Selected large-scale climate pattern indices and their relationships with discharge in Czechia based on wavelet transform

Código

I-EBHE0174

Autores

Ondrej Ledvinka, Katerina Vackova, Vit Stovicek

Tema

WG 1.10: Hydrologic Design - Solutions & Communication

Resumo

In climatology, it is known that large-scale climate patterns may be well correlated with variables such as air temperature and precipitation. Therefore, such patterns may serve as good explanatory variables when it comes to seasonal predictions since usually lagged correlations can be witnessed by the available data. As climate is one of the most important drivers of hydrological response, it is believed that hydrologists may also benefit from utilizing these patterns described by various indices. In Czechia, a few studies have been performed before to describe relationships between those indices and discharge. However, they took into account only a very limited number of catchments, which have not provided hydrologists with sufficient information on regional and temporal variability, or even changes, possibly considering different features of the catchments such as their area, prevailing geology, soils, vegetation cover, geomorphology, etc. We decided to bridge this gap and, by using freely available discharge data from Czechia, carry out a massive analysis of relationships between 303 uninterrupted time series of mean monthly discharge and monthly values of four selected climate indices: Atlantic Multidecadal Oscillation (AMO), Arctic Oscillation (AO), North Atlantic Oscillation (NAO), and El-Niño Southern Oscillation (ENSO; described by the Southern Oscillation Index, SOI). The analyzed period was January 1981 to December 2022, whose start was determined by the availability of discharge data, and the end was given by the last known value of the index related to AMO. Continuous wavelet transform (using the Morlet mother wavelet) was performed individually for all discharge series as well as for all series of climate indices. To look at the relationships and study the differences or similarities, we created plots showing wavelet coherence with regions where the coherence exceeded significance limits. Additionally, we included arrows indicating whether the variables were in phase, anti-phase, or if other important relationships appeared and disappeared in various periods on various aggregation scales. Since we were mostly interested in other than seasonal cycles known to be triggered by other forces, discharge series were standardized so that the values in individual months had a mean of zero and a variance of one before wavelet analyses. Now, all 4 + 303 + 4 * 303 = 1519 plots are being studied, and other analyses (e.g. incorporating gridded climate variables from Czechia) are in preparation. So far, it has been found that significant relationships are appearing in the plots as isolated islands while we want to extract the information valuable primarily for seasonal predictions to support water managers in Czechia.