Posts

Our 2020 in Review

/in /by

2020 was an out-of-the ordinary year but, despite the unusual challenges, here at Progressive we were able to achieve important milestones made possible by the commitment and dedication of our Team and our Partners. Such projects have laid the foundations for exciting initiatives coming up in 2021, which we are looking forward to sharing soon.

These are the key highlights of our 2020:

January and February

We continued assisting the growing community of Earth Observation Altimetry data enthusiasts, in the frame of the ESA Research and Service Support – RSS, the ESA service we operate since 2006 dedicated to supporting the Earth Observation (EO) community in exploiting EO data.

We kicked-off 2020 by putting together several processing services for Sentinel-3 and CryoSat-2 developed over the years by different PIs, under a unified umbrella: the SARvatore Services (SAR Versatile Altimetric TOolkit for Research and Exploitation) family. This collection of services enabled more than 200 users to get Earth surface heights over ocean, inland water bodies, rivers or ice sheets.

Figure 1 Fully Focused, Delay-Doppler and Conventional SAR Altimetry acquisitions.

March and April

Together with our partners Solenix, Qualteh JR, Terrasigna and GISAT, we released an updated version of the ESA Copernicus Sentinel App, the gateway to knowing the Copernicus Sentinel satellites. The App enables users to track the Sentinel satellite in orbit, search for products in a specific point over the 3D globe, visualize the latest product acquired from the satellite and save the searches. The new version is enriched with the possibility to browse Sentinel-5P products.

Figure 2 Copernicus Sentinel App – Acquisition of products feature over a specific area of the 3D globe.

Additionally, during the same months, we furthered our work within the Sentinel-3 Mission Performance Centre by contributing to increase the level of quality and maturity of the new Copernicus S3 Fire Radiative Power (FRP) product derived from S3 Sea and Land Surface Temperature Radiometer (SLSTR). We supported the development of a new algorithm for the validation of S3 FRP SLSTR with the objective to measure with higher accuracy the radiative power of land and ocean hotspots (wildfires, volcanoes, agricultural burning etc.) over an area size of 1 km2 on our planet.

May and June

We successfully run the 4th edition of the EODA (Earth Observation Data Analysis) Lab in the frame of the ESA Research and Service Support. Through the EODA Lab we opened a window on the sensors, characteristics, and main applications of the Copernicus Sentinel 1-2-3 missions to aspiring data scientists attending the EODA course taught by prof. Marzano within the Master Degree in Data Science of Sapienza University of Rome.

Figure 3 Postseismic ground displacement map following the Amatrice earthquake (Central Italy, 2016) estimated using SAR Differential Interferometry technique applied to Sentinel-1 SAR IW SLC data.
Figure 4 Mediterranean Sea Surface Temperature Composite. Contains Copernicus Sentinel-3 Data (2020).
Figure 5 Supervised classification over the area of Lake Vico with Sentinel-2 images.

Alongside our activities in the education field, the work we performed in the two following projects has been featured in two new publications.

Part of the work carried out in the frame of the ESA funded project AIDEO (AI and EO Data Innovative Methods for Monitoring West Nile Disease Spread in Italy) together with IZSAM, Istituto Zooprofilattico Sperimentale dell’Abruzzo e del Molise “Giuseppe Caporale, AImage Lab (University of Modena and Reggio Emilia) and ReMedia resulted in a research paper titled [Vincenzi, S., et al. 2020] “The color out of space: learning self-supervised representations for Earth Observation imagery” accepted at the 25th International Conference on PATTERN RECOGNITION (ICPR 2020). The publication provided new insights into the development of effective deep learning techniques for Remote Sensing, proposing for the first time a novel representation learning procedure on a colorization-based method which allows users to overcome the lack of large annotated datasets for satellite images.

The second one [Cigna, F., et al. 2020] “Supporting Recovery After 2016 Hurricane Matthew In Haiti With Big Sar Data Processing In The Geohazards Exploitation Platform (GEP)”, features a work led by the Italian Space Agency aimed at investigating the generation of experimental scientific products of change detection and ground deformation monitoring that Haitian end-users can exploit to support decision-making process and recovery progress after the Hurricane that hit Haiti back in 2016. More specifically, the project exploited two processing services available in the ESA’s Geohazards Exploitation Platform (GEP) that were integrated by us in the context of the RSS service support provisioning.

