Abstract

The Water Framework Directive (WFD) is the most substantial piece of water legislation ever produced by the European Commission, and will provide the major driver for achieving sustainable management of water in Member States for many years to come. It requires that all inland and coastal waters within defined river basin districts must reach at least good status by 2015 and defines how this should be achieved through the establishment of environmental objectives and ecological targets for surface waters. 
 
The WFD has established a system of biological (phytoplankton, macroalgae, marine angiosperms and macroinvertaebrate) hydromorphological and physiochemical indicators that determine the ecological quality of coastal water bodies. This study is focused on the coastal monitoring network which analyzes phytoplankton biomass through chlorophyll a (a surrogate indicator), established in the scope of the WFD. The objective is to provide an ecological classification for Valencian coastal water bodies.

Since the coastal monitoring network was activated in August 2005, this study manages historical data for three years. Ecological classification quality is assessed using 90th percentile valules (chlorophyll a) and ecological class thresholds intercalibrated between the Member States of the Mediterranean Geographical Intercalibration Group (MED-GIG). 

The calculation of the 90th percentile value of a sample presents some difficulties since the empirical CDF is represented by a ladder chart, which results in undefined situations. There are different criteria for calculating sample percentiles in these cases. In fact, most statistical software provides us with different results since they use criteria that are similar but not equal. These differences are caused by the lack of a universally accepted procedure. Because of this, the possibility exists that the ecological classification is inadequate. It is suggested that this issue should be discussed during the intercalibration process for MED-GIG and that common criteria to be used by all member states should be established.

The results of the exploratory study for the temporal reduction show unpredicted consequences in the ecological quality classification for the Valencian coastal waters. Integrating pluviometric, wave, salinity and chlorophyll a series analyzed for the three different climate conditions that were found for the Valencian coastal waters, it was realized that the trophic balance of the ecosystems depends on anthropogenic influence and environmental factors that are determined by climate seasonality and irregularity. Three types of coastal water body ecosystems have been determined: stable, unstable and intermediate. The results demonstrate that survey frequency can be reduced for both stable and unstable systems; even though temporal reduction is only recommended for stable water body ecosystems.
Additionally, in the exploratory study of possible reductions in the number of stations visited, the comparative study of the 90th percentile has confirmed that it was possible to reduce the number of stations with no change in ecological classification.  

A methodology has been developed to select the optimal subset of sampling stations to be surveyed. The reason for this is the fact that the ecological balance of coastal waters is subject to strong instability, and a simple descriptive study would lose its usefulness if the characteristics of the area were to change. The proposed methodology could be used routinely by scientists.

The particular study of the empirical CDF of the data collected for each water body has confirmed that it is desirable that the empirical CDF in the full sample does not differ substantially from the empirical CDF in the subsample in order to attain the greatest similarity between the respective 90th percentiles. Also, it would be interesting that the subsample includes the greatest values of the original sample. The method for selecting the stations to be surveyed has been approached from the perspective of an inter-observer variability problem. Different k-combinations of sampling stations have been approached as different observers which provide monthly measurements of chlorophyll a concentration in the area. In order to compare the concordance between two different observers, two measures have been considered: Euclidean distance and intra-correlation coefficient. To ease the interpretation of the Euclidean distance between observers, multidimensional scaling has been used.

To detect possible changes in unchecked sampling stations, we suggest estimating the concentrations of chlorophyll a in these stations using regression models. The virtual values obtained this way should be verified with occasional revisions to detect possible changes.