Ragnar Bang Huseby

Electoral thresholds have been used in national parliamentary elections in various countries. An electoral threshold is the minimum proportion of votes for political parties to compete for seats in the parliament. Instead of using a single electoral threshold, which is common, we investigate an alternative approach based on multiple electoral thresholds. In this work, we focus on the voting system in Norway. There are currently 169 seats in the Norwegian legislature. Of the 169 representatives, 150 are elected from 19 constituencies such that the representatives from a given constituency are determined by the popular vote within that constituency. The remaining 19 seats are levelling seats, which purpose is to achieve more proportional representation at the national level. An electoral threshold of four percent applies to the levelling seats. A weakness by using a single electoral threshold for allocating the levelling seats is that a small change in the proportion of votes could yield an excessive change in the proportion of allocated seats when a party crosses the electoral threshold. In order to reduce this weakness, we propose to introduce a set of multiple electoral thresholds, t1 < ... < tn, where n is the number of levelling seats. Each threshold applies to one of the levelling seats. Typically, the thresholds are evenly spaced, while t1 and tn are lower and greater than the current threshold of 4%, respectively. We compare the current and the proposed approaches on the most recent parliamentary elections in Norway as well as simulated elections. The simulations are based on a statistical model for the proportions of the political parties in Norway. The results of our investigations indicate that the proposed rule for allocating levelling seats is less sensitive to small changes in the vote share than the rule in use.

We investigate various approaches to reduce the problem of salmon louse for the salmon farming industry. Salmon lice are parasites that can be spread through water from one fish farm to wild fish and to neighbouring fish farms. Various measures have been used to control salmon lice in fish farms, but frequent use of the same type of treatment may cause resistance, especially for medicinal treatments and possibly for non-medicinal treatments (e.g. freshwater). Our work focuses on resistance development caused by genetic selection. This means that some genotypes of lice have a higher probability of surviving a given treatment than other lice, and consequently the proportion of lice of resistant genotypes tends to increase by using that kind of treatment. We consider treatment policies involving two types of treatments with similar efficacy properties, but with independent selection mechanisms. The investigation is done by simulating the population of lice in neighbouring fish farms using a population model for lice. When the density of lice on a small sample of fish exceeds a given threshold, a treatment, which type depends on the strategy, is applied. For each simulation we record the number of required treatments and the levels of resistance. Regarding delaying resistance our results suggest that applying the two treatment types in combination is more effective than any of the considered strategies with separate use of the two types. This is joint work with researchers at the Norwegian Veterinary Institute, Aqualife R&D, Pharmac and the Norwegian University of Life Sciences.

Arbeidet som beskrives her er utført av Norsk Regnesentral, i samarbeid med Aqualife R&D, og er en del av er del av prosjektet “Opplevd smittepress for lakse smolt og sjøaure i Nordfjord, Sogn, Bjørnafjorden og Hardanger 2012-2023” som finansieres av Vestaland fylkeskommune. Hovedmål i prosjektet har vært å modellere opplevd smittepress av lakselus fra oppdrett hos laksesmolt og sjøaure fra elver i Nordfjord, Sogn, Bjørnafjorden og Hardanger basert på observerte vand ringsforløp i fjordsystemene. Delmål i prosjektet har vært å: • Sammenstille datasett på vandringsruter til laks og sjøaure ervervet ved bruk av akustisk telemetri som metode i fjordsystem på Vestlandet • Beregne lusepåslag og videreutvikling til forekomst av skadelige bevegelige lus på laksesmolt og sjøaure med kjente vandringsruter • Genetiske analyser av sjøaure; for kartlegging av effektiv bestandsstørrelse og slektskap med sjøaure fra andre elver i fjordsystemet • Styrke samarbeidet mellom villfiskinteresser, forskningsmiljø, forvaltning og oppdrettsnæring ved å etablere en kunnskapsbasert dialog mellom aktørene I denne rapporten beskrives bruken av en avansert populasjonsmodell for lakse lus til utvikling av en smittemodell basert på sammenhengen mellom forekomster av hunnlus på oppdrettsfisk og 1. forekomster av lakselus på utvandrende post smolt av laks som er sporet til en oppvekstelv, og 2. luseforekomster på trålfanget laksesmolt. Videre beskrives hvordan smittemodellen anvendes til beregning av smittebyrde, både av kjønnsmodne hunnlus og andre mobile stadier av lakselus, på utvandrende laksesmolt fra i alt 9 elver og 35 elve-år i PO3 og 22 elver og 91 elve-år i PO4 i Vestland. Endelig beskrives en analyse av effekten av beregnet smittebyrde på antall gytelaks som returnerer til disse elvene i gitte år.

