The EUSeaMap portal has continued to be maintained online. In the maintenance phase some modifications have been made to the webGIS service itself (download options, metadata, optimisation), whilst improvements have now also been made to the maps and methods behind them for the North and Celtic Sea areas.
In the North and Celtic Seas, improvements to bathymetry data have allowed revised versions of the input layers for light percentage at the seabed and wavebase to be used in a re-run of the seabed habitat model. Corresponding threshold analysis has been conducted to review the boundaries applied to delimit the infralittoral and circalittoral zones, and these new methods could form the basis for future thresholds work. The thresholds analysis is critical to the creation of these types of habitat maps as it is the part of the process that integrates biological data with the physical parameters that are used to construct the final habitat maps. They also have a bearing on the confidence layers that accompany the maps. The quality of broad-scale habitat maps such as these that combine abiotic variables following the EUNIS system are heavily reliant on the interpretation of the thresholds between classes in the the classification system, and whenever improvements become available in either the input data layers or available habitat validation data, it is highly recommended that the relevant thresholds are re-analysed.
An improved substrata layer produced by the EMODnet Geology group was also incorporated. Future developments in sediment mapping that allow for dynamic classification (allowing users to specify sediment profiles according to different classifications such as Folk and Wentworth) could open up possibilities to test the thresholds for sediment classes as they relate to EUNIS habitats using similar methods to those used for the other input abiotic variables.
Biotope tagged (or simply classified habitat) samples are still relatively scarce in offshore waters across European waters as a whole, meaning there is a limited ability to undertake stastically robust threshold analysis for deeper waters. Within UK waters the situation is, by comparison, very good, with a large database (in the form of Marine Recorder) to draw on as a resource for threshold analysis, and there are further fine-scale seabed habitat maps that have been drawn together under the MESH project previously for North Atlantic waters. However, even in this case, there is still a bias towards near-shore areas, and a lack of consistency amongst classifying habitats based on the abiotic variables recorded (if any). As an example, when testing wavebase, the number of available samples for habitats within the circalittoral and deep circalittoral zone (that wavebase is used to define the boundary between) is only 165 when only sediment biotopes classified to EUNIS level 5 and 6 are used, i.e. with biological data also recorded, were available from Marine Recorder. Even when looking at samples that include level 3 and 4 classified habitats (i.e. using physical habitat data as well as biological), the sample size is only taken up to ~1000. To improve the circalittoral/deep circalittoral boundary in future iterations of the broad-scale habitat map, many more samples would be needed, ideally to the biotope level (EUNIS level 5 and 6). Ideally there would be some measurements of energy conditions taken with habitat samples as well to gain more in-situ measurements with which to calibrate the data layers, and improve threshold analysis. Further analysis would also be useful to look at seasonal variation, i.e. to further examine whether in the case of defining the circalittoral/deep circalittoral boundary it is extreme or average conditions that are the guiding constraint.
One of the intentions at the start of the maintenance phase was to join the Baltic and North/Celtic Seas models, as there would be advantages to dealing with as few boundary issues as possible. However, computational limitations were soon reached by the higher resolution datasets that were available over such large basin areas. Other software might be feasible but the model woudn’t be as accessible as it currently is. Outer (seaward) extents are generally constrained only by the limit of our focus, whereas inner (shoreline) extent is much more limited by the technical capability in terms of modelling/mapping our input layers. For most data, the inner coastal areas, the ‘white ribbon’ of missing data, is an ongoing limitation for mapping, especially when generating full coverage broad-scale layers. Ultimately, the overall objective of the EMODnet is to assemble fragmented and inaccessible marine data into interoperable, contiguous and publicly available data streams for complete maritime basins. The maps produced by the EUSeaMap project represent a realisation of this objective. Improvements to the input data not only improve the resolution of the habitat models, but also offer the opportunity to refine the thresholds that relate these data to habitats. With increased amounts of in-situ habitat data and improvements to full coverage data layers there will be a steady narrowing of the gap between broad-scale and fine-scale maps.