Tuesday 23 August 2005

GP2

PT0081
Modeling non-tidal sea level variations via fusion of the data coming from different satellite altimetry sensors
Ardalan, Alireza¹, Shojaee, Kamiar¹
1 University Of Tehran, Department of Surveying and Geomatics Eng., Iran
Author email: ardalan@ut.ac.ir
Instantaneous sea level is the result of the gravitational (tidal) and non-gravitational (non-tidal) forces such a winds, currents, variations of salinity, variations of depth, atmospheric changes, and the shape of the coast line. The gravitational effects can be well modeled through tide gauge observations and satellite altimetry data. As a result, nowadays, accurate models for estimation of the tidal variations are available. However, modeling of instantaneous sea level with high accuracy, besides the tidal models, requires accurate modeling of non-gravitational effects. In this research, modeling of sea level variations due to non-tidal effects is tackled. As the mathematical tool Auto-Regressive Integrated Moving Average (ARIMA) technique is used. The modeling is operational performed as follows: First from the satellite altimetry data, the tidal effects are removed. Next, the modeling of sea surface variations due to non-tidal sources is done in two modes: (1) direct mode, and (2) indirect mode. In the direct mode, by applying difference operator of various order, as well as trend analysis, the sea level variations due to non-tidal forces are made stationary and introduced to ARIMA machinery. Within the indirect mode, differencing and removal of trend is applied to atmospheric observations made by on board sensors of the altimetry satellites. The two computational strategies are tested by using 10 years of Topex/Poseidon satellite altimetry data over Persian Gulf and Oman Sea. According to the results at the test area in 90% of the covered test area, 'mode 1' is more successful, and 'mode 2' yields good results only at the coastal areas, and shallow waters. Theoretical details and the results of the case study will be presented.

Return to Poster Presentations