Methodology

The objectives posed in this project will be accomplished using the following methodologies:

M1: Global scale monitoring of the distribution of dominant phytoplankton functional types (PFTs)

The PhytoDOAS method (Phytoplankton Differential Optical Absorption Spectroscopy) developed by Bracher et al. (2009) allows the determination of the biomass of different PFTs from satellite data basically independent from a priori information. The specific optical signatures of different PFTs will be retrieved by the differential optical absorption spectroscopy (DOAS) from spectral highly resolved data (i.e., spectral resolution better than 1 nm), collected using the hyperspectral instrument SCIAMACHY (Scanning Imaging Absorption Spectrometer for Atmospheric Cartography) on ENVISAT satellite.

Diagram of PhytoDOAS method

Climatologies from different periods will be obtained, allowing a better assessment of the potential link between global phytoplankton dynamics and some environmental phenomena (e. g., algal blooms and climate-related cycle processes). In addition to further research devoted to optimize the PhytoDOAS method, the HCA cluster method by Torrecilla et al. (2011) (see details in methodology M2) or other methods will be applied to SCIAMACHY satellite data containing information on phytoplankton composition resolved on a reasonable spatial (e.g. 1° latitude to 1° longitude) and temporal (monthly) resolution.

M2: Hierarchical Cluster Analysis (HCA) to identify different marine phytoplankton assemblages

The feasibility of classifying different ocean environments in terms of phytoplankton assemblages will be examined by applying an unsupervised hierarchical cluster analysis (HCA) developed by Torrecilla et al. (2011) to hyperspectral absorption and remote-sensing reflectance data. To evaluate the utility of hyperspectral optical data for discriminating phytoplankton pigment assemblages, the cluster trees obtained for the different spectral optical data are compared with a reference cluster tree obtained using the pigment composition data. For this analysis, an objective criterion of cluster similarity will be utilized: the cophenetic index (i.e., a measure of how precisely two cluster partitions preserve the pairwise distances between data objects).

A schematic diagram illustrating the general approach to hierarchical cluster analysis and similarity determination. The cluster tree obtained with pigment composition as the input (upper pathway) is used as the reference for comparison with results obtained utilizing various hyperspectral optical data
Classification of different stations in the Eastern Atlantic Ocean using HCA method based on phytoplankton absorption spectral data. Source: Taylor et al. (2011)

By applying and optimizing the HCA cluster method to several global and in situ local optical data sets will contribute to answer the following scientific questions: Is it possible to define biogeochemical provinces based on the phytoplankton community composition information? Which are the phytoplankton groups that have a higher impact on global scale processes and which are the ones sensible to global climatic changes? Is the same methodology valid to address global and local studies?

M3: Local characterization of coastal environments using autonomous platforms

Traditional approaches using ship-based measurements for observing coastal dynamics and episodic phenomena have proven to be ineffective given evolving biological state, the need to measure various properties across the spatial extent of such phenomena, and most of all in dealing with logistical details centered on manned ships on fixed schedules. More recently, Autonomous Underwater Vehicles (AUVs) have shown to be more cost-effective and have increased persistent presence. With a suitable sensor payload, those platforms have been able to systematically observe phenomena at requisite scales of variability of biogeochemical processes.

Autonomous platforms may provide a new approach to characterize processes where temporal and spatial variability represent an important challenge for field measurements

Different field campaigns will be performed using Autonomous Underwater Vehicles (AUVs) in order to obtain a 3D (surface and time) and a 4D (volume and time) characterization of the evolution of optical and physical parameters in Alfacs Bay. The goal of this pilot program to monitor coastal areas will be to measure various properties across a wide spatial extend, where algal proliferations are a recurrent phenomenon. Moreover, a challenge still to overcome is to improve the extending of the range and endurance of the AUVs when the investigated area exhibits strong, time varying currents or high temporal and spatial variability. The following questions will be addressed from these datasets: Can we infer the phytoplankton dynamics from those high resolution observations? Is the hyperspectral approach a valid method to characterize phytoplankton assemblages in coastal areas? Can we establish a semi-automatic early-warning system for harmful algal blooms (HABs) based on this type of measurements?

Preliminary results of AUV-based measurements in Alfacs Bay showing the high spatial variability in Temperature, Salinity and Fluorescence