Displays

Climate forecast skill for the El Nino-Southern Oscillation (ENSO) is better than chance, but has increased little in recent decades. Further, the relative skill of dynamical and statistical models varies in skill assessments, depending on choices made about how to evaluate the forecasts. Using a suite of models from the North American Multi-Model Ensemble (NMME) archive we outline the consequences for skill of how the bias corrections and forecast anomalies are formed. We show that the method for computing forecast anomalies is such a critical part of the provenance of a skill score that any score for forecast anomalies lacking clarity about the method is open to wide interpretation. Many assessments of hindcast skill are likely to be overestimates of attainable forecast skill because the hindcast anomalies are informed by observations over the period assessed that would not be available to a real forecast. The relative skill rankings of forecast models can change between hindcast and forecast systems because the impact of model bias on skill is sensitive to the ways in which forecast anomalies are formed. Dynamical models are found to be more skillful than simple statistical models for forecasting the onset of El Nino events.

How to cite: Squire, D. and Risbey, J.: Towards onset: shades of ENSO skill, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-6358, https://doi.org/10.5194/egusphere-egu2020-6358, 2020.

A multivariate linear inverse model (LIM) is developed to demonstrate the mechanisms and seasonal predictability of the dominant modes of variability from the tropical and South Pacific Oceans. We construct a LIM whose covariance matrix is a combination of principal components derived from tropical and extra-tropical sea surface temperature, and South Pacific Ocean vertically-averaged temperature anomalies. Eigen-decomposition of the linear deterministic system yields stationary and/or propagating eigenmodes, of which the least damped modes resemble the El-Niño Southern Oscillation (ENSO) and the South Pacific Decadal Oscillation (SPDO). We show that although the oscillatory periods of ENSO and SPDO are distinct, they have very close damping time scales, indicating the predictive skill of the surface ENSO and SPDO is comparable. The most damped noise modes occur in the mid-latitude South Pacific Ocean, reflecting atmospheric eastward-propagating Rossby wave train variability. We argue that these ocean wave trains occur due to the atmospheric high-frequency variability of the Pacific South American pattern imprinting onto the surface ocean. The ENSO spring predictability barrier is apparent in LIM predictions initialized in Mar-May (MAM) but nevertheless displays significant correlation skill of up to ~3 months. For the SPDO, the predictability barrier tends to appear in June-September (JAS), indicating remote but delayed influences from the Tropics. We demonstrate that subsurface processes in the South Pacific Ocean are the main source of decadal variability, and further that by characterizing the upper ocean temperature contribution in the LIM the seasonal predictability of both ENSO and the SPDO variability is increased.

How to cite: Lou, J., O'Kane, T., and Holbrook, N.: A Linear Inverse Model of Tropical and South Pacific Seasonal Predictability, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-2235, https://doi.org/10.5194/egusphere-egu2020-2235, 2020.

An alternative to dynamical seasonal prediction of European climate is statistical modeling. Statistical modeling is an appealing and computationally effective approach for producing seasonal forecasts by exploiting the physical connections between the predictand variable and the predictors. We assess the seasonal predictability of summer European 2m temperature (T2m) using canonical correlation analysis. Seasonal means of spring Soil Moisture (SM), Sea Level Pressure (SLP) and Sea Surface Temperature (SST) are used as predictors of mean summer T2m. For SSTs, we test the potential predictability of T2m using three different regions. These regions include what we define as: Extratropical North Atlantic (ENA), Tropical North Atlantic (TNA), and North Atlantic (NA). The predictability is explored in the ERA20c reanalysis and in comprehensive Earth System Model (ESM) fields. The results are provided for the European domain on a horizontal grid of 1°x1° degrees.

In order to identify the local T2m predictability related to the different predictor variables, we first built Univariate Linear Regression models, one for every predictor. The regression models are calibrated and validated during 1902-1950 and a prediction is provided for the periods 1951-1998, 1951-2004, and 1951-2008, respectively. The resulting correlation maps between the original and the predicted T2m anomalies showed that for the predictor variables SLP, SM, and SST_ENA the results of the experiments using ESM data share similar T2m predictability patterns with the results of the experiments using reanalysis data. Most prominent disagreements between the predictability patterns resulting from ESMs and from ERA20c refers to the T2m prediction that utilizes tropical SSTs. SM is identified as the most important predictor for the summer European temperature predictability.

