Tallinn University of Technology

Exploration of the sea surface is primarily aimed at precise mapping of the water quality parameters as well as analysis of climate and ecosystem changes.

The research of TalTech oceanographers focused on the development of a new method for reconstruction of changes in sea surface temperature and salinity, the results of which can be applied directly in forecast models. In addition to mapping the undergoing environmental changes, such information will be of direct practical benefit, for example, to fishermen whose work depends on the movement of fish herds.

Merepinna temperatuuri näide uue EOF meetodi abil, kasutades ainult uurimislaevadelt tehtud in-situ mõõtmisi (asukohad näidatud punktidega)“. Joonis prof Jüri Elken.

Sea surface temperature example by the new EOF method, using only the data from shipborne in-situ observations (locations shown by dots). By prof Jüri Elken.

The head of the research group, Professor of TalTech Department of Marine Systems Jüri Elken said that nowadays oceanographers use primarily three types of data for studying the sea surface. "The oldest method is the in situ method, i.e. water characteristics are measured on-site from vessels or, in recent times, also from automated stations, either using sensor data or analysing the taken water samples. The second method is remote sensing: nowadays mostly data from satellites are used, the results of which can be presented, for example, in the form of detailed maps. The third method is numerical modeling, where changes in the sea state are found by sophisticated computational simulation of physical processes, the results of which can be presented in the form of animated maps, i.e. as a kind of cartoon.
All three methods of analysis have their benefits and drawbacks. For example, in-situ observations provide the most accurate data on seawater properties, but there are few observation points that are usually unevenly distributed, which is why the picture is rather fragmented. A satellite image provides detailed information, e.g. about sea surface temperature and chlorophyll content, but not about sea surface salinity, nutrients and a number of other important sea water properties. Numerical marine ecosystem models are increasingly being used in practice, but their accuracy is sometimes rather poor. "Therefore, we use all three methods to investigate the state of the sea and to make forecasts. The formula for success for researchers is the correct combination of the three methods," Jüri Elken says.
Oceanographers use statistical methods to assimilate different types of data. In this study, the traditional EOF (Empirical Orthogonal Functions) method was used, which was complemented: the repeated patterns of surface maps were obtained using the results from the numerical model on a regular computational grid; subsequently, the weights of basic patterns on specific time instances were obtained based on the recent-period observational data. By using this method, the observational data can be extended over regions and times, where no actual observations have been conducted. "If the sea regions are affected by the same climatic factors, like the sea areas surrounding Estonia, then response of sea surface variables in specific non-neighbouring locations is similar. For example, if in autumn the shallow coastal areas of Finland, somewhere near the city of Kotka, cool faster compared to the deeper offshore regions, a similar change is very likely to occur near Pärnu or in the Moonsund. Also, if sea surface salinity near the Narva River is below average, it is likely to be lower at the Daugava River as well. The method for EOF analysis developed by our research group takes these sophisticated correlation patterns into account," Professor Elken says. Thus, the observational data can be extended over a significantly broader study region by using the statistical patterns dominating in a certain area.

"As the output of the research, we developed a new type of algorithm for assimilation of observation data and a corresponding computer program. The first test results indicate that we can significantly reduce systematic (monthly mean) difference of the daily forecasts of sea surface temperature and salinity compared to the actual observational data, which has been a major challenge so far" Professor Elken says.

The international research group, which included, besides oceanographers from TalTech, also scientists from Danish Meteorological Institute, presented the results of its two-year research in the article "Reconstruction of Large-Scale Sea Surface Temperature and Salinity Fields Using Sub-Regional EOF Patterns from Models" published in the journal Frontiers in Earth Science.

Source: Frontiers in Earth Science

Additional information: Professor of TalTech Department of Marine Systems Jüri Elken, juri.elken@taltech.ee
Kersti Vähi, TalTech Research Administration Office