2021 Annual Report

,

MET Norway is responsible for weather, ocean and climate services for Norway and Norwegian areas of interest, which means in particular over the oceans and polar areas.Our services are based on leading science in the field, and our capability to transform research and innovation into operational products.Research-based user-oriented services and a short route from research to operations are among our main assets.We are dependent on close collaboration with leading research communities, and hence rely on valuable partnerships through good projects.CIRFA is a very good example of this type of collaboration, as it focuses on important value chains and connects scientific experts directly with important users.
Operational meteorology and oceanography builds on the analysis of large amounts of observations, including the assimilation of these into numerical forecast models of weather, ocean and sea ice.Over the polar ocean the capability to utilize satellite data is crucial.In Tromsø we have leading expertise in remote sensing of ocean and sea ice represented by CIRFA and its partners, and MET Norway is lucky to be involved in ongoing research efforts.
In CIRFA one important focus has been the modeling of ocean and sea ice.The work has led to new operational data assimilation methods in our regional modeling systems: the "Barents" model, which covers the Barents Sea and the areas around Svalbard, and the "Norkyst" model, which covers the fjords and shelf seas of mainland Norway.New and better ocean surface analysis is important for both ocean and weather forecasting.The coupling between ocean and atmosphere is crucial, and CIRFA is in this context contributing to the development of the total forecast capability for the earth system.A very important application of the forecasting models is to deliver input to the contingency models (OpenDrift) which are used for e.g. oil spill monitoring and in Search and Rescue operations.In CIRFA an ensemble prediction system has been developed for the Barents model which is providing uncertainty estimates which are of great value in an operational context.
The data assimilation and modeling capabilities is one part of the value chain.The direct user contact and guidance through the weather forecast centers is another important part.In Tromsø the MET Norway weather forecast office VNN (Vaervarslinga for Nord-Norge) is hosting the polar meteorology service including the Ice services.Being a part of the CIRFA team developing automatic sea ice classification from satellite data is mutually very beneficial for the success of CIRFA and MET Norway.
These are valuable examples of the importance of broad scientific collaboration in general, and CIRFA in particular, for the operational weather and ocean services.In the last part of the project there will be a focus on the CIRFA legacy, which brings us to another important point.The good relations that are built within the CIRFA consortium provide a solid foundation for collaborative research with high societal relevance in many years to come.

Lars-Anders Breivik
Another year has passed.A second year of Covid-19 restrictions, social distancing, few physical meetings, and almost no travel.Scientific work often allows for flexible working arrangements, and if the technical prerequisites are fulfilled, working from home has some advantages.One of the obvious disadvantages is that collaboration between colleagues is hampered.One loses valuable discussions during coffee and lunch breaks, or in informal meetings in the corridor.Hence, it is most welcome that we are now gradually getting back to seeing our colleagues again.It is also much welcomed that visiting conferences and social events are allowed.For early carrier scientists, the lack of the ability to travel to conferences has been most unfortunate, as it takes away the inspiration that feedback and discussions with peers and international colleagues give.Despite the challenges, CIRFA has continued to contribute to the scientific and technological development in Arctic remote sensing and climate research.In 2021, four PhD`s have successfully defended their thesis: Cornelius Quigley, Johannes Lohse, Artem Moiseev, and Rolf-Ole Rydeng Jenssen.CIRFA-researchers authored 16 journal papers, and are registered with 39 new entries in CRISTIN, the national research information system of Norway.The Centre's scientists were involved in several research proposals, and three applications were successful, two funded by RCN and one by EU.
In August, Camilla Brekke started in the position as Pro-Rector for Research and Development at UiT the Arctic University of Norway.This is a major career step for her and a great recognition of her competence and capacity.Camilla was work package leader for WP3 and her leave from CIRFA of course is a big loss to the Centre.However, we are delighted on Camilla's behalf, and wish her all the best in the new role as Pro-Rector.
CIRFA has had good progress through the years, and many of the deliverables described in the initial proposal for the Centre can be "ticked off " as achieved.We have developed models and algorithms for mapping and forecasting of sea ice, icebergs, and ocean surface dynamics.As CIRFA is approaching the end of its time of operation, the work ahead will be more focused on operational implementation and validation of algorithms and models.In this regard, CIRFA is currently planning a research cruise with R/V Kronprins Haakon to Fram Strait, Greenland Sea.During the cruise, we will combine shipboard observations with satellite acquisitions and aerial surveys using drones.We will deploy drifters, buoys, moorings as well as attaching GPS trackers to icebergs and floes.Comparing our models with real-world observations and collecting new data will help us to assess and improve the quality of our classification and forecast products.
As an SFI, CIRFA has facilitated collaboration between the industry sector and academia and connected multidisciplinary research groups in a synergetic collaboration that has been quite successful.One of the original motivations when CIRFA started was the oil and gas companies' needs for increased monitoring and forecasting capacity in the high north, associated with their planned activities in the Barents Sea.Other motivations were directed towards environmental management and climate science.In years to come, it will be important to find ways to maintain the collaborations that have been established, to exploit the remote sensing technology which has been developed, and use it in new application areas.Two such new applications that already have been identified are in offshore wind power and environmental monitoring of coastal waters.There should be many more.
I finally would like to end by extending my sincere thanks to colleagues and partners in CIRFA for all work and achievements made during the year 2021.A special thank you to Andrea Schneider for structuring and editing this annual report so nicely.