The ERA20c data show that the SM predictor field can be used for the T2m prediction over most of our study region west of 15° E and that the ENA SSTs can be used for the prediction over Europe east of 15° E. The resulting gridded correlation coefficients vary between 0.3 and 0.5. These results are not sensitive to the prediction period and to the number of Canonical Coefficients used in the regression model. Our approach complements existing numerical seasonal forecast frameworks and can be implemented for ensemble prediction studies.

How to cite: Pyrina, M., Wagner, S., and Zorita, E.: Statistical prediction of 20th century European summer temperatures based on ERA20c reanalysis data, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-4571, https://doi.org/10.5194/egusphere-egu2020-4571, 2020.

Seasonal Climate Predictions (SCPs) represent a challenging intermediate field where aspects typical of the short-term weather forecasts and long-term climate projections interact. Skillful SCPs represent an essential tool to reduce societal vulnerabilities to the inter-annual climate fluctuation through short-term (i.e., next season) climate impact mitigation measures. This is especially true over areas characterized by large climate inter-annual variability as the Mediterranean basin, which is also traditionally characterized by a poor seasonal predictability.

The primary research question of present study is to assess the capability of two dynamical downscaling approaches to improve the seasonal inter-annual variability signal coming from the global-scale driving SCP system on the Mediterranean basin.

In this work the Weather Research and Forecasting model (WRF3.9.1.1) and the Regional Climatic Model (RegCM4.1) were nested into NCEP’s operational seasonal forecast model Climate Forecast System version 2 (CFSv2) to dynamically downscale seasonal predictions over Mediterranean basin.

Using the initial and boundary conditions of an ensemble of the CFSv2 we compare the capability of the two downscaling approaches on improving the large scale CFSv2 prediction of a climatological period of 22-cold seasons (December–February) during 1982–2002.

The SCP systems (WRF- and RegCM-based) consist on a double dynamical downscaling where a height-member lagged ensemble of 3-month CFSv2 climate predictions represent the common driving fields. Both the nested models dynamically downscales CFSv2 climate prediction from the original 100 km resolution to 60 km over a domain covering the Mediterranean basin and Central Europe. The first downscaling feeds a second downscaling performed over a domain centered over Central Italy with a resolution of 12 km.

Climate variables considered are: 2 m temperature, precipitation, geopotential height at different pressure levels and mean sea level pressure. Results will be discussed by means of mean bias spatial distribution, inter-annual anomaly variability reproduction and probabilistic hit-rate of anomalous seasons, through tercile plots and reliability diagrams of the above mentioned variables.

Preliminary results, considering the RegCM, identify temperature variability reproduction benefiting from the downscaling. At the same time, precipitation shows an improved spatial distribution patterns but not improved inter-annual variability representation if compared to the driving CFSv2 reference period climate predictions.

How to cite: Sangelantoni, L., Mazzarella, V., Ricchi, A., Ferretti, R., and Redaelli, G.: Toward a regional-scale seasonal climate prediction system over the Mediterranean basin: evaluation and comparison of RegCM- and WRF-based dynamical downscaling approaches, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-10557, https://doi.org/10.5194/egusphere-egu2020-10557, 2020.

As the scientific and societal interest in skillful decadal predictions grows, a lot of effort is currently put into the development and advancement of such prediction systems worldwide. Studies evaluating the skill of basic atmospheric quantities, such as e.g. surface temperatures, in those prediction systems are numerous. However, dynamical quantities are discussed only rarely. Also, there is a lack of investigations which assess the exclusive impact of the model’s resolution on the forecast skill. 

In this study, we address both these issues: we analyse a set of four quantities of the extratropical circulation (storm track, blocking frequencies, cyclone frequencies, windstorm frequencies) and compare the deterministic forecast skill for lead winters 2-5 within the German MiKlip prediction system of two different spatial resolutions. While the lower resolution (LR, atm: T63L47, ocean: 1.5° L40) shows common deficits in the climatological representation, e.g. an overly zonal extratropical storm track and a deficit in blocking frequencies over the North Atlantic and Europe, the higher resolution version (HR, atm: T127L95, ocean: 0.4° L40) counteracts these biases. In return, the deterministic decadal prediction skill, which is measured in terms of anomaly correlation, increases (statistically significant) with the increase in resolution for all four quantities. 

