We will collect long-term time series of precipitation from rain gauges in major cities and towns along the Croatian coast, as well as water level and discharge time series from hydrological stations near the mouths of coastal rivers. We will also use long-term data from official tide gauges in Croatia. Although the observation network in Europe is among the best in the world, tide gauges are still sparse in Croatia and wave gauges have just been installed recently. Therefore, we will combine measured data with numerical reanalyses.

Sea level and salinity data for the entire Adriatic Sea will be obtained from the Copernicus Marine Service (Mediterranean Sea Physics Reanalysis).

Sea surface waves will be similarly obtained from the Copernicus Marine Service (Mediterranean Sea Waves Reanalysis,).  Since there are some discrepancies between the observations and numerically obtained data, we will adjust the numerical data (sea level, surface waves, and salinity) to the localised point observations and/or consider its uncertainty. In addition to sea surface waves, we will also obtain wind data from the ERA5 reanalysis.

Finally, we will also obtain climate projections from the Coupled Model Intercomparison Project (CMIP6) that support the IPCC sixth Assessment Report.

All these data will be systematised, preprocessed and integrated into a hydrometeorological database for analysis at regional and local scales.

Using the hydrometeorological data, we will assess the compound flooding potential for the Croatian coast. The Dependence Extreme Method and Joint Occurrence Method will be used to quantify the compound flooding potential index. We will consider the co-occurrence of storm surges and:

• large waves (source of coastal flooding and overtopping),
• high river discharge (source of estuarine flooding),
• low river discharge (source of saltwater intrusion), and
• high precipitation (source of urban flooding).

Neglecting the seasonality of rainfall and sea levels can lead to a significant underestimation of expected annual damages from compound flooding. Therefore, this project proposal will go a step further and evaluate the seasonal correlation and co-occurrence for all four combinations by examining the monthly maximum values. Finally, the correlation coefficients, the number of co-occurrences and the compound flooding potential index will be mapped along the Croatian coast and systematized in a GIS database.

At the local scale, we selected three pilot areas where we will map compound flooding hazards:

• Rovinj (high sea levels + extreme waves),
• Biograd na Moru (high sea levels + extreme precipitation),
• Neretva River Estuary (high sea levels + high discharge/low discharge).

All three locations were classified as very susceptible to flooding by the Preliminary Flood Risk Assessment for Croatia.

We will calculate the bivariate joint probabilities of high sea levels and: a) large waves, b) high river discharge, c) low discharge, and d) high precipitation, using copula-based methodology. We will derive boundary conditions for numerical modelling of compound flooding hazards by sampling the joint probability. This approach will be improved by considering non-stationary scenarios, defined by climate projections from CMIP6. Using climate projections for precipitation, we will apply appropriate machine-learning methods to evaluate potential future changes of extreme discharge in the Neretva River Estuary. To evaluate the compound flooding hazards we will create several numerical models and perform simulations using a wide range of scenarios.

• In the pilot area of Rovinj, we will model the compound flooding hazards resulting from a combination of high sea levels and extreme waves that can lead to coastal flooding and excessive overtopping.
• In the pilot area of Biograd na Moru, we will model compound flooding hazards resulting from a combination of high sea levels and extreme precipitation that can lead to urban flooding. 
• In the pilot area of Neretva River Estuary, we will model two flooding mechanisms, a combination of high sea levels and: a) high river flow that can result in large scale (fluvial and coastal) flooding, and b) low river flow that can lead to increased saltwater intrusion (high risk to agricultural activities). 

Finally, in all three pilot areas, we will map the obtained results and produce a hazard map illustrating spatial distribution of depths, velocities, and flood severity for three different probabilities in the present and future climate. 

To better understand the microtidal and baroclinic effects of flooding mechanisms in estuaries we will establish a field monitoring system in the Rječina River Estuary (RRE) in the city of Rijeka. In addition to the official observations, we will establish a new continuous measurement framework at the RRE consisting of:

• Flow meters, one installed upstream of the RRE mouth (outside the tidal limit), and the other installed near the river mouth. 
• Acoustic velocity profiler installed near the RRE mouth to measure the vertical velocity profile and river outflows.
• A series of conductivity-temperature-depth (CTD) data loggers, to measure salinity profile and estimate salt-wedge dynamics.

The total planned operating time of the monitoring system operation is 24 months to maximize the likelihood of capturing at least one extreme storm surge and high discharge; however, the flow meters will remain in the RRE after the completion of this project and may be integrated into a flood early warning system.

The field monitoring system data will be used to establish a numerical model of RRE. First, the model parameters (bed friction, interfacial friction, and entrainment) will be calibrated using field measurements. Second, the model will be validated for different scenarios (flow conditions). Third, the model will be used for probabilistic simulations to investigate the sensitivity of estuaries to different combinations of external forcings (tides, storm surges, and discharges). We will use these results to determine the importance and impact of stratification (baroclinic effect) in compound flooding in estuaries and apply the new insights to the local scale analysis.

All experience, knowledge, and insights on compound flooding will be summarized to produce a guidance document that will support the next generation of national flood risk management plans in Croatia. We will produce guidelines for Compound Flood Risk Analysis and Mapping that will include recommendations for:

• collecting and processing input data,
• calculating joint probabilities for different sources of flooding,
• defining appropriate scenarios,
• selecting, setting up, and running numerical models,
• mapping compound flood hazards.

These guidelines will be disseminated through two workshops.