Objectives and Visions in the Centre
CIRFA is developing remote sensing knowledge and technology that helps to make operations and activities in icy waters safe and environmentally friendly.The research tasks are centered around three application areas: CIRFA is a leading international research center on integrated remote sensing and forecasting for the Arctic.We focus on improving monitoring, understanding and forecasting of the geophysical processes at the ocean surface, such as surface currents, wind fields and sea ice.We also do research on remote sensing of oil spills in ice-covered waters.In addition, research is conducted on drone technology, validation of remote sensing data in the field, data assimilation methods and numerical weather forecasting.With an international team of researchers from academia and industry, the vision of the center has many aspects:

Research Highlights
Illustration of the training methodology of the Teacher-Student semi-supervised learning model for labelpropagation.Image from Khaleghian et al. 2021.Processes in the surface ocean and how they can be studied.The illustration was made by Gunnar Follesø (Follesø Animasjon).
Review Article on "Surface Currents in Operational Oceanography: Key Applications, Mechanisms, and Methods" A review article, published by the Journal of Operational Oceanography, combines remote sensing and modelling, and presents an overview of dynamics, observation methods and modelling of upper ocean surface currents.The article is an attempt to unify knowledge between various scientific communities, i.e. remote sensing, modellers, in-situ observationalists and theoreticians.In addition, applications for industries are discussed.By presenting a terminology for various aspects of surface currents, we aim to foster better exchange between these disciplines.Read more in Röhrs et al. 2021.A new Architecture for Deep Semi-Supervised Learning for Sea Ice Application Deep learning (DL) architectures refer to a subset of machine learning algorithms which are neural networks with three or more layers.These networks try to mimic the behavior of the human brain, they can learn from data, and are the foundation of many artificial intelligence applications.In the CIRFA context, these algorithms are highly interesting because they have proved efficient in classification and prediction tasks, and hence have been explored as a tool in sea ice applications.However, DL networks rely on lots of reliable training data to function properly.Generating lots of good quality training data for the sea ice case is challenging because of the difficulties of acquiring good ground truth information in the Arctic.
To address this problem, Khaleghian et al. have developed new semisupervised learning approach which can handle the situation where one wants to combine a small amount of reliable labeled data with a large amount of unlabeled data.The architecture consists of two 13-layer convolutional neural networks, a Teacher model, and a Student model.The Teacher model is trained in a two-step procedure.In the first step, the model is trained in a supervised manner using only the labeled data.Subsequently, both the labeled and unlabeled samples are fed to the trained Teacher model.Based on an evaluation of the samples' neighborhood relations and similarities in the feature space, pseudo-labels are generated for the unlabeled data through a label propagation procedure to train the Student model.Read more in Khaleghian et al. (2021).

Innovation Highlights
High resolution mapping of sea ice dynamics using drones As part of CIRFA, new methods and algorithms are developed for mapping and classifying sea ice using radar images recorded from satellites.The objective is to better understand the processes in the Arctic Ocean, but also provide operational data for marine operations in ice infested water.
High-resolution data, such as images captured from drones, is a valuable tool for understanding the radar backscatter recorded by the satellites and for providing local and real time information on sea ice dynamics.There are multiple operational and technical challenges with acquiring and processing high quality data using drones in the Arctic Ocean.
Covering large areas require a fixed wing drone capable of flying for several hours but operating from a ship deck usually requires a multirotor.
The sea ice is drifting fast, and standard methods for georeferencing images over larger areas does not work well for this application.Sharing high-resolution images from multiple cameras over long distances is not feasible in the high north, and for the near real-time requirements for operational data the images must be processed onboard the drone so that only the information regarding sea ice distribution and dynamics can be shared with the ship.
In CIRFA, work has been carried out developing robust technology for ship-based drone operations, accurate georeferencing of images for quantifying sea ice distribution and dynamics, path planning for mapping of drifting sea ice and onboard processing.With onboard data processing capabilities, the information regarding ice flow distribution and sea ice dynamics can be browsed in near real-time via a web interface on the ship.This system will be used during the CIRFA cruise in May 2022.

Left: A drone has been optimized for ship-based operations in the far north. With a flight time of 2 hours, carrying multiple cameras, the drone will be able to map an area close to 100 square km per flight. Right: The result is a map of ice floes (green) and open water areas (black) on a near real-time web interface. The map covers and area of 800m x 800m. Photos: Agnar Sievertsen, NORCE.
Photo by William Copeland

Deriving Oil Spill Thickness from SAR to help Resolving Slick Drift Patterns
Only a small fraction of detected oil slicks contains thick actionable oil, and by using the damping ratio these areas can be identified and targeted in a clean-up operation.The damping ratio is a well-known oil spill SAR parameter that can be used to infer relative oil thickness from SAR satellite images.The damping ratio is a calculated feature that measures the contrast between the oil slick and open ocean.
Much work has been carried out at CIRFA in collaboration with KSAT and NOFO (Norsk Oljevernforening For Operatørselskap / The Norwegian Clean Seas Association for Operating Companies) into making this parameter operational, and it is presently being implemented for the oil spill detection service at KSAT.At the same time, Cornelius Quigley is investigating an alternative robust approach for calculating the SAR damping ratio using verified oil slicks.The updated damping ratio will be used as input into the OilDrift model, where the first study initiating the OilDrift model based on varying thickness levels has been successfully carried out.This way information about the thickest parts of the oil can be incorporated into the forecast, thus focusing efforts on a potential clean-up scenario.This work was a successful collaborative effort between WP3, WP5 and WP7, and we expect further advances in 2022.

