The Miljašić jaruga is one of the most important watercourses in Zadar County, in the Dalmatian hinterland of the eastern Adriatic. The stream runs approximately 16 kilometers from an altitude of about 88 meters above sea level down to the sea roughly 900 meters northeast of the old town of Nin, draining a topographic catchment of approximately 191 square kilometers. Its lowermost 1.5 kilometers form a genuine estuary, directly influenced by the sea. Like most Croatian coastal streams, it flows through classic Dinaric karst terrain, a landscape of Eocene limestone, marl, and flysch where surface and underground waters interact rapidly. The hydrological regime is strongly seasonal: for most of the year, catchment waters flow out to sea, but in the dry summer months the bed is largely without outflow. It is precisely this pattern, a long dry period followed by extreme precipitation, that generates the sudden water-level surges and flash floods characteristic of karst regions.

The 2017 event was textbook flash flooding. Rain gauges and meteorological stations across the catchment recorded three-day totals exceeding 300 millimeters at Zadar and Zadar Airport, with the maximum daily value of 284 millimeters falling on September 11 at Zadar, the second-highest daily record in the station’s history after the 357 millimeters of September 10, 1986. Curiously, the rain gauge at Nin itself recorded almost negligible amounts on those days, showing how localized such storms can be. The rainfall activated violent torrential runoff across the entire 191-square-kilometer basin. At the Poljaki station on the upper stream, water levels began rising at 6 a.m. and climbed 357 centimeters within seven hours, peaking at 367 centimeters at noon. Downstream at Boljkovac, the peak of 353 centimeters arrived at 3 p.m., a typical flash-flood hydrograph shape with steep rise gradients of up to 121 millimeters per hour.

The discharge estimates are extraordinary for a stream whose average annual flow at Boljkovac is just 1 cubic meter per second. The extrapolated discharge curve of the Boljkovac station put the 2017 flood maximum at an estimated 112 cubic meters per second, with a total flood volume of approximately 10.6 million cubic meters. Upstream at Poljaki, the maximum was 91 cubic meters per second and the volume about 7.7 million cubic meters, while the Bokanjačko blato drainage tunnel, which diverts water from an indirect 59-square-kilometer sub-catchment into the stream through a 2-kilometer hydrotechnical tunnel, contributed 1.8 million cubic meters with a peak of 22.6 cubic meters per second. The time lag between the rainfall peak at 8 a.m. and the discharge peaks was four hours at Poljaki and seven hours at Boljkovac. The researchers caution that flows for such extreme levels are based on extrapolated rating curves and may carry relatively large errors, so the official maxima should be treated as estimates. Even so, they capture the order of magnitude of rare hydrological phenomena in this karst catchment.

The flood’s consequences were severe. Floodwater overtopped the embankment built along the stream bed on the eastern side of the Nin salt pan, damaging it along nearly its entire 700-meter length and completely destroying a 50-meter section. Water spilled laterally over the embankment, across the salt pan basin, and along the abandoned old river bed toward the town center. All the salt works pools were flooded and contaminated by torrential water, inflicting the greatest economic damage. Two historically valuable bridges in Nin were significantly damaged, their partial collapse aggravated by reduced flow capacity from sea-level rise since they were built. Roads, beaches, and numerous residential and commercial buildings were also hit. The salt exploitation at Nin dates back more than 1,500 years, and the modern salt works established in the mid-twentieth century cover 55 hectares, producing excellent-quality salt from seawater, sun, and wind in a chain of shallow pools.

To understand how rare the 2017 flood actually was, the team turned to flood frequency analysis at the Boljkovac hydrological station, comparing hydrological series of 26 and 27 years using three theoretical distributions: normal, log-normal, and Gamma. They deliberately analyzed water levels rather than discharges, because annual maximum discharges are obtained by extrapolating the rating curve, whereas water levels are continuously recorded by limnigraph and are therefore more reliable input data. The agreement between empirical and theoretical distributions was tested with the Kolmogorov–Smirnov test, and all three distributions proved statistically acceptable, with the Gamma distribution showing the best fit. But the calculated design levels diverged sharply. The 2017 flood level of 353 centimeters came out with a return period greater than 500 years under the normal distribution, roughly 500 years under the Gamma distribution, and slightly less than 200 years under the log-normal distribution.