At the same time the RSS team collaborated with La Sapienza University of Rome in the research project SMIVIA – Snow-mantle Modeling, Inversion and Validation using multi-frequency multimission InSAR in central Apennine addressing snow-deposits monitoring using multi-frequency multi-mission InSAR in central Apennine, by supporting the research team with the automation of the InSAR processing chain and with elaborations and expert technical assistance for the use of the RSS CloudToolbox service.

July and August

The first of August marked an important milestone for the ESA Research and Service Support service which entered into an extension phase until Summer 2021. The tasks we have been in charge so far will remain unchanged during this extension phase. We will continue to be responsible for the service operation and for providing research support by facilitating Earth Observation data exploitation, the development of new algorithms and the generations of value-added processing results.

In these months, we also laid important groundwork for three new Earth Observation tools which we plan to deliver as operational services in 2021:

  • Integration of ACOLITE (Atmospheric Correction for Operational Land Imager) in the processing environment, allowing for a simple and fast processing of coastal and inland water applications;
  • Development of CoDeMinT (Coastline Detection and Monitoring Tool) extracting and analysing shorelines from Landsat and Sentinel-2 imagery;
  • the Cloudy Earth Observation Time Series Restoration Tool, a working implementation of the algorithm proposed in [Bertoluzza et al. 2019] to restore the missing pixels of multispectral imagery time series acquisitions covered by clouds.
Figure 6 Chl-a concentration (left) Suspended Particulate Matter – SPM (right) from a Sentinel-2 image over Venetian Lagoon processed by ACOLITE in G-POD.
Figure 7 Intersections between user-defined transects and detected shorelines from Sentinel-2 images. Contains Copernicus Data [2017, 2020]
Figure 8 Original Image (left); Original Image with masked clouds (centre); Reconstructed Image (right)

September and October

Our team participated in the digital edition of the ESA EO Phi-Week, the ESA event dedicated to the latest achievements in Earth Observation science, technology and applications. We showcased the tools, services and projects derived from some of our latest research and engineering activities:

In the same period, the ESA/ESRIN contract “Maintenance and Evolution of the Copernicus Apps for Mobile Devices” was successfully completed. The contract, started in 2017, has rebranded and upgraded the Copernicus Sentinel App which became one of the most successful ESA mobile apps as of today, with 16 different app versions delivered over the years. In parallel to the evolution of the Copernicus Sentinel App, two new prototypes were also developed within the contract: the Copernicus Eye and the Copernicus Dashboard. The former visualizes information on land, atmospheric and ocean variables taken by The Copernicus Services. The latter is a web-base version of Copernicus Sentinel App. We played an important role as Customer’s interface and in the software independent testing and validation activities to ensure a proper functioning of the apps. The maintenance of the current features of the Copernicus Sentinel App is guaranteed, therefore it is still possible to download the most recent app release on the App Store and Google Play. Its newest features allow users to track the different Copernicus Sentinels directly from their ground position, browse new products for Sentinel-3 and Sentinel-5P through the search feature and download the different Sentinel products on the computer with the Sentinel Download Manager.

November and December

Figure 9 Graphical Abstract of the solution/method adopted in [Candeloro, L., et al. 2020]

The results of our collaboration with IZSAM, AImage Lab and ReMedia within the AIDEO Project were presented in a final review meeting with ESA. Combining past West Nile Virus occurrences, Earth Observation data and Machine Learning algorithms, the project aimed at putting in place a model to predict areas at risk of West Nile Disease circulation two weeks in advance.
The results were promising and represented an important first step towards the development of an early warning system that could support public authorities in better targeting surveillance measures for the prevention of West Nile Disease. Our contribution was focused on the generation and provisioning of Analysis-Ready-Data tailored to the artificial intelligence requirements, which considerably facilitated EO data exploitation and value-added information retrieval.
A conclusive AIDEO paper [Candeloro, L., et al. 2020] “Predicting WNV Circulation in Italy Using Earth Observation Data and Extreme Gradient Boosting Model” was also published in the MDPI Remote Sensing journal and is available here.