Surface oil seepages, as well as other natural phenomena, are shown in a Radarsat-2 image. Left, the damping ratio for the image is shown. Images processed by Cornelius Quigley. The damping ratio is used to derive oil spill shapes from satellite images and provide input to the OpenOil drift model. The damping ratio shape is seen as the solid dark red area. The Barents-2.5 km Ensemble Prediction System is used as forcing, and results from 6 different ensemble members are used. The output for the 6 different ensemble members are seen as different colours (dots). Image processed by
Victor de Aguiar, UiT.

Ocean Remote Sensing Background and Objectives
The ocean surface is the complex boundary between two very dynamic and stochastic media, the ocean and the atmosphere.
Better forecasting of the ocean state and improved understanding of the physical processes at the ocean/atmosphere interface require combined capacity in remote sensing, numerical modelling, and insitu observations.
The Norwegian coastal area is very industrialized and important for the economy.Ship traffic, fisheries, renewable energy, offshore oil/gas, and public recreation are all activities conducted in the coastal areas.Many users ask for more precise and higher resolution prediction of wind/wave/currents in the coastal areas to better support their activities.
The prediction skills of numerical models are still limited by uncertainties in the parameterisations used to represent the processes not resolved explicitly by the models.These unresolved scales must first be observed and measured, before approximate models can lead to efficient parameterisations.The newly launched Sentinel satellites will greatly improve the capabilities of providing such high-resolution information from space due to the enhanced temporal and spatial coverage.
This work package will develop the use of satellite technology to advance the understanding of the Arctic Ocean processes and dynamics, and contribute to better prediction of polar lows, now-casting, and short-range forecasting of ocean state through coupling with highresolution numerical models.

Key research tasks
Develop physical and statistical methods to improve the reliability of satellite-derived met-ocean parameters.Develop algorithms, products, and a processing system for providing ocean state parameters from satellite observations beyond what is achievable today.Perform extensive satellite product calibration and validation analysis using independent measurements and models.Study the dynamics of polar lows by combining remote sensing and numerical modeling.Support short range forecasting of ocean state through coupling with high-resolution numerical models in collaboration with WP 5.

Achievements in 2021
The R&D has been concentrated on the following three key activities: 1) Assessment of Sentinel-1 ocean coastal and global current measurements, 2) Assessment of Sentinel-1 data from polar low wind detection, 2) Operationalization of met-ocean processing system at Kongsberg Satellite Services (KSAT) based on achievements from CIRFA R&D.The research will contribute to better characterization of and possibly improve the modelling of the Norwegian Coastal Current and Polar Low events.

WP leader
Harald Johnsen Professor, NORCE One year of Sentinel-1 Doppler measurements acquired over Skagerrak/North-Sea in IW mode and globally in WV mode are calibrated using restituted attitude data.These data led to a journal paper recently accepted for publication in JGR -Ocean.These results also contribute to the preparations for the surface current assimilation experiment.The PhD study on polar lows has resulted in a publication in Remote Sensing.A third paper has evolved from the cooperation with Chalmers University on the comparison of ocean surface velocities derived from Sentinel-1 Doppler measurements and from TerraSAR-X ATI measurements.
The operational met-ocean processing system established at KSAT has been upgraded and extended to better support user and R&D needs.This includes the processing of wind field and ocean current field as for oil spill and ship detection services that KSAT supplies in near real-time to the European Maritime Survey Agency (EMSA) as well as BarentsWatch and Kystverket (the Norwegian Coastal Administration) to increase safety for Arctic maritime traffic.

Background and Objectives
Because of the thinning and retreating of sea ice in the Arctic, marine traffic and activities such as fishing, shipping, oil and mineral exploitation are intensifying and are expected to further increase in the future.To handle navigation challenges in icecovered waters, vessel crews need appropriate planning support.
The goal of this work package is to refine remote sensing methodologies and algorithms to characterize and map Arctic sea-ice conditions, and to provide improved methodologies for detecting icebergs.Requirements regarding the potential output and performance of algorithms are defined in collaboration with partners from operational ice services and industry.
Besides SAR images as key data source in this WP, also optical and thermal satellite images, passive microwave and altimeter data are employed.Studies focus on semi-and fully automated sea ice classification, on sea ice drift and deformation retrieval, on observing temporal and spatial variations of sea ice conditions, and on finding optimal algorithms and radar configurations for monitoring icebergs in open water and in sea ice.The methods used include multivariate statistical analysis, anomaly detection, CFAR (constant false alarm rate) approaches, and machine learning/deep learning techniques, supplemented by investigations of radar signatures of ice types and icebergs measured under different meteorological conditions.WP 2 and WP 3 are jointly working on issues regarding detection of thin ice as look-alike of oil spills and the use of compact polarimetry.Testing and validating algorithms takes place in collaboration with other work packages and MET Norway.The methodologies, tools and products developed within WP 2 will be integrated with the modelling activities of WP 5 to produce information products for the pilot services of WP 7.