The improvements found in our study for the different metrics follow a physically consistent line of argument, and the areas of improved forecast skill are crucial regions for the genesis and intensification of synoptic weather systems over the North Atlantic and for their impact on Europe. Thus, we identified a significant improvement of the storm track skill along the North Atlantic Current (i.e., the source region of synoptic eddies), a downstream improvement of the cyclone frequency skill over the central North Atlantic (where the synoptic systems intensify), and finally improved skill of the cyclone, windstorm and blocking frequencies over the European continent (i.e., the impact area).

Not only is the skill improved with the increase in resolution (HR vs. LR), but also the HR system itself offers significant deterministic decadal forecast skill for the extratropical circulation metrics in large regions over the North Atlantic and Europe (HR vs. ERA-Interim) for the considered lead time of two to five winters. 

Our results are encouraging for the advancement of decadal prediction systems as they document that even small improvements in the bias of the model, through an increased spatial resolution and possibly a better representation of smaller scales, can have a substantial effect on the representation of dynamical processes and can ultimately lead to a significant improvement of the decadal prediction skill for extratropical features and extreme events.

How to cite: Schuster, M., Grieger, J., Richling, A., Schartner, T., Illing, S., Kadow, C., Müller, W. A., Pohlmann, H., Pfahl, S., and Ulbrich, U.: Improvement in the decadal prediction skill of the North Atlantic extratropical winter circulation through increased model resolution, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-5649, https://doi.org/10.5194/egusphere-egu2020-5649, 2020.

Decadal climate prediction can provide invaluable information for decisions made by government agencies and industry. Modes of internal variability of the ocean play an important role in determining the climate on decadal time scales. This study explores the possibility of using self-organizing maps (SOMs) to identify decadal climate variability with the ultimate goal of improving decadal climate prediction. SOM is applied to 11-year running mean winter Sea Surface Temperature (SST) in the North Pacific and North Atlantic within the Community Earth System Model 1850 pre-industrial simulation to identify patterns of internal variability in SSTs. Transition probability tables are calculated to identify preferred paths through the SOM with time.  Results show both persistence and preferred evolutions of SST depending on the initial SST pattern.  This method also provides a measure of the predictability of these SST patterns, with the North Atlantic being predictable at longer lead times than the North Pacific. In addition, decadal SST predictions using persistence and lagged transition probabilities are conducted.

How to cite: Gu, Q. and Gervais, M.: Exploring North Atlantic and North Pacific Decadal Climate Prediction Using Self-Organizing Maps, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-20876, https://doi.org/10.5194/egusphere-egu2020-20876, 2020.

The Pacific Decadal Oscillation (PDO) is the most prominent form of decadal variability over the North Pacific, characterized by its horseshoe-like sea surface temperature (SST) anomaly pattern. The PDO exerts a substantial influence on marine ecosystems, fisheries, and agriculture. Through modulating global mean temperature, the phase shift of the PDO at the end of the 20th century is suggested to be an influential factor in the recent surface warming hiatus. Therefore, determining the predictability of the PDO in a warming climate is of great importance. By analyzing future climate under different emission scenarios simulated by the Coupled Model Intercomparison Project phase 5 (CMIP5), we show that the prediction lead time and the associated amplitude of the PDO decreases sharply under greenhouse warming conditions. This decrease is largely attributable to a warming-induced intensification of oceanic stratification, which accelerates propagation of Rossby waves, shortening the PDO lifespan and suppressing its amplitude by limiting its growth time. Our results suggest that greenhouse warming will make prediction of the PDO more challenging, with far-reaching ramifications.  

How to cite: Li, S.: The Pacific Decadal Oscillation less predictable under greenhouse warming, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-6399, https://doi.org/10.5194/egusphere-egu2020-6399, 2020.

