Silius Mortensønn Vandeskog

Reanalysis products such as the ERA5 reanalysis are commonly used as proxies for observed atmospheric conditions. These products are convenient to use due to their global coverage, the large number of available atmospheric variables and the physical consistency between these variables, as well as their relatively high spatial and temporal resolutions. However, despite the continuous improvements in accuracy and increasing spatial and temporal resolutions of reanalysis products, they may not always capture local atmospheric conditions, especially for highly localised variables such as precipitation. This paper proposes a computationally efficient stochastic downscaling of ERA5 temperature and precipitation. The method combines information from ERA5 and surface observations from nearby stations in a non-linear regression framework that combines generalised additive models (GAMs) with regression splines and auto-regressive moving average (ARMA) models to produce realistic time series of local daily temperature and precipitation. Using a wide range of evaluation criteria that address different properties of the data, the proposed framework is shown to improve the representation of local temperature and precipitation compared to ERA5 at over 4000 locations in Europe over a period of more than 70 years.

En ny KI‑modell kombinerer vannførings- og kraftverksdata for å beregne vannkraft og vannføring også der vi mangler målinger. Dette gir bedre innsikt i energi, flom og naturressurser over hele Norge

Climate change will substantially impact agriculture, requiring the development of new crop cultivars adapted for resilience to future stresses. Developing a new crop cultivar is a decades-long process. Therefore climate change considerations must already be incorporated into breeding programs. National breeding programs have typically involved field trials in a single country, which has functioned satisfactorily during periods of stable climate variability. However, it is widely acknowledged that when field trial information across borders is made available, it can significantly aid plant breeding programs. This paper presents a pan-Nordic multi-environment dataset that consolidates national trial data and data from VCU (Value for Cultivation and Use) trials for spring barley, red clover, and potato, during 1970–2024. The dataset harmonizes information on yield across diverse environments, enabling comparative analyses and cross-regional assessments. By providing a unified resource for the Nordic region, this dataset supports the development of more resilient crop varieties and facilitates data-driven breeding strategies under changing climatic conditions.

Climate change projections of extreme precipitation are commonly derived from large multi-model ensembles such as EURO-CORDEX, where individual model combinations show systematic biases relative to observations. It is not clear how bias correction affects both model agreement and the projected change signal itself, or whether raw model performance provides a reliable basis for ranking ensemble members. Using annual maximum 1-day precipitation (Rx1day) as a case study, we evaluate a EURO-CORDEX ensemble against the E-OBS and ERA5 observational datasets, both in raw form and after bias correction with Empirical Quantile Mapping (EQM), mean correction mapping, and Quantile Delta Mapping (QDM). We find that historical model performance is not predictive of the magnitude of projected change and that bias correction reduces inter-model spread in the historical period, bringing most models to a level of agreement with observations comparable to the disagreement between the observational datasets themselves. Further, systematic differences in the projected change signal between certain GCMs are clear, particularly in the tails of the distribution and in regions such as the Alps and the Adriatic Sea, showing that differences in model skill remain both prior to and after bias correction. Finally, the choice of correction method matters: EQM distorts the projected change signal more than the trend-preserving QDM. Taken together, bias correction should be used as a tool for understanding how models perform, but it should not be relied on as an absolute truth.

We propose a Long Short-Term Memory (LSTM) network to estimate historical daily streamflow and hydropower generation in Norway, with particular focus on run-of-river (ROR) plants. Historical records from such plants are often limited, and typically only contain hydropower generation data, which are truncated at the plants’ capacity limits and therefore do not capture high-flow conditions. The proposed LSTM model improves predictions in data-sparse and ungauged catchments, and for high-flow conditions, by learning from both hydropower generation data from ROR plants and streamflow data from other Norwegian catchments. Our model builds upon the neuralhydrology package, by adding a component that transforms streamflow into hydropower generation before loss calculations. The model is trained using streamflow and hydropower generation data from 190 Norwegian catchments and 136 ROR plants, with precipitation, temperature and catchment attributes as input variables. The LSTM model outperforms more traditional hydrological models for predictions in both gauged and ungauged catchments. Furthermore, the combined LSTM model yields hydropower generation estimates that are comparable to or better than those from a model trained only on hydropower generation data, while producing considerably better streamflow estimates. Our approach highlights the added value of additional data sources for hydrological modeling for both local calibration and the task of regionalization, and demonstrates that data-driven methods are suitable for leveraging their potential.

Norsk Regnesentral har etablert en statistisk risikomodell for vannskader etter overvannsflom på bygninger i Norge. Modellen kobler forsikringsdata fra Gjensidige sammen med nedbørdata fra seNorge og annen lokal eksponeringsinformasjon. Vi finner at risikoen for vannskader lar seg beskrive gjennom sesongvise mål på mengde kraftig nedbør og avvik fra typisk kraftig nedbør. Kombinert med klimaframskrivninger levert av Norsk Klimaservicesenter simulerer modellen forventede endringer i skadebeløp fra referanseperioden 1991–2020 til to framtidige scenarioperioder under et lavt, middels og høyt utslippsscenario for CO2. På nasjonalt nivå antyder simuleringene en økning på opptil 33 % fram mot slutten av århundret. Skadeframskrivningene er følsomme for variabiliteten i klimaframskrivningene, og vi anbefaler å utvise forsiktighet med bruk av lave og høye kvantiler av endringene i skadebeløp på kommune- og fylkesnivå.