Key research tasks
Development and improvement of (semi-)automated algorithms for sea ice type classification, using satellite SAR images at different frequencies and complementary satellite sensors, and testing them under operational conditions.Building a data base of optical and SAR image pairs for validating the algorithms' outputs.Determination of ice drift and deformation from sequences of SAR images in combination with ice classification and sea ice model simulations for short-term forecasting of ice conditions.Selection and improvement of algorithms for the detection of icebergs in open water and in sea ice, considering the detection rates at different radar frequencies and spatial resolutions.

Achievements in 2021
The Norwegian Sea Ice Service at MET in Tromsø has taken CIRFA algorithms one step closer to operation in 2021.We have routinely shared and updated our algorithms and demonstrated using such algorithms to automatically process and provide results to a ship for navigation and research assistance during a teaching cruise called UAK 2021 with the Norwegian coast guard vessel KV Svalbard.This collaboration had many technical challenges to test and solve.The CIRFA team is now well prepared for new versions and tests in the future.
More in detail, further progress was made on four main fronts: 1. Sea ice classification approaches extended to multifrequency and multi-sensor data 2. Sea ice concentration confidence being developed using our algorithms 3. Ice drift PhD in place and building CIRFA drift algorithm knowledge 4. Iceberg PhD in place and exploring false-alarm-rate algorithms and L-band iceberg detection The success of including incidence angle into the supervised sea ice classification algorithm for Sentinel-1 data has led to our ambition to extend this same approach to other sensors.CIRFA members have successfully explored transfer learning and re-training for RADARSAT-2 data (also a C-band system like Sentinel-1) and are now extending to TerraSAR-X (X-band) and ALOS-2-PALSAR-2 (L-band) data.Although re-training is generally required, the background algorithm is consistent.Our next step is to explore using multiple trainingsets, for the common algorithm and per sensor, to potentially overcome the temperature-dependent seasonal differences for year-round sea ice classification.
Another new method for sea ice classification using combinations of optical and SAR images, multi-frequency SAR images, and passive microwave and SAR data was developed.This method reveals a significant flexibility, i.e., it adapts to local environmental and imaging conditions by automatically selecting only those input parameters (radar intensity, texture) that are optimal for classification in the respective area.
After classification, we are also interested in the confidence of the class decision to provide good ice classification products and data input for modelling systems.This will have at least two components, an external error that can only be measured in comparison to ground-truth auxiliary data sets, and an internal error, or confidence, in the classification decision.This second internal component has been explored statistically and the error has been propagated to the numerical error, or uncertainty, in a sea ice concentration product and is ready for rigorous evaluation.
For the ESA L-C-Ice project, a new version of a coupled pyramid-cascade ice drift retrieval algorithm was implemented for providing short-term forecast of sea ice conditions using SAR imagery and forecasts of ice drift.This development is based on experiments with the CIRFA drift algorithm.
Our work on iceberg detection focusses now on the comparison of constant-false-alarm-rate algorithms that are applied to SAR images acquired at C-and L-band, considering iceberg occurrences in open water and sea ice.
Tests are based on the use of optical satellite images and supplemented by investigations of iceberg detectability dependent on radar frequency and polarization.In PALSAR-2 L-band satellite images, we observed ghost reflections next to the radar returns from icebergs (see Fig. 2-1).In 1991, a study reported the observation of ghost reflections in airborne L-band SAR images of icebergs that, however, were not observed in SEASAT satellite images.Our investigations have not been finished yet, but one plausible explanation is the occurrence of multiple reflections in the interior of the iceberg, considering that the penetration depths of radar waves at L-band are much larger than of the shorter waves at C-and X-band.In fact, we did not observe shifts and ghosts in C-and X-band images.

WORK PACKAGE 3
Oil Spill Remote sensing

Background and Objectives
In-situ data and fieldwork are of major importance for technology development for marine monitoring.The objective of this work package is to develop accurate remote sensing technologies for reliable oil slick detection and characterization, and to improve modeling of oil behavior and fate in sea icy waters.
Marine pollution is a significant problem in the World's oceans.As the sea ice in the Arctic retreats, maritime traffic and offshore operations are expected to increase.At the same time, the risk of oil spills increases.In addition, offshore oil and gas platforms are known to release produced water that still contains hydrocarbons despite treatment.During oil spill clean-up operations, the authorities need to know where thicker mineral oil is located, and where it may be moving due to wind and ocean currents.Radar remote sensing technology is a key solution to this problem, as it can see through clouds and operate during the polar night.By integrating remote sensing imagery and numerical drift modelling, we can predict where an oil slick is heading.
In the coastal areas, which in Norway are most often composed of long and narrow fjords, the seawater can be a mix of saline and more brackish water.In fjord sea ice with mixed salinities, oil behaves differently as in sea ice conditions.This may have an impact in case of a near-costal oil pollution incident.A natural extension of CIRFA`s research is to include marine primary production and water quality studies.

