| one publication added to basket [350494] | Identifying the causal network of sea level variability domains in the Southeast Pacific.An application of satellite altimetry
Schütt, E. (2021). Identifying the causal network of sea level variability domains in the Southeast Pacific.
An application of satellite altimetry
. Thesis. NIOZ Royal Netherlands Institute for Sea Research: Yerseke. 24 pp.
|
| Abstract |
Accurate projections of sea level rise are highly important for decision makers and the population in the low-lying coastal zone. However, due to the complex causes for the variability of sea level, reliable regional predictions remain a challenge. Understanding local and regional factors that contribute to this variability will improve the understanding and predictability of sea level. Sea level variations can be caused by many different factors, e.g. local and remote variability of atmospheric or oceanic circulations causing wind stress anomalies or changes in temperature or salinity.In this project, more than 26 years of satellite data measurements of monthly mean sea level anomalies and the dimensionality reduction algorithm deltaMaps (Fountalis et al. 2018, Falasca et al. 2019) to derive regions with similar SLV dynamics, so called domains. The sea level signals of the different domains are then fed into the causal inference algorithm PCMCI (Runge et al. 2019) to identify the causal network between the domains. A subset of the causal network in the Southeast Pacific is then analysed using current velocity data from the CMEMS' GLORYS reprocessing and wind and sea level pressure data from ERA5 provided by ECMWF. The results highlight the importance of the El Niño-Southern Oscillation (ENSO) on both atmospheric and oceanic processes and suggest that SLV patterns in the Southeast Pacific are sensitive to the "flavour" of ENSO. However, this data driven approach can not accurately determine and quantify the physical processes at work and some questions remain unanswered. An approach that couples observational data and modelling may help to overcome this constraint. |
|