Climate sensitivity and aerosol forcing are two of the most central, but uncertain, quantities in climate science that are crucial for assessing historical climate as well as future climate projections. Here, we use a Bayesian approach to estimate inferred climate sensitivity and aerosol forcing using observations of temperature and global ocean heat content as well as prior knowledge of effective radiative forcing (ERF) over the industrial period. Due to limited information on uncertainties related to the time evolution of aerosol forcing, we perform a range of sensitivity analyses with idealized aerosol time evolution. The estimates are sensitive to the aerosol forcing pathway, with the mean estimate of inferred effective climate sensitivity ranging from 2.0 to 2.4 K, present-day (2019 relative to 1750) aerosol ERF ranging from −0.7 to −1.1 W m−2, and anthropogenic ERF ranging from 2.6 to 3.1 W m−2. Using observations and forcing up to and including 2022, the inferred effective climate sensitivity is 2.2 K with a 1.6 to 3.0 K 90 % uncertainty range. Analysis with more freely evolving aerosol forcing between 1950 and 2014 shows that a strong negative aerosol forcing trend in the latter part of the 20th century is not consistent with observations. Although we test our estimation method with strongly idealized aerosol ERF pathways, our posteriori estimates of the climate sensitivity consistently end up in the weaker end of the range assessed in the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC AR6). As our method only includes climate feedbacks that have occurred over the historical period, it does not include the pattern effect, i.e., where climate feedbacks are dependent on the pattern of warming which will likely change into the future. Adding the best estimate of the pattern effect from IPCC AR6, our climate sensitivity estimate is almost identical to the IPCC AR6 best estimate and very likely range.

The effective control of ectoparasitic salmon lice, Lepeophtheirus salmonis, in fish farms is challenged by the salmon lice having developed resistance towards several antiparasitic drugs and by the effectiveness of non-medicinal treatments being limited by considerations of fish welfare. When new antiparasitics are introduced to the market, these should be used sparingly to slow resistance development. Using a population model for salmon lice parameterised for salmonid fish farms in Norway, we quantified how reduced treatment effectiveness influences treatment frequency and lice abundance. Furthermore, we investigated when in the production cycle a highly effective lice treatment leads to the largest reduction in the total number of treatments, mean lice abundance and lice larvae production. Results showed that reductions in treatment effectiveness to lower than 50% led to the steepest increases in treatment frequency and mean lice abundance, as well as to increased risk that lice abundance increased beyond control. The timing of the most effective treatment had only moderate effects on the total treatment need and the mean number of adult female lice through the production cycle, but large effect on the production of lice larvae in spring. These findings imply that farmers can optimise the timing of the most effective treatment to reduce the release of lice larvae in the period of year when wild salmonids are in coastal waters, without compromising total treatment need or mean lice levels.

Vi beregner her i hvilken grad passering av brønnbåter eller nærhet til slakterier øker risikoen for at fisken på et oppdrettsanlegg (lokalitet) blir smittet med PD (pancreas disease). Beregningene gjøres på grunnlag av data fra produksjonsområdene fra 2 til 7 fra starten av 2013 til midten av 2022. Vi bruker så en smittemodell for PD til å etablere sammenheng mellom registrert PD-smitte og ulike smittekilder. Den klart viktigste smittekilden er smitte fra naboanlegg, og 85 % av smittetilfellene tilskrives dette. Videre anslår vi at 12 % av smittetilfellene er relatert til at det var PD på forrige utsett på samme anlegg. Smitte relatert til nærhet til slakteri tilskrives 0,3 % av smittetilfellene. Når det gjelder brønnbåter tilskrives 0,2 % av smittetilfellene passering av brønnbåter og 0,2 % besøk av brønnbåter. Til sammen kan slakterier og brønnbåter tilskrives 0,7 % av smittetilfellene. Øvrige smittekilder utgjør 2,6 %. Det er viktig å redusere risikoen fra alle smittekilder, men når det gjelder PD er det aller viktigst er å hindre smitte mellom oppdrettsanlegg.

