than 5% to disperse naturally since the HFO is highly viscous even when fresh. Therefore, in the absence of recovery of the oil from the sea surface 90-95% of an HFO spill will typically impact the shoreline.
However, the precise fate of the HFO at the SEA EMPRESS spill will have been influenced by whether it was released outside the Haven (230 t from the ship's gauges), while the SEA EMPRESS remained grounded, or inside the Haven alongside the jetty (250 t). The HFO released at the mouth of the Haven between 15-22 February was, it is believed, released at the same time as the Forties Blend crude oil. Since the oils are readily miscible the HFO would have been mixed into the Forties Blend crude and affect the properties of the oil mixture on the sea surface. However, since the HFO formed a small percentage (on average 0.3%) of the oil mixture, the properties of the mixture would have been dominated by the properties of the Forties Blend crude.
At the BRAER incident a similar mixing of HFO into the crude oil cargo (combined with the high sea states) explained how the HFO was dispersed completely into the water column by natural dispersion. In the absence of the crude oil, even under those high wind conditions at the BRAER we would have expected a shoreline impact from the HFO released from the tanker.
However, the 250 t of HFO which was released on the 22 February inside the Haven was not accompanied by a release of Forties Blend. The HFO impacted along most of the coastline inside the Haven. The fraction of the oil pollution that was HFO ranged from almost entirely HFO (Pwllcrochan) to around a third of the total pollution at Gelliswick. A rough estimate of the level of oiling from HFO can be gained from the length of coastline inside the Haven, around 50 km, giving a typical level of oiling of 5-7 t/km.
The persistence of the HFO meant that inside the Haven, where it was not mixed with the crude oil to a great extent, it resulted in a shoreline impact proportionately much higher than the Forties Blend crude oil.
MOBILISATION OF THE DISPERSANT OPERATION
In response to the grounding of the SEA EMPRESS at the mouth of Milford Haven, MPCU activated the UK national contingency plan and immediately deployed surveillance aircraft to fly over the vessel to estimate the extent of the spill. Information on the position and size of the slick which was neededn for making operational decisions was provided by MPCU's 2 dedicated remote sensing aircraft equipped with Side-Looking Airborne Radar (SLAR), and downward-looking Video, IR, and UV cameras. 7 DC3 dispersant aircraft were loaded with dispersant and flown to the scene in readiness to begin spraying operations at first light, if required.
Predictions of where the major areas of oil contamination were to move and the likely weathering state of the oil were provided by the oil spill model OSIS. This model is used as the operational response model by MPCU. OSIS has been developed by AEA Technology (NETCEN) and British Maritime Technology (Leech & Walker 1992) .
As an example of the interplay between modelling and remote sensing model runs, on the evening of 15 and early 16 February OSIS predicted that the oil released would move SE towards Linney Head and beach at this location under the prevailing westerly wind at 18-20 kt (Fig 1) . As the spill was close to shore where local oceanographic features can have an important impact on the trajectory of the spill, it was important to obtain early confirmation of the slick trajectory using MPCU's remote sensing aircraft. The predicted trajectory was confirmed by side-looking airborne radar (SLAR) imagery at 23:02 on 15 February (Fig.2) and from daylight reports on oil beaching at 07:00 on 16 February.
This combination of remote sensing and predictive modelling was used throughout the incident to help to plan response operations. The incident demonstrated that in the case where a major spill occurs and pollution authorities are immediately notified then existing techniques of remote sensing and sea trial validated predictive modelling (i.e.OSIS) provide the information necessary to make operational decisions. One of the lessons learnt from the SEA EMPRESS spill is that the detailed analysis and dissemination of this remote sensing and modelling information requires a dedicated team as part of the response operation. This team will be mobilised at future major oil spills in the UK.
The strategy used in the UK for applying dispersant is for the remote sensing planes to direct the spray aircraft to the areas of thickest oil and for the spray aircraft to repeatedly pass over the region of thickest oil until the surface oil has been dispersed. As expected, the dispersants were most effective on the fresh oil emerging from the grounded tanker. Therefore, the MPCU strategy for dispersant application was, in the first instance, to