Achievements in 2021
Oil slick and produced water detection Cornelius Quigley successfully defended his PhD thesis about oil spill remote sensing using SAR images in March 2021 and presented his work at two international conferences.As a PostDoc, Cornelius continues improving oil spill thickness retrievals in collaboration with KSAT and other researchers in WP3.Cornelius' work will also be connected to newly appointed PhD student Victor de Aguiar's work about oil spill drift.Victor is working with produced water slick surface drift pattern estimates in collaboration with WP5.Master student Brynjar Andersen Saus contributed to produced water detection through his master's thesis work that also was presented to industry partners at Equinor, KSAT and NOFO.In summary, improved oil spill thickness and drift estimates are valuable tools for oil spill clean-ups.

Fjord ice work
Another work focus is the PhD thesis of Megan O'Sadnick on fjord ice -where and when it forms, its properties, and impacts on surrounding communities.The direct usefulness of her work was demonstrated in early 2022, where after a large avalanche in February the road to the community in Beisfjord (Nordland, Norway) was closed.The army tried to provide supplies to residents by boat but was prevented from accessing the community by boat because of ice surrounding the dock.This forced the army to land further away from the town on a rocky coastline that is not always accessible due to tides.

Ocean Color Remote Sensing
CIRFA is consolidating its work on Ocean Colour with the work of associate scientists Katalin Blix, PhD candidate Muhammad Asim, and Camilla Brekke.Muhammad Asim joined a research cruise where he collected water samples and measured radiance to better understand the biological structure of the water column.Katalin was granted the project ITEX (Innovative Technological solutions to monitor water quality in optically complex waters) by the Norwegian Research Council to organize and conduct a 1-year data collection campaign at Lake Balaton in Hungary starting in summer 2021.Katalin developed machine learning algorithms that were transferred to the optically complex waters of the Barents Sea and surrounding oceans to understand their influences.

WP leader
Tor Arne Johansen Professor, NTNU Remotely Piloted Aircraft Systems Technology

Background and Objectives
Both satellite based systems and RPAS (Remotely Piloted Aircraft Systems) have their strengths and weaknesses.Satellites have superior coverage and repeatability, but limitations when it comes to accurate fine spatial and temporal scale measurements of thickness distribution, drift, convergence and divergence.RPAS can achieve accurate high-resolution measurements, but have limited spatial coverage and range, and are weather sensitive.
The systems needed by industrial operators in the Arctic should be robust and reliable, and the system should be able to handle disruption in service by individual components.This work package aims to develop robust and efficient RPAS and sensor technologies suited to support arctic operations, that are flexible and that can handle the widest possible ranges of environmental conditions enabling high quality measurements of sea-ice and iceberg properties, as well as detecting and monitoring oil spills in ice affected areas.RPAS combined with satellites and state of the art models provide the optimal data based situational awareness tool.

Key research tasks
Develop platforms with improved take-off and landing capabilities, de-icing performance, wind tolerance, and fault tolerance.Improve communication links, robustness and bandwidth in Arctic RPAS operations.Develop RPAS sensors for sea ice characterization, ocean surface parameters measurements, and oil-in-ice detection and tracking.Develop onboard data processing and improve 'concepts of operation' allowing for realtime operation support and ultimately integration into non-segregated airspace.

UWB radar development for measurements of snow and sea-ice properties
The PhD project on Ultra-wide-band (UWB) radar was successfully completed by Rolf-Ole Rydeng Jensen in May 2021.The radar was used to measure snow pack parameters such as snow depth and characterize sea-ice properties.A method to reliably and remotely estimate snow water equivalent was also developed.This led to improved accuracy in snow density estimates.

Multisensor sea-ice properties retrieval for real time operational support
A sensor system has been designed for mounting on a fixed wing VTOL drone to be operated from the R/V Kronprins Haakon during the CIRFA cruise planned for 2022.The goal is to provide real time operational support for vessels navigating through sea-ice, and to demonstrate this on the 2022 cruise.The system consists of four cameras, one nadir looking and three forward looking as shown in Figure 4-2.This system will give better situational awareness at the same time as accurate quantification of the surface below the aircraft can be retrieved.Georeferenced data will be displayed in real time in the web based NLIVE application on the ship bridge.A dual antenna GPS compass has been integrated in the drone to improve accuracy in orientation and avoid problems with use of magnetic compass in ship environment and at high latitudes.Figure 4-2 shows a NORCE Shark drone that is instrumented for ship based operations for supporting sea-ice navigation, oil spill detection and mapping.On the right side, the footprints of the mounted camera system are presented.Photos: NORCE.

Background and Objectives
Forecast models for the ocean and sea ice cover predict oceanic conditions in the near future (1-3 days ahead) and use two steps.In the first step, the forecast systems ingest observations from remote sensing and in-situ platforms to estimate the current state of the ocean.In the second step, the ocean's state is projected forward in time by solving physical laws numerically.
Our overall objective is to improve operational ocean, sea-ice, and weather forecast models by developing new data assimilation (DA) algorithms and by using new types of observations provided in other work packages.DA techniques combine the model state with various types of ocean and sea-ice observations, providing an analysis of the ocean's state and sea ice cover.For Arctic operations, critical forecast parameters are the position of the sea ice edge and ocean currents as well as the weather.Quantification of uncertainty in model predictions is key for dealing with the ocean's chaotic nature, and it is therefore necessary to calculate multiple realization of the ocean's state -commonly known as ensemble prediction systems (EPS).