Polar Regions are viewed by many as "observational deserts", as in-situ measurements there are indeed scarce relative to other regions. The increasing availability of satellite observations is salutary but does not entirely solve the problem due to persistent uncertainties in the derived products. Climate models have been instrumental in completing the big picture. However, models are themselves subject to errors, some of which are systematic. How to take advantage of the respective strengths of observations and models, while minimizing their respective weaknesses? To illustrate this point, I will discuss how recent advances in data assimilation, model evaluation, and numerical modeling have enabled major progress in tackling important questions in polar research, such as: What are the causes of the recent Antarctic sea ice variability? What might the future of Arctic sea ice look like? How to improve the skill of seasonal sea ice predictions? How should the existing observational network be improved at high latitudes? What are the priorities in terms of sea ice modeling for climate change studies? By running through these cases, I will provide evidence for the emerging hypothesis that "the whole is greater than the sum of its parts": treating observations and climate models as two noisy instances of the same, but unknown truth, gives insights that would not be possible if each source was used separately.

How to cite: Massonnet, F.: Making an informed use of observations and climate models to advance understanding of past and future sea ice changes, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-14046, https://doi.org/10.5194/egusphere-egu2020-14046, 2020.

Statistical relationships between climate variables are good source of seasonal predictability, but can we trust them to be valid in the future? In two recent papers, we investigated the stationarity of some well-known lagged relationships. The predictors were Arctic sea surface temperatures (SSTs) and sea ice cover during autumn, and the predictands were the North Atlantic Oscillation (NAO) and European temperature in winter. The reason for studying these variables was that in recent decades, reduced sea ice and above-normal SSTs in autumn have often preceded negative NAO conditions and cold temperatures in Northern Europe in the following winter. When we looked further back in time, however, we found that the relationships between SST/ice and NAO/temperatures have been highly changeable and sometimes even the complete opposite to that seen recently. One key finding was that, according to two 20th century reanalyses, the strength of the negative lagged correlation between Barents Sea SST anomalies in fall and European temperature anomalies in winter after 1979 is unprecedented since 1900. An analysis of hundreds of simulations from multiple climate models confirms that the relationships vary with time, just due to natural climate variability. This led us to question the causality and/or robustness of the links between the variables and to caution against indiscriminately predicting wintertime weather based on Arctic sea ice and SST anomalies.

How to cite: Kolstad, E. W., Screen, J. A., and Årthun, M.: Nonstationary lagged relationships between the Arctic and the midlatitudes, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-11159, https://doi.org/10.5194/egusphere-egu2020-11159, 2020.

Advances in the development of seasonal forecast systems allow skillful predictions of the atmospheric flow in the extratropics. Recent studies have highlighted the importance of stratospheric processes in climate variability at seasonal time scales, while their representation and impact in seasonal prediction is yet to be understood. Here stratospheric variability and predictability in boreal winter are evaluated on the seasonal range, using multi-model retrospective forecasts initialised in November. A novel focus is adopted to assess troposphere-stratosphere coupling (i.e., the interaction between upper-tropospheric eddy heat flux and the stratospheric polar vortex) on the basis of the empirical relation derived by Hinssen and Ambaum (2010)^[1]. Results indicate that dynamical predictions perform better than persistence forecasts and show significant skill up to lead season one (December to February). We find that seasonal anomalies of stratospheric zonal-mean zonal wind in the extratropics are mostly explained by anomalous tropospheric eddy heat flux; the response to tropospheric wave forcing is weaker in models than in reanalysis. Furthermore, we demonstrate that skillful seasonal stratospheric forecasts benefit from residual predictability of the heat flux over the Pacific sector, while further improvements are limited by current unpredictability of the Eurasian heat flux on the seasonal time scale. Sources of long-term predictability are examined and reveal a potential influence of the QBO, Arctic sea ice, Eurasian snow cover and ENSO. This work is realised using data from the seasonal Copernicus Climate Change Service multi-model (November initialisations from 1993 to 2016) and from ERA-Interim reanalysis.

[1] Hinssen,  Y. B. L. and Ambaum,  M. H. P.:  Relation between the 100-hPa heat flux and stratospheric potential vorticity, J. Atmos.Sci., 67, 4017–4027, 2010.