We here estimate the effectiveness of ten types of salmon lice treatments currently used in the salmonid industry by analysing daily and cage-wise data from 90 full production cycles from farms spread along the Norwegian coast. The calculations are based on a stage-structured population model for salmon lice and accounts for the structure of the data, including the uncertainties that arise from the weekly counting of lice on a subset of the fish. Results suggest that the most commonly used treatment methods in the data set, i.e. thermal, mechanical and freshwater treatments, kill 70–80% of the lice in average, but with high variability. Feed treatments with emamectin benzoate are also commonly used, but are only estimated to kill around 35% of the lice in average. Bath treatments with hydrogen peroxide are estimated to kill around 74% and pyrethroids 50% of the lice in average. The other medicinal treatments were infrequently used in the data set and the estimates are therefore more uncertain. Of note is that the recently licenced bath treatment with imidacloprid is estimated to kill more than 99% of sessile and 98% of pre-adult and adult lice in average. The estimated effects of hydrogen peroxide, pyrethroids and azamethiphos, here based on data from 2017–2020, are lower than estimates from a previous analysis of production data from 2011–2014, possibly because of resistance development. In contrast, there is no indication of reduced effects of thermal, mechanical or freshwater treatments compared to previous analyses of production data from 2013–2018. These results allow comparing the effectiveness of the different treatment methods in a consistent and comprehensive way, hence enabling fish farmers and authorities to better balance the expected benefits of the treatments in terms of lice control against the economic costs, fish welfare and risk of resistance development.

Infeksiøs lakseanemi (ILA) er en alvorlig virussykdom som forårsaker økonomisk tap og redusert velferd hos atlantisk oppdrettslaks. Vi har undersøkt forskjellige kontrollstrategier, eller scenarier, for mulige tiltak for å bekjempe ILA. Strategiene inkluderer vaksinering, utslakting etter klinisk ILA-utbrudd, screening med påfølgende utslakting ved påvist ILA-virus og kombinasjoner av de ulike strategiene. Vi beregner hva hver strategi medfører av antall smittetilfeller og registrerte kliniske utbrudd av ILA. De samme beregningene gjøres for en strategi der vi ikke gjennomfører tiltak i det hele tatt. For å komme frem til resultatene simulerer utviklingen av smitte fremover i tid ved hjelp av en statistisk modell for spredning av ILA. Modellen beskriver hvordan smitte- og utbruddsdynamikken påvirkes av ulike faktorer som inkluderer smittestatus på de ulike oppdrettslokalitetene, anleggenes beliggenhet, størrelsen av anleggene og tid på året. For å estimere modellen benytter vi historiske produksjonsdata som dekker alle oppdrettsanlegg i Norge i perioden f.o.m. januar 2004 t.o.m. februar 2019. I denne perioden ble det påvist 142 kliniske ILA-utbrudd. Fra den estimerte modellen finner vi at omtrent halvparten av all smitte kommer fra smittet fisk på naboanlegg. En stor del av den resterende smitten skyldes trolig infeksjon med virusvarianten HPR0 som har utviklet seg til HPRdel-varianten. For hvert scenario vi undersøker, tar vi utgangspunkt i en gitt nåsituasjon og simulerer hvordan ILA utvikler seg ti år frem i tid. For at simuleringene skal bli mest mulig realistiske, antar vi at beliggenheten av anleggene og produksjonsperiodene er som i perioden f.o.m. mars 2009 t.o.m. februar 2019. Samtidig antar vi at gjennomsnittlig antall fisk per lokalitet er omtrent som i februar 2019 gjennom hele simuleringsperioden. Simuleringene viser at smitten stabiliserer seg på et nytt nivå etter et par år avhengig av hvilken kontrollstrategi som benyttes. Vi finner at 18,4% av alle utsett i Norge vil få påvist et klinisk utbrudd av ILA dersom vi ikke bruker kontrolltiltak. Dersom vi gir alle utsett en vaksine som reduserer sannsynligheten for å bli smittet med 50%, vil andelen med utbrudd bli rundt 3,8%. Dersom vi istedenfor vaksinering velger utslakting seks uker etter deteksjon av utbrudd, reduseres andelen med utbrudd til 3,4%. Andelen reduseres ytterligere til 0,36% ved screening av 20 fisk hver fjerde uke og utslakting seks uker etter deteksjon av virus. Vår studie viser hvordan scenariosimulering kan benyttes som verktøy for beslutningsstøtte.