Key research tasks
Application of the Ensemble Kalman Filter (EnKF) DA scheme in the regional ocean and ice forecast model Barents-2.5Quantification of forecast skill in ocean current predictions from the Barents-2.5EPS Application of ensemble forecast methods in trajectory modeling Assimilation of sea surface temperature (SST) observations from various platforms in the coastal forecast model NorKyst-DA, using a 4D-var DA scheme.Improving SAR-based wind retrieval algorithms in polar low and extreme wind situations, working towards DA of SAR-winds in atmospheric forecasts.

Achievements in 2021
After nearly 5 years of research on data assimilation methodology in the Barents-2.5model, we have launched an operational implementation of the EnKF in Barents 2.5 in August 2021.At present, real-time sea ice concentration, SST, and in-situ data are assimilated.This system is based on the Barents 2.5 EPS developed in the previous year, consisting of 24 daily members that allow us to assert the uncertainty in predictions for sea ice, currents, and temperature fields (Figure 5-1).
Newly employed PostDoc Martina Idzanovic has started to investigate predictive skill of the Barents 2.5 model to forecast surface currents, using HF radar observations of surface currents as reference for validation.PhD candidate Silje Iversen investigates how to best use the information from various SST observing platforms including both passive microwave and infrared sensors.
PhD student Mathias Tollinger works on obtaining surface wind information with high spatial resolution over sea from SAR observations.These can provide new detailed wind information regarding high-latitude extreme wind events such as polar lows.Mathias has provided important improvement to the wind-retrieval algorithm and works on assimilating these high-resolution, SAR-based wind retrievals into the Harmonie-Arome weather prediction model, used operationally at MET Norway, to improve predictions of weather events with strong wind gradients.
In collaboration with WP3, PhD candidate Victor de Aguiar uses ocean model ensembles to drive respective trajectory model simulations for marine pollution studies.In addition to understanding the drift of pollution, the ensemble allows us to quantify uncertainty and asses the true range of possible drift scenarios.In 2021, several cruises with RV Kronprins Haakon (Nansen Legacy/Norwegian Polar Institute) were conducted with direct links to CIRFA work.In addition, the planning for a specific CIRFA cruise in April-May 2022 began.

Key research tasks
Plan and conduct dedicated field campaigns on Arctic sea ice, oceans, and oil spills to combine accurate direct measurements of surface properties with airborne data and satellite-based data.Improve validation shortcomings by seeking and implementing new and refined measurement concepts and methods using new technologies and platforms.Provide quality ground-truth data from archives and new campaigns for assessing the theoretical work in other work packages.

Achievements in 2021
During three cruises within the Nansen Legacy project from the northern Barents Sea and transition to the deep Arctic Basin (March and April-May), and from the central Arctic Ocean (August/September), sea ice datasets were collected.This includes ice core samples, snow depth and ice thickness measurements and transects, as well as aerial photography

Malin Johansson
Researcher, UiT (from August 2021) from a Remote-Piloted Aircraft System (RPAS).CIRFA Postdoc Anca Cristea participated in the JC2-2 cruise and had the opportunity to measure and sample some of the northernmost Arctic summer sea ice.Furthermore, scientists Dmitry Divine and Sebastian Gerland participated in the cruise Nansen Legacy Q1 in March.
The UAK 2021 campaign with the Norwegian coast guard vessel KV Svalbard, led by Hanne Sagen and Stein Sandven from NERSC, took place in June.CIRFA PhD students Laust Faerch and Anna Telegina as well as associate PhD student Jozef Rusin were onboard.During the cruise, results from the CIRFA sea ice classification algorithm (Lohse et al., 2020) were sent directly to the ship for near-real time validation.In addition, sea ice classifications using RADARSAT-2 images were sent to the ship for validation making this a successful collaboration between WP2, WP6 and WP7.
More is to come in 2022 when the algorithms using both Cand X-band SAR satellite data will be validated during the planned CIRFA 2022 cruise.On all cruises mentioned here, standardized IceWatch-ASSIST sea ice observations were done regularly and uploaded to IceWatch online portal.Furthermore, in-situ work in Kongsfjorden on landfast sea ice was carried out by NPI researchers in spring 2021 adding onto earlier CIRFA work where SAR satellite observations to give additional information on the sea ice concentration in the fjord.
Satellite and oil slick detection work was carried out, both in collaboration with Wintershall Dea, Equinor and CAGE (an SFF based at UiT).The collaboration with Wintershall Dea and Equinor saw further overlaps between in-situ data of produced water and satellite images from the Brage and Norne platforms.Together with scientists from CAGE, overlapping in situ and satellite data was collected over naturally occurring oil and methane seepages in the Barents Sea, with special focus on Hopendjupet during in-situ data campaigns lead by CAGE in July and December.
Beyond the fieldwork, time was spend in WP6 on data processing and curation, developing and writing of scientific manuscripts, presenting results, and preparing for the CIRFA 2022 cruise with RV Kronprins Haakon.

Pilot Service Demonstration
Background and Objectives Satellite-based operational capabilities including oil spill detection, ship traffic monitoring and sea ice mapping have been demonstrated and developed into regular use.However, there are still requirements for industrial maritime operations in the environmentally sensitive Arctic that have not been met, such as monitoring technologies integrated into their day-to-day operations for decision support.
In this work package, we will demonstrate pilot services showing the provision of integrated environmental information to end-users involved in Arctic operations.The objective of this work package is to demonstrate the implementation of R&D results into pilot services to be delivered to end-users with operational needs.The services will be based on multi-sensor data acquired from various sensors and platforms, accessed via improved communication infrastructure and brought into analysis and decision through dedicated interfaces.

