[thread] Last month colleagues from @OGS_IT, @NIB_MBP_SI, Paris ENSTA-Tech and @meteoSI published a case study in @EGU_NHESS: a simulation of an actual windsurfer's drift in the ocean boundary layer in the Adriatic Scirocco storm in Oct 2018.
https://nhess.copernicus.org/articles/20/2335/2020/nhess-20-2335-2020.html
(1/n, n>>1)

I'll go through the work quickly - it's a very simple paper, perhaps some of it will be interesting to someone. No prior specific knowledge is needed.

On 29 October 2018 a windsurfer's mast broke about 1 km offshore from Istria during a severe Scirocco storm and drifted in severe marine conditions until he eventually beached alive and well in Sistiana (Italy) 24h later. We wanted to see if we could compute his trajectory.

He was through quite an ordeal. He was drifting in gale force Scirocco [SE, waves 3-5m, winds 15-25 m/s], clutching his windsurf board by the footstraps most of the night. In the morning, he drifted into a cold Soča river plume - with max discharge in months, see satellite pic.

Within the Soča plume, he was struggling to keep upstream, because the plume forms a westward inertial coastal current (due to Earth rotation), dragging him away from Sistiana and towards Grado. He broke out of the plume at around 10-12 am on 30 Oct by paddling intensely.

So he is somewhere offshore Monfalcone at 12 am. And again he starts to drift away from the Sistiana coast toward Grado. Why? This seems consistent with the tidal motion. See tidal elevation for 30 Oct 2018 below. Between 12h and 20h there is an ebb coming.

This means the sea level in the Gulf of Trieste is dropping and the water mass is leaving and emptying the Gulf - carrying the survivor along and away from his eventual beaching site in Sistiana. So it wasn't over till it was over - he had to paddle all the way to the coast.

We conducted an interview with the survivor and he reconstructed his trajectory for us by memory - red line in the right panel. However, we had to verify his subjective trajectory reconstruction independently.

To do this, we used ( @OGS_IT and @NIB_MBP_SI operated) high-frequency radar (HFR) measurements of surface currents in the Gulf of Trieste at the time of the drift. HFRs deduce surface currents by measuring Doppler shift of radio waves scattered off the moving ocean surface.

Once you know surface current and wind fields, you can release virtual particles into the modeled ocean and allow them to get carried around by these currents/winds, and see where they end up. They drift downwind (drag) *and* crosswind (lift) - this depends on the object shape.

A human body will drift differently than a person on a surfboard. Or an oil spill. Or a cargo container. US Coast Guard has performed a number of experiments to measure the drifting properties of each of these different objects. So this is known.

Windsurfer's accident location is outside of the HFR domain, but windsurfer's beahing site and beaching time are exactly known and within the HFR domain. We hence released virtual particles in the ocean at the *beaching* site and made them drift *upstream* and *backward* in time.

This back-propagation simulation provided an independent verification of survivor's trajectory estimate. Turns out the survivors trajectory was actually quite good, see pic below: red line is his estimate, yellow line is the one from our back-propagation simulation.

We then proceeded with a forward-propagation simulation: we released virtual particles into modeled ocean again, this time at the *accident* location, and propagated them *forward* in time. For this simulation one needs to model the full state of atmosphere and the ocean first.

It turns out that the majority of particles lagged a few hours behind the trajectory but there were clear streaks of particles reaching Sistiana coast, indicating this region as a potential search and rescue (SAR) area. The diffusion is high, however. Search needs to be rapid.

What's more, wind-only simulations were worse than full wind and current simulations. Ocean currents matter, especially in topographically constrained gulfs - and in spite the fact that wind contribution to the drift was dominant in this case.

I'll conclude by saying that this accident, and survivor's willingness to help us, kick-started a development of a numerical environment which may provide useful SAR estimates if this happens again - hopefully not ever, or at least not anytime soon and not during Bora conditions.