How to cite: Portal, A., Ruggieri, P., Palmeiro, F., Garcìa-Serrano, J., Domeisen, D., and Gualdi, S.: Seasonal prediction of boreal winter stratosphere, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-739, https://doi.org/10.5194/egusphere-egu2020-739, 2020.

Multiyear to decadal predictability of the North Atlantic sea surface temperature (SST) is commonly attributed to buoyancy-forced changes of the Atlantic Meridional Overturning Circulation and associated poleward heat transport. Here we investigate the role of the wind stress anomalies in decadal hindcasts for the prediction of annual extratropical North Atlantic SST anomalies. A global climate model is forced by ERA-interim wind stress anomalies over the period 1979-2017. The resulting climate states serve as initial conditions for the decadal hindcasts. We find significant skill in predicting annual SST anomalies over the central extratropical North Atlantic with anomaly correlation coefficients exceeding 0.6 at lead times of 4 to 7 years. The skill of annual SSTs is basically insensitive to the calendar month of initialization. This skill is potentially linked to a gyre-driven upper-ocean heat content anomaly that leads anomalous SSTs by several years.

How to cite: Reintges, A., Latif, M., Bordbar, M. H., and Park, W.: Multiyear predictability of extratropical North Atlantic sea surface temperatures in hindcasts initialized with wind stress anomalies, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-7160, https://doi.org/10.5194/egusphere-egu2020-7160, 2020.

There is great interest from a wide range of stakeholders in near-term climate prediction ranging from seasonal to decadal timescales. While seasonal forecasting is done operationally since more than 20 years now, decadal climate prediction still has to be considered mainly a research subject. The vast majority of existing decadal prediction studies focusses on skill of temporally (typically multi-annual) averaged parameters. This is in line with the general understanding of climate prediction skill to be expectable only for low-frequency climate variability on larger spatial scales. However, while such predictions of multi-annual means might be skilful, they contain little information on shorter timescale extremes.

We present a different approach, that is the temporal pooling of seasonal means to form probabilistic forecasts. Thus, rather than for example analyzing the anomaly of the summer temperature averaged over a decade, we examine the probabilities of extreme seasonal summer temperatures within this decade (exceedance of some quantile of the climatological summer temperature probability distribution). This approach complements the common multi-annual means and hence extends the usability of decadal predictions.

For this study we use decadal climate predictions produced by the CMIP5 multi-model ensemble as well as available CMIP6-DCPP contributions. We analyze these large ensembles’ skill regarding forecasting the probability of extremely warm and extremely dry seasons. A season is considered to be “extreme” if the seasonal mean temperature (precipitation) is above (below) the 5^th (1^st) sextile of the climatological probability distribution.

We will show that the forecast skill in this respect is comparable to that obtained for the common approach, based on multi-annual year averages. This means the existence of significant skill for many regions globally when considering the probability of extremely warm temperatures. Skill regarding predicting extremely dry seasons (i.e. low precipitation) is rather limited, though.

These results generally agree with studies applying the common multi-annual averaging approach for assessing skill of temperature and precipitation climate predictions but extend the existing knowledge by covering probabilities of seasonal mean extremes. Hence, this approach states an important contribution towards the extended utility of decadal climate predictions. An additional benefit of the framework proposed here is the larger sample size when pooling instead of averaging. This allows to consider extreme events of higher magnitude before reaching the limitations of statistical uncertainty hampering the derivation of robust results.

How to cite: Kruschke, T., Befort, D., Nikulin, G., and Koenigk, T.: Multi-model decadal predictions of probabilities for seasonal mean temperature and precipitation extremes, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-17685, https://doi.org/10.5194/egusphere-egu2020-17685, 2020.

Reliable predictions of terrestrial water storage (TWS) changes for the next couple of years would be extremely valuable for agriculture and water management. Decadal predictions have already shown to be meaningful for predicting e.g. sea surface and air temperature, but have not yet been intensively investigated regarding TWS. Here we evaluate decadal hindcasts of TWS related variables from an ensemble of five CMIP5 (Coupled Model Intercomparison Project Phase 5) climate models against a TWS data set that is based on GRACE (Gravity Recovery And Climate Experiment) satellite observations.