Key research tasks
Establish an infrastructure that allows the WP partners to access and perform processing on the project data in a technically efficient way close to the data storage.
Integrate the R&D results from the other WPs into service demonstrations at KSAT and/or MET Norway to show the provision of integrated environmental information to end-users involved in Arctic operations.Develop a visualization solution associated with the integrated pilot services demonstrations.
Detecting the presence of oil spills, the thickness of the oil and its drift direction and speed are helpful information in case of an oil spill.Detecting mineral oil slicks in ice covered seas from space The thinning and retreat of the arctic sea ice has led to increased human presence in Arctic seas.Marine traffic is most likely to increase in the future, as are activities such as fishing, oil and mineral exploitation.All these activities increase the risk for oil spills in ice-covered waters.Yet, the technology used to monitor for marine oil spills in the World's oceans is not yet applicable for ice-covered seas.A new study (Johansson et al, 2020) presents a first approach to overcome this limitation.
EGU Blog Post: Mapping Sea ice from Space Reduced and thinner sea ice makes Arctic waters increasingly appealing for shipping, fishing, tourism, and mineral exploration.However, with increased accessibility and more dynamic ice conditions comes a greater risk for ship crews to encounter sea ice and icebergs outside of their usual seasonal limits.To help them navigate, timely and reliable sea ice information is key.Have you wondered how sea ice information can be provided for vast and remote areas such as the Arctic Ocean?CIRFA has a post on the award-winning blog of the European Geosciences Union (EGU) about mapping sea ice from space.

National and International Collaboration
In its research field, CIRFA is associated with 37 active national and international research and innovation projects or varying funding amount, time line and complexity.CIRFA researchers were successful to secure funding from DLR, RCN, Framsenter, Norwegian Space Agency and others.About half of them are international projects that involve organisations such as  The Horizon 2020 Green Deal Project IMPETUS The European Union (EU) strategy on adaptation to climate change recently adopted by the European Commission (EC) addresses the importance of improved adaptation and resilience to climate change.The impact of climate change is already visible on economies, communities and ecosystems worldwide with consequences on citizens' wellbeing as well as the natural and built environment.Remote sensing can play a key role in this context.
The H2020 Green Deal project IMPETUS aims to integrate remote sensing datasets and products to accelerate the transition towards a climate-neutral and sustainable economy.Remote sensing data will be complemented by additional data collected on the ground and assessment methods to support decision and policy making within a process of co-creation with local stakeholder communities.The result of this approach will be the co-creation of regional Adaptation Pathways, aiming at optimizing adaptation and mitigation approaches to climate change in a specific region based on the needs of local communities.IMPETUS is expected to lead to increased community empowerment in all seven EU biogeographical regions (Continental, Coastal, Mediterranean, Atlantic, Arctic, Boreal, Mountainous) covering key community systems, climate threats, and multi-level governance regimes.
The key research topics of the IMPETUS project.
Photo by Christian Zoelly Ocean Surface Currents in the Coastal Zone From the Sentinel-1B Doppler Shift Observations.a) is a Sentinel-1B SAR scene acquired in IW mode on 10 December 2017 and provided by Copernicus; b) shows the ocean current velocity from the Radar; c) shows ocean currents from a model; d) shows the night time ocean surface temperature; e) shows the ocean current velocity from a model.Image credit: Moiseev et al. 2021.

Figure 2 -
Figure 2-1 shows icebergs in L-band SAR images.Left: Optical images, Right: Coarse and high spatial resolution.In our recent work on iceberg detection in L-band satellite SAR images we found shifts of iceberg positions in comparison to optical data, and next to the main shifted radar return (in yellow) ghost reflections (in purple).Sentinel-2 (left), and PALSAR-2 Wide Beam (Center) and Fine Beam (right).
Detection and characterization of oil spills on open waterDetection of oil spills in fjord sea ice Integration of drift modelling and remote sensing for marine environmental monitoring

Figure 3
Figure 3-1, Left: The fjord two days after the avalanche with the location of avalanche and the dock noted.Right: A navy boat delivers supplies at an alternative landing spot.Photos: Vidar Løkeng, Fremover.

Figure 3 -
Figure 3-2 shows impressions from Ocean Color fieldwork on lake Balaton (top row, Photos: Katalin Blix, UiT) and in the Barents Sea (bottom row, Photos: Muhammad Asim, UiT).

Figure 4 -
Figure 4-1 shows the drone that carries the snow radar being prepared for its next flight.Photo: Anders Martinsen.

Figure 5 -
Figure 5-1 shows surface current speed (green shading) and sea surface height (red contours) at the coast of Finnmark, northern Norway, from 24 ensemble members of Barents 2.5 at 2021-11-15 18:00.The ensemble members represent different realizations of ocean forecasts, where each is one possible outcome of equal likelihood to occur.The members exhibit both similarities (e.g.position of coastal current jet) as well as distinct differences (e.g.strength of coastal current).

