Observations requirements for marine litter concentration characterization in the Mediterranean Sea.

Abstract

The main problem of characterizing the spatial distribution and variability of the marine litter (ML) in seas and oceans is the scarcity of observations. ML sampling campaigns are usually concentrated near coastal regions and are carried out during spring – summer months, when the navigation conditions are more favorable. As a result, the spatial and temporal resolution of the observations doesn’t allow a statistically robust description of the ML distribution and time evolution. Considering the limited resources and the high cost of the observation campaigns, developing an optimized sampling strategy is a key step to capitalize resources and obtain a robust ML characterization. This study analyzes the temporal and spatial requirements that a sampling should fulfill to obtain accurate estimates of ML concentration in different areas of the Mediterranean Sea. Provided that there are not enough observations to define the underlying statistics of ML concentration we use the outputs of the realistic numerical model as a synthetic reality. Then, we conduct several Monte Carlo experiments simulating different sampling strategies on the model data to obtain the mean ML concentration in a certain region. The spread of values from the ensemble of Monte Carlo members will be considered as the uncertainty associated to the estimated mean. Our results suggest that for the same number of observations (i.e. the same observational effort), is better to maintain long observational records rather than to intensify the sampling (i.e. reducing the sampling interval). If the spatial distribution of ML is aimed at, the required spatial density of the sampling depends on the characteristic correlation length scale. Therefore, those regions where the ML concentration structures are larger would require less dense observational samplings.