Short Course on Marine Data Literacy

• University of Cádiz (Laura Howard)
• University of Malta (Godfrey Baldacchino)
• University of Kiel (Ralph Schneider)
• Université de Bretagne Occidentale in Brest (Yves-Marie Paulet)
• University of Gdańsk (Arnold Kłonczyński)
• University of Split (Đurđica Miletić)

• University of Gdańsk (Aleksandra Cupiał)
• University of Split (Hrvoje Kalinić)

• University of Malta (Rami Kalfouni)
• University of Kiel (Stella Buchwald)

Project proposal presentations

Project 1

• Jesús Gómez-Enri (University of Cadiz, Puerto Real, Spain)
• Alfredo Izquierdo González (University of Cadiz, Puerto Real, Spain)

Project 2

• Maciej Markowski (Institute of Geography, University of Gdansk, Poland)
• Anna Panasiuk (Institute of Oceanography, University of Gdansk, Poland)
• Marcin Paszkuta (Institute of Oceanography, University of Gdansk, Poland)

Project 3

• Frano Matić (Institute of Oceanography and Fisheries, Split, Croatia)
• Jadranka Šepić (Faculty of Science, University of Split, Split, Croatia)

Project 4

• Tomás Fernández Montblanc (Earth Sciences Department, University of Cádiz)
• Alfredo Izquierdo González (University of Cadiz, Puerto Real, Spain)

• Why do we need marine data?
• How do we measure and model the sea?
• Different types of marine data: physical, biogeochemical, ecological; models; observations (in situ + remote)
• What is operational oceanography?
• Technologies used to produce marine data
• Temporal and spatial constraints in data collection and interpretation; how to measure and interpret data avoiding pitfalls
• The value addition chain of data; downstream services and data product delivery

• Met-ocean data: types and characteristics
• Types of products PHYS CMEMS
• Types of products INSITU CMEMS
• "Data lexicon": Quality information document and product user manual

• Introduction to IODE and UN Ocean Decade and its data chapter
• Introduction to the FAIR Guiding Principles for scientific data management and stewardship
• Best practices and standards for management and analysis of marine data
• Introduction to EMODnet

• What oceanographic data is freely available? How can this be accessed?
• Demonstration of professional online data interfaces to visualise in near-real-time ocean data products
• Portals from where numeric data derived from in-situ measurements, remote sensing, and forecasting models, can be downloaded
• Demonstration of visualisation software including Panoply, the Sentinel Application Toolbox (SNAP), and QGIS, that can be used to process downloaded data
• Simple data processing techniques to added-value products (such as colour composites, vegetation and water quality indices, etc…) 

• Brief introduction to web scraping and API
• Brief introduction to data wrangling
• Interacting with data servers part 1 (openDAP)
• Introduction to cloud storage and cloud computing
• Interacting with data servers part 2 (Zarr over HTTP)
• Introduction to Pangeo, Binder, Google Colab

• Data source for the blue ocean (SST, sea level, waves, etc.)
• Data source of the green ocean (chlorophyll-a concentration)
• First steps with Sentinel Data Hub for data downloading
• First steps with Sentinel Application Platform (SNAP)

• Introduction
• What is Artificial Intelligence? Demystifying the hype surrounding AI
• AI needs huge amounts of data to operate, and oceanographic data is readily available
• Overview of different AI applications used in oceanography

• IT tools and systems
• Relational Database Management Systems
  • Data model
  • Basic properties of transactions in RDMS
• Geo-Information systems
  • Spatial, temporal, and spatio-temporal data models
  • NoSQL Databases
  • Data models
  • Cap-theorem (consistency and availability)

• Data Preprocessing
• Overview on Distributed Data Processing
  • Hadoop, map-reduce
  • Spark
  • Flink

• What is unsupervised learning?
• Distinction between clustering and classification
• Cluster properties and basic approaches to cluster analysis
• Examples of some algorithm and applications
• What is supervised learning?
• Distinction between classification and regression problem
• Learning algorithms and why are the introduced
• Learning as an optimization problem. Gradient descent. Local optimum
• Linear separability and complexity of neural network architecture
• Simple (shallow) neural network implementation

• Introduction - Aldo Drago
• Anaconda and packages - Adam Gauci & Audrey Zammit
• Ocean Data View - Julia Gottschalk & Sally Close
• SNAP - Jesús Gómez-Enri
• MATLAB - Tomás Fernández Montblanc

• Log in the Marine Copernicus Portal.
• Download daily averaged sea surface temperature (SST) data from a high-resolution model that is available for ten consecutive days.
• Download satellite observed SST data (L4) generated over the same domain and ten-day period.
• Load and visualise the model and satellite netcdf files in Panoply.
• Identify a grid cell in the model and satellite datasets, and extract the SST time series over the ten days being considered.
• Generate a time series plot that overlays the SST predicted by the model and observed by the satellite over the ten-day period being considered.
• Generate a scatter plot to visualize the correlation between the model and the satellite.
• Compute the Mean Square Error between the model and satellite SST.

• Students will be introduced to various open access sea-level data repositories and given information on how to choose the best data for their research
• Students will download sea-level data from the selected repository
• Students will plot the sea-level data and visually assess it
• Students will then be familiarized with different ocean processes detectable in sea-level time series and learn how to recognize and extract these processes by filtering procedures
• Students will be introduced to basic statistical parameters, the normal distribution, Student's t-test and linear regression
• Students will estimate stationarity and trends of selected component of downloaded sea-level time series and asses the significance of estimated trends

• How is primary production measured?
• How is primary production modelled?
• How to optimize model performance and extract parameter values from data?
• How to fill data gaps using models?