearest grid node to each tide gauge station is considered and the corresponding time-series used.
Most of the six tide gauge time series have jumps before 1995. We therefore use the tide gauge data starting from 1995.5, that correspond to the start of the ERS-2 data. The linear sea level change in mm/yr at the altimetric grid node near to each tide gauge station is evaluated in the interval 1995.5 - 2001.5. The agreement between the T/P and the ERS-2 corrected time-series data is better that 0.8, no significant improvement is observed with the ERS-2 corrected data.
To estimate the bias between the two time-series due to the change in the reference points in Summer 1998, two intervals of three years each are considered before and after July 1 1995. At each station the mean of the differences between the tide gauge and the T/P monthly means is evaluated on both intervals. The difference between the two means gives the height difference between the two reference points and the tide gauge series after July 1995 are corrected by this amount. The standard deviation of the differences and the correlation between the two time-series are indicators of the agreement between the altimetry and tide gauge time-series. To estimate the goodness of the corrected tide gauge time-series the long-term sea level change is estimated as well as the correlation between the stations. The values obtained for the linear change are comparable to the altimetric values and the correlation between the stations is higher than 0.8 for all station pairs.
More abstracts