This report describes the research related to use case 1 within pilot ♯6 of the FAME project: Embedding Climatic Predictions in Property Insurance Products. For a dataset with house value statistics over a high-resolution grid over California (USA), a hedonic regression model was fitted that explains the median house value over a grid cell through factors like population density, median income, and ocean proximity. Additional variability can be explained by a component in the regression model that quantifies the reduction in house value due to frequent episodes with extreme heat at this grid cell. Daily mean temperature simulations from several regional climate models are then statistically further downscaled to the resolution of the house price grid, and the projected increase in the number of days per year with excessive heat over the next decades is studied. When combining this information with the fitted regression model, the associated projected decrease of house values can be calculated.

The stock assessment model SAM contains multiple age-dependent parameters that must be manually grouped together to obtain robust inference. This can make the model selection process slow, non-extensive and highly subjective, while producing unrealistic parameter estimates with discrete jumps. We propose to model age-dependent SAM parameters using spline functions, which can produce smoother parameter estimates, while making the model selection process faster, more automatic and less subjective. We develop a SAM spline model and compare it, using simulation studies and cross- and forward-validation methods, with published SAM models for 17 different fish stocks. The results show that our automated spline models overall outcompete the final accepted SAM models from stockassessment.org. We also demonstrate how our proposed spline model can be employed as a diagnostics tool for improving and better understanding properties of other SAM models.

Aiming to deliver improved precipitation simulations for hydrological impact assessment studies, we develop a methodology for modelling and simulating high-dimensional spatial precipitation extremes, focusing on both their marginal distributions and tail dependence structures. Tail dependence is crucial for assessing the consequences of extreme precipitation events, yet most stochastic weather generators do not attempt to capture this property. The spatial distribution of precipitation occurrences is modelled with four competing models, while the spatial distribution of nonzero extreme precipitation intensities are modelled with a latent Gaussian version of the spatial conditional extremes model. Nonzero precipitation marginal distributions are modelled using latent Gaussian models with gamma and generalized Pareto likelihoods. Fast inference is achieved using integrated nested Laplace approximations. We model and simulate spatial precipitation extremes in Central Norway, using 13 years of hourly radar data with a spatial resolution of 1 × 1 km2, over an area of size 6,461 km2, to describe the behaviour of extreme precipitation over a small drainage area. Inference on this high-dimensional data set is achieved within hours, and the simulations capture the main trends of the observed precipitation well.

Diurnal temperature range is an important variable in climate science that canprovide information regarding climate variability and climate change. Changesindiurnaltemperaturerangecanhaveimplicationsforhydrology,humanhealthand ecology, among others. Yet, the statistical literature on modeling diurnaltemperature range is lacking. In this article we propose to model the distri-bution of diurnal temperature range using the five-parameter lambda (FPL)distribution. Additionally, in order to model diurnal temperature range withexplanatory variables, we propose a distributional quantile regression modelthat combines quantile regression with marginal modeling using the FPL distri-bution. Inference is performed using the method of quantiles. The models arefitted to 30 years of daily observations of diurnal temperature range from 112weather stations in the southern part of Norway. The flexible FPL distributionshows great promise as a model for diurnal temperature range, and performswell against competing models. The distributional quantile regression model isfitted to diurnal temperature range data using geographic, orographic, and cli-matological explanatory variables. It performs well and captures much of thespatial variation in the distribution of diurnal temperature range in Norway.

Global circulation models (GCMs) are used for projecting climate changes on a global scale. However, when we need information for local climate changes, a dynamical downscaling through a regional climate model (RCM) may be used to gain more precise information. Therefore it is important to make good RCMs that are unbiased when projecting climate changes. This note investigates the skill of precipitation projections from five combinations of global and regional climate models from EURO-CORDEX and four bias correction methods applied to some of these. This is performed by comparing the model outputs with data from the E-OBS and NGCD data products using integrated quadratic distance.

Global climate models are used for projecting climate changes on a global scale. However, when we need information for local climate changes, a regional climate model with a finer grid that uses the global climate model as boundary conditions is necessary to gain more precise information. Therefore it is important to make good regional climate models that are unbiased when projecting climate changes. This note investigates the fit of precipitation projections from nine combinations of global and regional climate models from EURO-CORDEX and four bias correction methods applied to some of these. This is performed by comparing the models with data from the E-OBS dataset using integrated quadratic distance (IQD). All the climate models had difficulties with their projections in the areas of Fennoscandia with high amount of daily precipitation or long drought periods, while the IQD was improved for all climate models after bias correction, the IQD was still highest in the areas with more extreme data after bias correction. The LSCE-IPSL-CDFt-EOBS10-1971-2005 bias correction method obtained the best results for all our tests. However, this method was calibrated using the E-OBS dataset. As the other bias correction methods used different datasets the current results should be compared to an evaluation using alternative observation-based datasets.