The statistics may help prevent future river tragedies

Hundreds of tons of dead fish poured into the Oder River in early July and August 2022. The scale of the event was devastating. Despite quick efforts, Polish and German authorities struggled to determine the cause of this environmental disaster for several weeks.

It is now known that a fatal combination of multiple factors contributed to the accident. Is there a way to prevent similar tragedies in the future?

Researchers from Bydgoszcz University of Science and Technology (PBS) in Poland believe there is potential to avoid it, but it will require a change in the approach to analyzing parameters recorded at river monitoring stations. Details of their proposal were explained in the magazine Holistic ecology.

Water is a vital element of Earth's environment, and plays a critical role everywhere in supporting life. Despite its fundamental importance, accurate methods for assessing water quality have not been developed. In countries such as Poland, a direct classification into five categories has been used.

Globally, the Water Quality Index (WQI), which ranges from 0 to 100, is widely accepted. This indicator is a simple weighted average based on different water parameters measured by monitoring stations. However, the challenge lies in potential differences in statistical weights and parameter sets, leading to multiple definitions of the WQI parameter in practice.

“More importantly, the water quality index suffers from a structural flaw: the basic summation method used in its calculation allows a positive shift in one parameter to compensate for a negative shift in another, and vice versa. As a result, the aquatic ecosystem may exhibit significant dynamics” that They are detected in the changes reflected by the indicator,” explains Dr. Grazyna Czerniak (PBS), lead author of the article presenting the innovative statistical analysis.

“Recognizing the shortcomings of the WQI parameter, we chose to develop a new index, which is statistical, more versatile and responsive to changes in the values ​​of individual parameters,” she adds.

As a chemometrics specialist, Dr. Czernyak is involved in multidimensional statistical analysis of physical and chemical phenomena. This abstract term includes mathematical techniques designed to extract maximum information regarding the dynamics of phenomena and their interrelationships from measurement data involving different physical quantities within a single system.

Rivers provide a suitable example in this context. Describing their state requires knowledge of many physical quantities (such as water temperature or electrical conductivity), chemical quantities (such as water acidity, oxygen levels, or concentrations of specific chemical compounds), and biological quantities (such as the number of diatoms or algae).

Each available parameter can be treated as an independent dimension within a defined space, where points correspond to individual measurements – water samples. Through principal components analysis, two or three new artificial variables are usually calculated for individual points or samples.

These variables are designed to better capture the correlations between several parameters. The final step involves checking whether the values ​​of these new variables, converted into a single indicator for individual samples, meet the specified statistical quality criteria. In the analysis of water from the Oder River, the T2 criterion, introduced by the American statistician Harold Hotelling, turned out to be decisive.

The basis of the research conducted at PBS was data from physical and chemical measurements collected by more than 30 Polish stations monitoring the Oder River. This data spanned from the early days of August 2022 (after the initial observations of dead fish) until mid-May of the following year. Information from the German monitoring station at Frankfurt (Oder) proved to be a valuable addition, including many biological parameters and covering in particular the period leading up to the disaster.

“Our statistical tests showed that alarms about exceeding the Hotelling criterion could be linked to press reports about ongoing pollution in the Oder River. What is important here is the fact that the criterion worked as expected both on the maximum set of parameters available and when we limited this set to three Only physical and chemical quantities,” emphasizes Dr. Cherniak.

Analysis of data leading up to the Oder disaster yielded results that are both interesting and promising. It was revealed that a month before the initial fish kill, the evolving statistical indicator consistently exceeded the alarm criterion, maintaining a slightly fluctuating pattern until the collapse of the river ecosystem, which was marked by a noticeable peak in the histogram.

This pattern indicates that the chosen parameter is indeed sensitive to synergies that may not be immediately apparent, which occur between changes in various physical and chemical parameters of the river.

“Statistical parameters may sometimes generate inaccurate alerts, and a single alert may be overlooked. However, if results consistently exceed alert levels throughout the week, this should raise serious concerns. In Oder's case, signs of changes became apparent a month before,” said Dr. Cherniak: “If our statistical method had been applied at that time, there would have been a two or three-week period for potential preventative action.”

The method presented by the physicists from PBS is easy to use, does not require large computational resources, and is applicable to monitoring various rivers. However, the statistical tool needs to be pre-trained on a reference set of appropriate quality, containing parameter values ​​representative of the typical case of a given river.

For streams already under intensive monitoring, adapting the instrument can be quick and cost-effective.