Search

In this article we present a method for numerical simulations of extreme sea levels using synthetic low-pressure systems as atmospheric forcing. Our simulations can be considered to be estimates of the high sea levels that may be reached when a low-pressure system of high intensity and optimal track passes the studied region. We test the method using sites located along the Baltic Sea coast and simulate synthetic cyclones with various tracks. To model the effects of the cyclone properties on sea level, we simulate internal Baltic Sea water level variations with a numerical two-dimensional hydrodynamic model, forced by an ensemble of time-dependent wind and air-pressure fields from synthetic cyclones. The storm surges caused by the synthetic cyclones come on top of the mean water level of the Baltic Sea, for which we used a fixed upper estimate of 100 cm. We find high extremes in the northern Bothnian Bay and in the eastern Gulf of Finland, where the sea level extreme due to the synthetic cyclone reaches up to 3.5 m. In the event that the mean water level of the Baltic Sea has a maximal value (1 m) during the cyclone, the highest sea levels of 4.5 m could thus be reached. We find our method to be suitable for use in further studies of sea level extremes. [ABSTRACT FROM AUTHOR]

Accurate estimates of net ecosystem CO 2 exchange (NEE) would improve the understanding of natural carbon sources and sinks and their role in the regulation of global atmospheric carbon. In this work, we use and compare the random forest (RF) and the gradient boosting (GB) machine learning (ML) methods for predicting year-round 6 h NEE over 1996–2018 in a pine-dominated boreal forest in southern Finland and analyze the predictability of NEE. Additionally, aggregation to weekly NEE values was applied to get information about longer term behavior of the method. The meteorological ERA5 reanalysis variables were used as predictors. Spatial and temporal neighborhood (predictor lagging) was used to provide the models more data to learn from, which was found to improve considerably the accuracy of both ML approaches compared to using only the nearest grid cell and time step. Both ML methods can explain temporal variability of NEE in the observational site of this study with meteorological predictors, but the GB method was more accurate. Only minor signs of overfitting could be detected for the GB algorithm when redundant variables were included. The accuracy of the approaches, measured mainly using cross-validated R2 score between the model result and the observed NEE, was high, reaching a best estimate value of 0.92 for GB and 0.88 for RF. In addition to the standard RF approach, we recommend using GB for modeling the CO 2 fluxes of the ecosystems due to its potential for better performance. [ABSTRACT FROM AUTHOR]

Accurate estimates of the net ecosystem CO2 exchange (NEE) would improve the understanding of the natural carbon sources and sinks and their role in the regulation of the global atmospheric carbon. In this work, we use and compare the random forest (RF) and the gradient boosting (GB) machine learning (ML) methods for predicting the year-round 6 hourly NEE over 1996–2018 in a pine-dominated boreal forest in southern Finland and analyze the predictability of the NEE. Additionally, aggregation to weekly NEE values was applied to get information about longer term behavior of the method. The meteorological ERA5 reanalysis variables were used as predictors. Spatial and temporal neighborhood (predictor lagging) was used to provide the models more data to learn from, which was found to improve the accuracy compared to using only the nearest grid cell and time step. Both ML methods can explain the temporal variability of the NEE in the observational site of this study with the meteorological predictors, but the GB method was more accurate. It was more effective in separating the important predictors from non-important ones, showing no signs of overfitting despite many redundant variables. The accuracy of the GB (RF), here measured mainly using cross-validated Pearson correlation coefficient between the model result and the observed NEE, was high (good), reaching a best estimate value of 0.96 (0.94) and the root mean square value of 1.18 µmol m-2 s-1 (1.35 µmol m² s-1). We recommend using GB instead of RF for modeling the CO2 fluxes of the ecosystems due to its better performance. [ABSTRACT FROM AUTHOR]

The skill scores of the Extended-Range Forecasts (ERF) of the European Centre for Medium-Range Weather Forecasts (ECMWF) are still quite modest for the forecast weeks 3-6 in Northern Europe. As there are known stratospheric precursors impacting the surface weather with potential to improve ERFs, we aim to quantify the effect of these predictors and post-process the ERFs with them. During boreal winter the quasi-biennial oscillation (QBO) affects the stratospheric polar vortex; the easterly (westerly) QBO often coincides with weaker (stronger) than average polar vortex. Consequently, the weaker (stronger) than average stratospheric polar vortex is connected to negative (positive) Arctic Oscillation (AO) and colder (warmer) than average surface temperatures in Northern Europe. We developed a stratospheric wind indicator, SWI, based on the previous months' stratospheric wind observations and the phase of the AO during the following weeks. We demonstrate that there was a statistically significant difference in the observed surface temperature within the 3-6 weeks depending on the SWI at the start of the forecast. These temperature anomalies were underestimated by the ECMWF's reforecasts. When our new SWI was applied in post-processing the ECMWF's two-week mean temperature reforecasts for weeks 3-4 and weeks 5-6 in Northern Europe during boreal winter, the skill scores of those weeks were slightly improved. This indicates there is some room to improve the ERFs, if the stratosphere-troposphere links were better captured in the modelling. [ABSTRACT FROM AUTHOR]