Pancreas Disease (PD) is a viral disease caused by Salmonid Alphavirus (SAV). It affects farmed salmonids in the North Atlantic, and leads to reduced feed intake and increased mortality with reduced production and welfare as a consequence. In 2013, the estimated cost of an outbreak on an average salmon farm was about 6.6 mil €. In Norway, PD has been notifiable since 2008, and regulations to mitigate disease spread are in place. However, despite the regulations, 140–170 farms are affected by PD every year. The aquaculture industry is growing continuously, introducing farms in new geographical areas, and fish are moved between hydrographically separated zones for trade and slaughter. All such movements and relocations need to be approved by the competent authorities. Thus, there is a demand for support to farmers and competent authorities when making decisions on disease management and especially on the effect of moving infected fish. We have used a disease-transmission model for outbreak-simulation in real time for assessing the probability of disease transmission from a farm that gets infected with PD. We have also simulated the effects of three different control-regimes: no stamping-out, delayed stamping-out or immediate stamping-out, on the transmission of PD to surrounding farms. Simulations showed that the immediate stamping out of an infected farm led to effective containment of an outbreak. No stamping out led to up to 32.1% of farms within 100 km of the index farm to become effected. We have used real production data for the model building and the scenario simulations, and the results illustrate that a risk assessment of horizontal disease transmission must be undertaken on a case-by-case basis, because the time and place of the outbreak has a large influence on the risk of transmission.

Infectious salmon anaemia (ISA) is an important viral disease causing economic losses and reduced welfare in farmed Atlantic salmon. Here, we present a spatio-temporal stochastic model for the spread of ISA between and within marine aquaculture farms. The model is estimated on historical production data for all marine salmonid farms in Norway from 2004 to February 2019. In this time 142 outbreaks of ISA occurred. We find that transmission from infected neighbouring farms accounts for around 50% of the infections, whereas transmission from “non-specified sources” accounts for around 40%. We hypothesise that the most important of the latter are viruses mutating from the non-virulent ISAV HPR0 to the virulent ISAV HPRdel. The model is used for scenario simulation, or what-if analysis, to investigate the effects of potential strategies to combat ISA, including screening, vaccination and culling. Changing from the current strategy of culling farms with detected ISA-outbreaks to mandatory screening and culling when virus is detected will reduce the fraction of cohorts with a clinical ISA outbreak from 3.8 to 0.36%. Introducing mandatory vaccination would have approximately the same effect as the current stamping-out strategy. The scenario simulation is a useful tool for deciding on appropriate mitigation measures.

Purpose The purpose of this paper is to develop, describe and validate a machine learning model for prioritising which financial transactions should be manually investigated for potential money laundering. The model is applied to a large data set from Norway’s largest bank, DNB. Design/methodology/approach A supervised machine learning model is trained by using three types of historic data: “normal” legal transactions; those flagged as suspicious by the bank’s internal alert system; and potential money laundering cases reported to the authorities. The model is trained to predict the probability that a new transaction should be reported, using information such as background information about the sender/receiver, their earlier behaviour and their transaction history. Findings The paper demonstrates that the common approach of not using non-reported alerts (i.e. transactions that are investigated but not reported) in the training of the model can lead to sub-optimal results. The same applies to the use of normal (un-investigated) transactions. Our developed method outperforms the bank’s current approach in terms of a fair measure of performance. Originality/value This research study is one of very few published anti-money laundering (AML) models for suspicious transactions that have been applied to a realistically sized data set. The paper also presents a new performance measure specifically tailored to compare the proposed method to the bank’s existing AML system.