The variability grew worse at longer return periods and, strikingly, hinged on a single data point. Comparing the 27-year series, which includes the 2017 flood, with the 26-year series that excludes it, the estimated 100-year water levels differed by 39 to 60 centimeters depending on the distribution. Translated into discharges through the Boljkovac rating curve, the 100-year high-water flow ranged from 79 cubic meters per second under the normal distribution to 101 cubic meters per second under the log-normal distribution, a difference of roughly 20 percent. Excluding the 2017 event dropped the 100-year flows to between 60 and 85 cubic meters per second, a decrease of 16 to 29 percent, even though the two input series differed by only one term. Adopting the Gamma distribution as the most relevant, the team settled on characteristic flows of 67, 81, and 92 cubic meters per second for return periods of 20, 50, and 100 years, and characteristic levels of 252, 284, and 308 centimeters for the same periods.

For engineering practice, these numbers matter enormously, because they directly determine how flood defense structures are sized, including the height of embankments protecting populations along rivers and estuaries. After the 2017 flood, all damage to the existing embankment was repaired and a new clay embankment was constructed on the destroyed section, with a projected crown width of 2.5 meters and an adopted crown height of 3.0 meters above sea level, calculated to accommodate a 100-year high-water event. Yet all three distributions indicate that the 2017 flood had a return period significantly longer than 100 years. The authors note that reconstruction or construction of new flood defense facilities in rural or smaller urban areas is commonly designed for return periods of up to 100 years, and that experience from Dalmatian catchments and the current issue of climate change suggest extreme events are occurring more frequently and with greater intensity than before.

Geographically, the study also underscores why Croatian estuaries are so vulnerable to fluvial flooding. In the Adriatic, fluvial flooding dominates owing to strong orographic influence: mountain massifs higher than 1,000 meters, running northwest-southeast along the Dinaric coast, sit in the immediate vicinity of the sea and squeeze intense short-term precipitation out of passing weather systems. Meanwhile, the sea’s influence as the downstream boundary is comparatively weak, because tidal amplitudes rarely exceed 40 centimeters in the southern Adriatic and reach about 1 meter in the north. Rivers therefore exert the dominant influence on Croatian estuaries, and combinations of high sea levels and intense basin precipitation can still produce fluvio-tidal floods that raise estuary water levels further. Rising sea levels and intensifying extreme climatic conditions are expected to increase inundation hazards in estuaries, and the authors argue that future research should consider multivariate flood frequency analysis for estimating coastal-estuarine flooding risks, rather than the classic univariate approach applied here because the 2017 flood was caused exclusively by inflow from the basin.

The team’s recommendations reach beyond the embankment. They propose catchment-wide water management, including the construction of retention basins and barriers to calm flows and prevent sedimentation, and note that a retention basin called Rašinovac is planned for the Miljašić jaruga, approximately 3.2 kilometers upstream from the river mouth. High-quality catchment management, they stress, requires engagement across several professions, from spatial planning and climatology to hydrology. The broader lesson is methodological: the procedure for determining relevant hydrological quantities is complex and ambiguous, and continuous validation with the widest possible analysis of all available climatic, hydrological, geological, and land-use parameters is required to quantify runoff processes as precisely as possible. Hydrological bases used for designing hydrotechnical systems should be updated and validated as frequently as possible, especially under obvious climatic variation. For the residents of Nin, whose town sits on the alluvium of a stream that can rise three and a half meters in a single afternoon, that statistical humility is not an academic nicety but the difference between a defense system that holds and one that fails when the next historic storm arrives.

Subject of Research: Flash flood hydrology and flood frequency analysis of the Miljašić jaruga estuary in Nin, Croatia

Article Title: Estuary fluvial flooding: the example of the watercourse Miljašić jaruga in the town of Nin (Croatia)

Article References: Ljubenkov, I., Oskoruš, D., Rubinić, J., Peša, I., & Papić, J. (2024). Estuary fluvial flooding: the example of the watercourse Miljašić jaruga in the town of Nin (Croatia). Estuarine Management and Technologies, 1(1), 19-37. https://doi.org/10.3897/emt.1.120413

Image Credits: AI Generated

DOI: 10.3897/emt.1.120413

Keywords: flash flood, estuary flooding, karst hydrology, Croatia, flood frequency analysis, climate change, Zadar, Nin salt works, Estuary, fluvial, flooding, example