Abstract: Climate change may have notable impacts on carbon cycling in freshwater ecosystems, especially in the boreal zone. Higher atmospheric temperature and changes in annual discharge patterns and carbon loading from the catchment affect the thermal and biogeochemical conditions in a lake. We developed an extension of a one‐dimensional process‐based lake model MyLake for simulating carbon dioxide (CO2) dynamics of a boreal lake. We calibrated the model for Lake Kuivajärvi, a small humic boreal lake, for the years 2013–2014, using the extensive data available on carbon inflow and concentrations of water column CO2 and dissolved organic carbon. The lake is a constant source of CO2 to the atmosphere in the present climate. We studied the potential effects of climate change‐induced warming on lake CO2 concentration and air‐water flux using downscaled air temperature data from three recent‐generation global climate models with two alternative representative concentration pathway forcing scenarios. Literature estimates were used for climate change impacts on the lake inflow. The scenario simulations showed a 20–35% increase in the CO2 flux from the lake to the atmosphere in the scenario period 2070–2099 compared to the control period 1980–2009. In addition, we estimated possible implications of different changes in terrestrial inorganic and organic carbon loadings to the lake. The scenarios with plausible increases of 10% and 20% in CO2 and dissolved organic carbon loadings, respectively, produced increases of 2.1–2.5% and 2.2–2.3% in the annual CO2 flux. [ABSTRACT FROM AUTHOR]

We examined the feasibility of long-range forecasts of temperature for needs of the energy sector in Helsinki, Finland. The work was done jointly by Finnish Meteorological Institute (FMI) and Helen Ltd, the main Helsinki metropolitan area energy provider, and especially provider of district heating and cooling. Because temperatures govern the need of heating and cooling and, therefore, the energy demand, better longrange forecasts of temperature would be highly useful for Helen Ltd. Heating degree day (HDD) is a parameter that indicates the demand of energy to heat a building. We examined the forecasted monthly HDD values for Helsinki using UK Met Office seasonal forecasts with the lead time up to two months. The long-range forecasts of monthly HDD showed some skill in Helsinki in winter 2015-2016, especially if the very cold January is excluded. [ABSTRACT FROM AUTHOR]

A method for estimating the occurrence of freezing rain (FZRA) in gridded atmospheric datasets was evaluated, calibrated against SYNOP weather station observations, and applied to the ERA-Interim reanalysis for climatological studies of the phenomenon. The algorithm, originally developed for detecting the precipitation type in numerical weather prediction at the Finnish Meteorological Institute, uses vertical profiles of relative humidity and temperature as input. Reanalysis data in 6-hourly time resolution was analyzed over Europe for the period 1979-2014. Mean annual and monthly numbers of FZRA events, as well as probabilities of duration and spatial extent of events, were then derived. The algorithm was able to reproduce accurately the observed, spatially averaged interannual variability of FZRA (correlation = 0.93) during the 36-year period, but at station level rather low validation and cross-validation statistics were achieved (mean correlation = 0.44). Coarse grid resolution of the reanalysis, and misclassifications to other freezing phenomena in SYNOP observations, such as ice pellets and freezing drizzle, contribute to the low validation results at station scale. Although the derived gridded climatology is preliminary, it may be useful, for example, in safety assessments of critical infrastructure. [ABSTRACT FROM AUTHOR]

ABSTRACT Global warming leads to a prolongation and intensification of the thermal growing season. In this study, we present projections for the growing season length and growing degree day sum ( GDD) in Europe by the end of the 21st century using two threshold temperatures, 5 and 10 °C. The analysis was based on simulations performed with 22-23 CMIP5 global models under the RCP4.5 and RCP8.5 scenarios. Systematic errors in the temporal mean and variability of modelled temperatures were eliminated, and the data were downscaled spatially by employing a bias-correction method. To determine the onset, termination and GDD of the growing season, two methods have been used. The previously developed Fourier method is suited for exploring long-term means, while the novel temperature deviation integral method is applicable to inter-annual variations. According to the multi-model mean of the RCP8.5 simulations in the late 21st century, for the majority of Europe the growing season is prolonged by 1.5-2 months, the GDD above 5 °C increasing by 60-100%. Responses to RCP4.5 are qualitatively similar but smaller. A decomposition of the uncertainty variance reveals that in the near-term future the contribution of internal variability is pronounced, but by the end of the century inter-model differences dominate. In studying growing-season conditions on an annual basis, we found that in coming decades years with a GDD below the recent past (1971-2000) mean become very uncommon. In the majority of years, GDD will exceed the 10-year or even the 20- or 50-year return level derived from recent past data. [ABSTRACT FROM AUTHOR]

Climate change projections for Finland have been calculated from simulations performed with 28 recent-generation (CMIP5) global climate models. During the next few decades, projected changes are fairly similar under all four RCP forcing scenarios examined. Conversely, in the second half of this century, the evolution of climate is highly dependent on greenhouse gas emissions. Under the high-emission RCP8.5 forcing scenario for the period 2040-2069, surface air temperatures in winter are projected to increase by 2-7°C relative to 1981-2010, while precipitation would increase by 4-30% and diurnal temperature range diminish by 5-34%. For incident solar radiation, the projected change falls between -17 and +2%. In summer warming is more modest, 1-4°C; for the other quantities, the sign of changes remains uncertain, even though minor increases are likely for both precipitation and solar radiation. For mean temperature and solar radiation, uncertainty in projected changes mainly stems from modelling differences, while for precipitation, the contribution of internal variability is prominent. Compared to the previous model generation, the new CMIP5 models simulate somewhat larger warming for summer, while the precipitation projections are nearly equal throughout the year. For surface wind speed, the multimodel-mean change is close to zero, in contrast to a slight increase inferred from the previous model generation. However, inter-model differences in the wind speed projections are large, up to ±15%. The geographical pattern of the modelled change can be downscaled spatially by using regional climate models. However, in the simulations based on a single regional model that were examined here, dynamical downscaling proved to lead to a substantial underestimation of the projection uncertainty for solar radiation and warm-season temperature. [ABSTRACT FROM AUTHOR]