Next Article in Journal
Laboratory Measurements of Subsurface Spatial Moisture Content by Ground-Penetrating Radar (GPR) Diffraction and Reflection Imaging of Agricultural Soils
Previous Article in Journal
Nationwide Projection of Rice Yield Using a Crop Model Integrated with Geostationary Satellite Imagery: A Case Study in South Korea
Previous Article in Special Issue
Improved Detection of Tiny Macroalgae Patches in Korea Bay and Gyeonggi Bay by Modification of Floating Algae Index
Open AccessArticle

Modelling the Vertical Distribution of Phytoplankton Biomass in the Mediterranean Sea from Satellite Data: A Neural Network Approach

1
Istituto di Scienze Marine, Consiglio Nazionale delle Ricerche (ISMAR-CNR), 00133 Rome, Italy
2
Istituto di Scienze Marine, Consiglio Nazionale delle Ricerche (ISMAR-CNR), 30122 Venice, Italy
3
Dipartimento di Biologia, Università di Roma “Tor Vergata”, 00133 Rome, Italy
4
Consorzio Nazionale Interuniversitario Per Le Scienze Del Mare (CoNISMa), 00196 Rome, Italy
5
Agenzia nazionale per le nuove tecnologie, l’energia e lo sviluppo economico sostenibile (ENEA), Centro Ricerche Frascati, 00044 Frascati, Italy
6
Istituto di Scienze dell’Atmosfera e del Clima, Consiglio Nazionale delle Ricerche (ISAC-CNR), 00133 Rome, Italy
*
Author to whom correspondence should be addressed.
Remote Sens. 2018, 10(10), 1666; https://doi.org/10.3390/rs10101666
Received: 30 September 2018 / Revised: 9 October 2018 / Accepted: 14 October 2018 / Published: 21 October 2018
(This article belongs to the Special Issue Remote Sensing of Ocean Colour)
Knowledge of the vertical structure of the bio-chemical properties of the ocean is crucial for the estimation of primary production, phytoplankton distribution, and biological modelling. The vertical profiles of chlorophyll-a (Chla) are available via in situ measurements that are usually quite rare and not uniformly distributed in space and time. Therefore, obtaining estimates of the vertical profile of the Chla field from surface observations is a new challenge. In this study, we employed an Artificial Neural Network (ANN) to reconstruct the 3-Dimensional (3D) Chla field in the Mediterranean Sea from surface satellite estimates. This technique is able to reproduce the highly nonlinear nature of the relationship between different input variables. A large in situ dataset of temperature and Chla calibrated fluorescence profiles, covering almost all Mediterranean Sea seasonal conditions, was used for the training and test of the network. To separate sources of errors due to surface Chla and temperature satellite estimates, from errors due to the ANN itself, the method was first applied using in situ surface data and then using satellite data. In both cases, the validation against in situ observations shows comparable statistical results with respect to the training, highlighting the feasibility of applying an ANN to infer the vertical Chla field from surface in situ and satellite estimates. We also analyzed the usefulness of our approach to resolve the Chla prediction at small temporal scales (e.g., day) by comparing it with the most widely used Mediterranean climatology (MEDATLAS). The results demonstrated that, generally, our method is able to reproduce the most reliable profile of Chla from synoptical satellite observations, thus resolving finer spatial and temporal scales with respect to climatology, which can be crucial for several marine applications. We demonstrated that our 3D reconstructed Chla field could represent a valid alternative to overcome the absence or discontinuity of in situ sampling. View Full-Text
Keywords: ocean colour; SST; chlorophyll-a vertical profile; Mediterranean Sea; Artificial Neural Network; error-backpropagation ocean colour; SST; chlorophyll-a vertical profile; Mediterranean Sea; Artificial Neural Network; error-backpropagation
Show Figures

Graphical abstract

MDPI and ACS Style

Sammartino, M.; Marullo, S.; Santoleri, R.; Scardi, M. Modelling the Vertical Distribution of Phytoplankton Biomass in the Mediterranean Sea from Satellite Data: A Neural Network Approach. Remote Sens. 2018, 10, 1666.

Show more citation formats Show less citations formats
Note that from the first issue of 2016, MDPI journals use article numbers instead of page numbers. See further details here.

Article Access Map by Country/Region

1
Search more from Scilit
 
Search
Back to TopTop