Figure 6
Figure 6-1: During two cruises, regional-scale Helicopter-borne Electro Magnetic (HEM) ice thickness measurements were also performed in overlap with remote sensing imagery and ground data.This overlap aides in the validation of remote sensing measurements and algorithms.Photos: Sebastian Gerland, NPI.

Figure 7 -
Figure 7-1 shows drift trajectories around Jacarta, Indonesia, calculated using the OpenDrift model.The model was initialized based on oil detected with two thickness classes from an optical Sentinel-2 image compiled by Martine Espeseth, KSAT.
Left: A cargo ship navigating through icy waters, seen from a drone.Photo: Alex Perz/Unsplash.Right: Example of the sea ice classification in satellite images North of Svalbard.The different colours represent different ice types and open water.The red line outlines the ship track.Data used to produce this figure come from Sentinel (Copernicus) and Radarsat-2 (NSC/KSAT, Norwegian-Canadian Radarsat agreement 2021).The map is made by Anna Telegina and Johannes Lohse (UiT).
e r s e w e a t h e r m is s io n p la n n in g fo r UA V's Ic in g e e c ts , d et ec tio n, an d mit igation on UAV 's CO SM IC IETX INT ER AA C H a rm o n y A r k t is 2 0 3 0 S e n t i n e l -1 L -b a n d S A R A p p l i c a t i o n s R e q u ir e m e n t s C o n so li d at io n St udy S 1 -4 S C I O c e a n S t u d y S E O M S A T R o S S Se nt in el 1 Ra dia l Velo city An overview of related projects and funding sources.Image credit: Torger Grytå, YUiT.

Team members
The OpenOil model is now set up at KSAT in collaboration with MET for a testing phase.It has been improved by including the damping ratio.Results about "Oil Spill drift model Initiated with Thickness Variations" were presented at the Esri European Petroleum GIS Conference by Martine M. Espeseth, Hugo Isaksen, Alastair Cannell, and Charlotte Bishop.

workshop on sea ice classification.
The workshop had two practical sessions and was organised by Johannes Lohse.It reviews the basics of SAR image classification, the installation of data processing packages, and the main workflow steps of data analyses for SAR image classification.The six lectures covered the basics of Ocean Color remote sensing, atmospheric properties and correction, ocean biology from space, inversion algorithms and calibration, validation and errors.With the lecturers based in the US, Hungary and Norway, the course was run digitally.The traditional CIRFA Young Scientist Forum 2021 took place in December.PhD candidates Megan O`Sadnick and Muhammad Asim arranged a 2-day workshop with focus on external research funding, career design thinking, stress management and mindfulness.In the evening, games with Bumperball, Lasertag, and a pizza dinner offered the participants pleasant diversion and teambuilding.The CIRFA Annual Conference 2021 took place during October 19-21 at the Sommarøy Arctic Hotel near Tromsø.This year's conference featured sessions on key research topics such as ocean modelling and data assimilation, recent advances in sea ice classification, iceberg detection, oil slicks and natural oil seeps in the Barents Sea, and relevant developments and applications of machine learning.The conference started with a poster session with mingling, had a session on research cruise planning, early career researcher presentations, and two workshop sessions to advance ongoing projects.This year's conference in a hybrid format had 42 in-person and 22 remote participants.This podcast series is an initiative of the European Commission and part of the European Research and Innovation Days.
Throughout the autumn semester between September and November, Katalin Blix, Muhammad Asim, Atsushi Matsuoka and Torbjørn Eltoft were giving a special course on Ocean Color Remote Sensing.A screenshot from the workshop on sea ice classification (Sentinel-1 & Copernicus, a satellite`s perspective on ocean color (data from the European Space Agency, processed by Katalin Blix); and a photo from the Young Scientist Forum (Johannes Lohse).In the podcast series "Behind the Scenes" of the EU Research and Innovation Exhibition, journalist Jennifer Baker spends time with some of Europe's most brilliant scientists and innovators whose discoveries are having a lasting and important impact on our daily lives.The episode "A village on ice" is connected to the MOSAiC expedition and open questions in Earth climate sciences, but it also talks about Polona Itkin`s inspiration to become a researcher.snow/ice/ecosystem models using winter-to-summer ARctic observations of coupled snow, ice, and ecosystem processes) took part during the MOSAiC expedition.The aim was to collect unique data on snow and sea ice over a full year to improve our understanding of the vulnerability of the Arctic sea ice system, and decrease uncertainties in climate projections.Participants of the CIRFA Annual Conference 2021 at the Sommarøy Arctic Hotel near Tromsø.Photo: CIRFA.
DLR, IICWG, ESA and JAXA.Many of the projects employ artificial intelligence and/or machine learning to analyze climate and environmental data, and research training and competence building.The largest associated projects are The Nansen Legacy, KEPLER, ExtremeEarth, and the new Horizon 2020 Green Deal project IMPETUS.Five new projects were awarded funding in 2021: SLABAR -Integrating SeaLevel Anomalies into the BARents model with funding from the Norwegian Space Agency to Martina Idzanovic ITEX -Innovative Technological solutions to monitor water quality in optically complex waters with funding from RCN to Katalin Blix INTERAAC -Air-snow-ice-ocean INTERactions transforming Atlantic Arctic Climate with funding from NFR to Jack Landy; the Peder Sather foundation grant to Andrea Marinoni; and the IMPETUS H2020 grant to Andrea Marinoni.