We develop, describe and validate a machine learning model for prioritising which financial transactions should be manually investigated for potential money laundering. The model is applied to a large data set from Norway's largest bank, DNB. This is one of very few published anti-money laundering models for suspicious transactions that has been trained and validated on a realistically sized data set. We demonstrate that the common approach of not utilising non-reported alerts, i.e.~transactions that are investigated but not reported, in the training of the model can lead to sub-optimal results. We also demonstrate the benefit of including normal (uninvestigated) transactions in the training of the model, and study whether explicitly modelling the probability that a transaction is a normal transaction can improve the money laundering detection rate. Therefore, we present a new performance measure for comparing our method to the existing anti-money laundering system in the bank. Using this performance measure, we clearly outperform the bank's current approach.

Salmon farming has become a prosperous international industry over the last decades. Along with growth in the production farmed salmon, however, an increasing threat by pathogens has emerged. Of special concern is the propagation and spread of the salmon louse, Lepeophtheirus salmonis. To gain insight into this parasite’s population dynamics in large scale salmon farming system, we present a fully mechanistic stage-structured population model for the salmon louse, also allowing for complexities involved in the hierarchical structure of full scale salmon farming. The model estimates parameters controlling a wide range of processes, including temperature dependent demographic rates, fish size and abundance effects on louse transmission rates, effect sizes of various salmon louse control measures, and distance based between farm transmission rates. Model parameters were estimated from data including 32 salmon farms, except the last production months for five farms, which were used to evaluate model predictions. We used a Bayesian estimation approach, combining the prior distributions and the data likelihood into a joint posterior distribution for all model parameters. The model generated expected values that fitted the observed infection levels of the chalimus, adult female and other mobile stages of salmon lice, reasonably well. Predictions for the periods not used for fitting the model were also consistent with the observational data. We argue that the present model for the population dynamics of the salmon louse in aquaculture farm systems may contribute to resolve the complexity of processes that drive this host-parasite relationship, and hence may improve strategies to control the parasite in this production system. Population model Aquaculture Stochastic model Sea lice counts

Rapporten oppgir beregnede effekter av en del ulike strategier for lakselusbekjempelse. Effektene angis her som hvor mye (i prosent) kan antall behandlinger mot lus reduseres ved å bytte fra én strategi til en annen, mens lusenivået holdes uendret. Med en behandling mener vi her en behandling som gir en umiddelbar ekstra dødelighet for all lus på fisken. Dette er beregnet som en samlet effekt for et geografisk område hvor det antas at alle oppdrettsanlegg i området i følger samme strategi. Beregningene er gjort ved scenariosimuleringer (hva hvis-analyse) fra en populasjonsmodell for lus. Beregningene viser blant annet, på betingelse at det telles lus ofte nok og på nok fisk, at det er gunstig å følge en merdvis strategi i forhold til en anleggsvis strategi. Med en merdvis strategi menes her at hver merd monitoreres for seg og at lusebehandling kun gjennomføres for merder hvor det observerte lusenivået i merda er over en viss terskel, mens en anleggsvis strategi vil si at alle merder behandles samtidig hvis det observerte lusenivået i gjennomsnitt over alle talte merder er over en terskel. Videre kan antall behandlinger reduseres hvis i) det telles lus på 20 i stedet for 10 fisk fra hver merd, ii) hvis det telles hver uke per merd i stedet for annenhver uke, og iii) hvis avgjørelsen om å gjennomføre en behandling baseres på det observerte nivået av alle bevegelige lus, ikke bare av voksne hunnlus. Hvis alle disse endringene gjennomføres kan antall behandlinger reduseres med omkring 30-35% uten at lusenivået øker. Tilsvarende effektberegninger er gjort for en del andre tiltak, inkludert bruk av rensefisk og skjørt.

Pancreas disease (PD) is a viral disease associated with significant economic losses in Scottish, Irish, and Norwegian marine salmon aquaculture. In this paper, we investigate how disease-triggered harvest strategies (systematic depopulation of infected marine salmon farms) towards PD can affect disease dynamics and salmon producer profits in an endemic area in the southwestern part of Norway. Four different types of disease-triggered harvest strategies were evaluated over a four-year period (2011–2014), each scenario with different disease-screening procedures, timing for initiating the harvest interventions on infected cohorts, and levels of farmer compliance to the strategy. Our approach applies a spatio-temporal stochastic model for simulating the spread of PD in the separate scenarios. Results from these simulations were then used in cost-benefit analyses to estimate the net benefits of different harvest strategies over time. We find that the most aggressive strategy, in which infected farms are harvested without delay, was most efficient in terms of reducing infection pressure in the area and providing economic benefits for the studied group of salmon producers. On the other hand, lower farm compliance leads to higher infection pressure and less economic benefits. Model results further highlight trade-offs in strategies between those that primarily benefit individual producers and those that have collective benefits, suggesting a need for institutional mechanisms that address these potential tensions.