In the same months, the first phase of the validation activities for the new S3 SLSTR FRP nighttime product has been successfully concluded in collaboration with the S3 SLSTR Mission Performance Centre experts, and its results have been included in an interactive Web Application. The project officially entered its second phase: the development of the daytime algorithm is currently ongoing, the validation algorithm is being revised and updated accordingly, and the web application is being upgraded.

In the meantime, the RSS altrimetry processing services offer continued to expand improving accuracy and spatial resolution. The SARvatore service in G-POD now offers the possibility to include in output an additional product obtained with ALES+ SAR retracker, a waveform retracker for open ocean and coastal zone SAR altimetry data. The family of SARvatore services was enriched with a Fully Focused SAR altimetry algorithm for CryoSat-2 (FFSAR, developed by Aresys), which allows for a consistent improvement of the along-track spatial resolution (from about 300 m to order of meters), and the Sentinel-6 High Resolution L1A Ground Processor Prototype (GPP) processor.

Figure 10 Comparison between S6 low resolution and high resolution acquisition modes.

Over the last year the entire Progressive Systems team joined forces to bring to light a new platform for the analysis of Earth Observation Data which will be released in 2021 and that will incorporate some of the tools and competencies we have developed in 2020. This is Earth Console, a platform bringing Earth Observation data, algorithm development, integration and processing services, analysis and visualization tools, powerful computing resources and high-speed network connections, all in one place. A comprehensive service characterized by a great degree of flexibility allowing its users to exploit satellite data at diverse extent depending on their needs, from developing their own algorithm, to accessing specific know-how to scale up their application to massive data processing or wide service exposure.

Stay tuned and we really look forward to an even more fruitful 2021!

The views expressed herein can in no way be taken to reflect the official opinion of the European Space Agency, the European Union or any other space agency or entity mentioned in the text.

Figure in the cover image:
– Reconstructed Image using the Cloudy EO Time Series Restoration Tool
– Suspended Particulate Matter – SPM (right) from a Sentinel-2 image over Venetian Lagoon processed by ACOLITE in G-POD.
– Shapefiles of the coastline for different acquisition times obtained with CoDeMinT.
– Path followed by 2016 Hurricane Matthew and detailed satellite view before it struck southwestern Haiti (credit NOAA/ NHC – NASA);
– Comparison between S6 low resolution and high resolution acquisition modes (RSS altimetry processing service)
Earth Console Monitoring Dashboard
– Tracking from Ground feature of the Copernicus Sentinel App

Progressive meets Society – Interview with Frank S. Marzano

/in /by

If “Data science is the sexiest job in the 21st century” as often mentioned, then we may say that “Earth Observation is certainly the funniest”

FRANK S. MARZANO

The demand for professional figures able to manage and interpret the growing availability and complexity of data has been increasing rapidly with the advent of new technologies. Sapienza University of Rome was the first Italian University and one among the first in Europe to address this labour market need with the establishment of the Laurea Magistrale in Data Science back in 2015. The Master degree holds a direct link with the space domain, having enriched its offer with the Earth Observation Data Analysis Lab organized in collaboration with the ESA Research and Service Support service.

We have asked Prof. Frank S. Marzano, professor from the Department of Engineering at Sapienza University of Rome, how the Master Degree was able to attract a growing participation of students year after year and in which way the link with the Earth Observation domain contributed to such a success.

The Sapienza University of Rome was the first Italian university and one of the first in Europe to launch a Master’s Degree in Data Science back in 2015, can you tell us more about the work you do in this regard?

The Laurea Magistrale in Data Science at Sapienza University of Rome was one of the first in Europe, but the first in Italy. I was contributing to enlarge the offer to include Earth Observation (EO) into Data Science courses and students background.

Why was a Master’s Degree in Data Science needed back then?

The need of MSc in Data Science was due to the remarkable increase in the volume and complexity of available data and new technologies that have been developed. Processing them requires a combined multi-disciplinary approach to design an overall strategy aimed at transforming data into useful information. Key ingredients to develop a successful strategy are data manipulation and visualization, large scale computing, statistical modelling, learning techniques and algorithmic thinking.

How did the educational program evolve over time to keep pace with the technological advances in the field of data science?

