Detection of trends in observed river floods in Poland

Poland is characterized by hydrometeorological variability, where conditions such as ice melt, snowmelt, extreme precipitation, and other factors might cause river floods. Climate change has had a direct influence on it, due to the shifts in parameters such as precipitation and temperature over the world. Therefore, evaluating the historical flood trends allow to establish a link between hydrology and water resource management.The purpose of this study was to present trends in selected river flood indicators (magnitude, frequency, and timing) for a data set of 146 stream flow gauges, using the annual maximum daily flow and peak-over-threshold approaches. The gauges are also free from major human influences on flow regimes such as large dams. Two periods (1956-2019 and 1981-2019) were analyzed. The first one focused on a large temporal coverage (58 flow gauges), while the second one maximized spatial coverage (146 flow gauges). Trends were calculated using the non-parametric Mann-Kendall test (MK) and Sen slope for changes in magnitude and timing, while flood frequency trend was detected by chi-square test on parametric Poisson regression.The flood trend detection in river flood indicators allowed us to identify areas with predominant decreases or increases. The northeast region of Poland showed a downward trend in flood magnitude with changes between -5 to -15% per decade using both methods, whereas the south part showed a predominantly upward trend. The results are in general agreement with previous studies based on annual maximum flow. No evident trends were detected in flood frequency, where detected changes generally did not exceed ± 0.2 events per decade. An annual flood timing average on February-April was found in the northern and central parts of Poland, while May-June was the average in the southern-most basins. The flood timing showed a strong pattern of earlier flood occurrence in the southern half of the country, and an increasing trend (later flood occurrence) in the north-eastern and north-western parts, with changes of 4 and 8 days. The results showed the first analyses of a rather under-studied topic in Poland and leave an opportunity to study the attribution of changes in floods.