Infectious diseases are a constant threat to industrialised farming, which is characterised by high densities of farms and farm animals. Several mathematical and statistical models on spatio-temporal dynamics of infectious diseases in various farmed host populations have been developed during the last decades. Here we present a spatio-temporal stochastic model for the spread of a disease between and within aquaculture farms. The spread between farms is divided into several transmission pathways, including (i) distance related spread and (ii) other types of contagious contacts. The within-farm infection dynamics is modelled by a susceptible–infected–recovered (SIR) model. We apply this framework to model the spread of pancreas disease (PD) in salmon farming, using data covering all farms producing salmonids over 9 years in Norway. The motivation for the study was partly to unravel the spatio-temporal dynamics of PD in salmon farming and partly to use the model for scenario simulation of PD control strategies. We find, for example, that within-farm infection dynamics vary with season and we provide estimates of the timing from unobserved infection events to disease outbreaks on farms are detected. The simulations suggest that if a strategy involving culling of infectious cohorts is implemented, the number of detected disease outbreaks per year may be reduced by 57% after the full effect has been reached. We argue that the high detail and coverage of data on salmonid production and disease occurrence should encourage the use of simulation modelling as a means of testing effects of extensive control measures before they are implemented in the salmon farming industry.

Suppose estimates are available for correlations between pairs of variables but that the matrix of correlation estimates is not positive definite. In various applications, having a valid correlation matrix is important in connection with follow-up analyses that might, for example, involve sampling from a valid distribution. We present new methods for adjusting the initial estimates to form a proper, that is, nonnegative definite, correlation matrix. These are based on constructing certain pseudo-likelihood functions, formed by multiplying together exact or approximate likelihood contributions associated with the individual correlations. Such pseudo-likelihoods may then be maximized over the range of proper correlation matrices. They may also be utilized to form pseudo-posterior distributions for the unknown correlation matrix, by factoring in relevant prior information for the separate correlations. We illustrate our methods on two examples from a financial time series and genomic pathway analysis.

We have constructed a statistical model to forecast, with uncertainty, the stock of Norwegian farmed Atlantic salmon (Salmo salar). The model provided good predictions of future biomass of Norwegian farmed salmon and can also be used to perform “what-if” analysis exploring the impact of varying scenarios for stocking and slaughtering. The model is based on the number of fish in each mass class (0–1, 1–2, …, 10+ kg) and their average mass. The model, which is related to standard size-structured models, computes the number of fish growing into the next mass class the next month and the number of fish remaining in the same mass class. In addition, the number of new fish stocked, fish lost, slaughtered, and wasted, as well as the sea temperature related to the growth, were modelled. All the model parameters were estimated based on monthly data from 2002 to 2007, and the model was validated statistically. Any animal production involving cycles may benefit from this forecasting tool.

This paper describes a system for co-registration of time series satellite images which uses a learning-based strategy. During a training phase, the system learns to recognize regions in an image suited for registration. It also learns the relationship between image characteristics and registration performance for a set of different registration algorithms. This enables intelligent selection of an appropriate registration algorithm for each region in the image, while regions unsuited for registration can be discarded. The approach is intended for co-registration of sequences of images acquired from identical or similar earth observation sensors. It has been tested for such sequences from different types of sensors, both optical and radar, with varying resolution. For images with moderate differences in content, the registration accuracy is, in general, good with an RMS error of one pixel or less.

The Catch Limit Algorithm for commercial baleen whaling developed by the International Whaling Commission converges slowly to a steady depletion (proportion of carrying capacity), and consequently 300 years of management is proposed as horizon for tuning and computer simulation. Long-term depletion is rather insensitive to the parameter currently used for tuning, and an alternative control parameter is suggested for this purpose.