The Laurea Magistrale in Data Science is a Master degree taught in English. It is a joint initiative within the i3S Faculty combining the expertise of four Departments:

  • Department of Computer Science (DI)
  • Department of Computer, Control and Management Engineering (DIAG)
  • Information Engineering, Electronics and Telecommunications (DIET)
  • Statistics (DSS)

This Master program provides a solid and modern preparation to understand and manage the multi-facet aspects of carrying out a complete data analysis, including acquisition, management, and statistical analysis. Its educational program benefits from this inter-department and inter-disciplinary approach to keep pace with the scientific and technological innovation.

The Master’s Degree is currently in its fifth edition and it has been receiving a very positive response with the number of students doubling year after year. In your opinion, which is the reason behind such a success?

The success is probably due to the innovative approach of this Master’s program in Data Science aimed at mixing all the necessary ingredients for a successful learning: a solid multi-disciplinary theoretical background combined with a frequent use of laboratory activity and special emphasis on developing a final data-science thesis project. The program is taught in English to attract the motivated students from everywhere and help them develop the necessary ability to interact in an international multidisciplinary environment. It is a 2-year, 120 ECTS program ending with the development and discussion of a final thesis project.

What kind of role partnerships with Earth Observation experts played in providing a real hands-on experience with the space domain of applications to students?

The role of the Earth Observation experts has been and is essential as they can provide a real hands-on experience to students showing the most updated tools, such as the Sentinel Application Platform (SNAP) platform, as well as using Sentinel data for a variety of applications. In my Earth Observation (EO) Data Analysis course the partnership with the ESA Research and Service Support (RSS) group in ESRIN was greatly appreciated by all students and is probably one of the reasons for the increase of the number of EO students from 5 to 30 in only 3 years.

With a look to the future, what jobs will be the most in-demand in the field of Earth Observation according to you?

Most appealing and requested EO jobs will be probably related to EO big-data analysis both in the midstream and downstream domain. This means that professionals should be able to develop new retrieval algorithms and to understand the physical modelling behind as well as to be capable to apply new machine learning techniques and set up robust data pipelines for data processing.

More generally, where do you see the society taking the most benefit from Earth Observation in 5/10 years from now?

Earth system monitoring applications as well as security, civil protection, urban planning and agricultural production will be the most explored domains. EO data support to governmental policy makers will be probably more and more requested.

For a close, is there anything else you would like to add?

In conclusion, if “Data science is the sexiest job in the 21st century” as often mentioned, then we may say that “EO is certainly the funniest”.

————————————————————————-

To learn more about Frank S. Marzano

Prof. Frank S. Marzano received the Laurea degree (cum laude) in Electrical Engineering (1988) and the Ph.D. degree (1993) in Applied Electromagnetics both from the Sapienza University of Rome, Italy. During 1993 he collaborated with the Institute of Atmospheric Physics, National Council of Research (CNR), Rome, Italy. After being a lecturer at the University of Perugia, Italy, in 1997 he joined the Department of Electrical Engineering, University of L’Aquila, Italy teaching courses on electromagnetic fields as Assistant Professor. In 2002 he got the qualification to Associate Professorship and co-founded Center of Excellence on Remote Sensing and Hydro-Meteorological Modeling (CETEMPS), L’Aquila. In 2005 he finally joined the Dept. of Information engineering, Electronics and Telecommunications (DIET), Sapienza Univ. of Rome, Italy where he presently is a full professor and teaches courses on antennas, propagation and remote sensing. Since 2013 he is the director of Centre of Excellence CETEMPS of the University of L’Aquila, Italy. His current research concerns passive and active remote sensing of the atmosphere from ground-based, airborne, and space-borne platforms and electromagnetic propagation studies. Prof. Marzano has published more than 150 papers on refereed International Journals, more than 30 contributions to international Book chapters and more than 300 extended abstract on international and national congress proceedings. Since 2014 he is Associate Editor of IEEE Transactions on Geoscience and Remote Sensing (TGRS) as well as the journal EGU Atmospheric Measurements Techniques. Dr. Marzano is Fellow of RMetS (Royal Meteorological Society) since 2012 and Fellow of IEEE since 2015.