As the overlap time span of 9 years for the model time series and GRACE observations is not long enough for a robust comparison, we also use a GRACE-based reconstruction of TWS utilizing precipitation and temperature data sets (Humphrey and Gudmundsson, 2019) available back to the year 1900. Thus we are able to compare the full 41 year (1970-2011) time span covered by CMIP5 decadal predictions to the TWS reconstruction. Correlations and root mean squared deviations (RMSD) are calculated for yearly global averages and for individual climate zones. Furthermore, we derive global maps of correlations and RMSD.

We find that at least for the first two prediction years the decadal model experiments clearly outperform the classical climate projections, regionally even for the third year. However, the spread among the models is large and absolute similarities between model output and GRACE TWS reconstructions are quite low.

We also perform a preliminary skill assessment for the first CMIP6 decadal hindcasts publicly available, finding a slightly reduced skill for the first forecast year in comparison to the CMIP5 models, while for the second forecast year an improvement is seen. This result is generally encouraging, but requires confirmation as soon as more CMIP6 decadal hindcasts become available.

Humphrey, V., Gudmundsson, L., 2019. GRACE-REC: a reconstruction of climate-driven water storage changes over the last century. Earth System Science Data Discussions 1–41. https://doi.org/10.5194/essd-2019-25

How to cite: Jensen, L., Eicker, A., Stacke, T., and Dobslaw, H.: Assessment of prediction skill for land water storage in CMIP5 models based on GRACE satellite observations, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-2605, https://doi.org/10.5194/egusphere-egu2020-2605, 2020.

Two machine learning (ML) models (Support Vector Regression and Extreme Gradient Boosting; SVR and XGBoost hereafter) have been developed to perform seasonal forecast for the winter (December–January–February, DJF) surface air temperature (SAT) in North America (NA) in this study. The seasonal forecast skills of the two ML models are evaluated in a cross-validated fashion. Forecast results from one Linear Regression (LR and hereafter) model and two Canadian dynamic climate models are used for the purpose of a comparison. In the take-one-out hindcast experiment, the two ML models and the LR model show reasonable seasonal forecast skills for the winter SAT in NA. Comparing to the two Canadian dynamic models, the two ML models and the LR model have better forecast skill for the winter SAT over the central NA which mainly get contribution of a skillful forecast of the second Empirical Orthogonal Function (EOF) mode of winter SAT over NA. In general, the SVR model and XGBoost model hindcasts show better forecast performances than LR model. However, the LR model shows less dependence on the size of the training dataset than SVR and XGBoost models. In the real forecast experiments during the period 2011-2017, compared to the two Canadian dynamic climate models, the two ML models clearly improve the forecast skill of winter SAT over northern and central NA. The results of this study suggest that ML models may provide real-time supplementary forecast tools to improve the forecast skill and may operationally facilitate the seasonal forecast of the winter climate of NA. 

How to cite: Qian, Q., Jia, X., and Lin, H.: Forecasting North America Winter Surface Air Temperature Using Machine Learning Methods, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-4465, https://doi.org/10.5194/egusphere-egu2020-4465, 2020.

Current state-of-the-art dynamical seasonal prediction systems still show limited skill, particularly over Europe in summer. To circumvent this, we propose a neural network-based classification of individual ensemble members at the initialisation of summer climate predictions, prior to performing a skill analysis. Different from European winter climate, largely dominated by the North Atlantic Oscillation, predictability of European summer climate has been associated with several physical mechanisms, including teleconnections with the tropics. Recent studies have shown that predictive skill improves when the dominant physical processes in a given season are identified at the initialisation of a prediction. Each of these dominant physical processes is associated with large-scale circulation patterns, often depicted by modes of Empirical Orthogonal Functions (EOF). We argue that Self-Organising Maps (SOM), a type of neural network classifier, can provide further insight on interpreting the predictive skill of mixed resolution hindcast ensemble simulations generated by MPI-ESM. This is achieved by identifying which circulation patterns over the North Atlantic-European sector (NAE) at the initialisation of hindcasts lead to more predictable states than others, their preferable transition states, and whether the spatial structure of each SOM mode play a role in shaping climate over Europe. We train SOM networks on sea level pressure fields of ERA-20C reanalysis at the initialisation of the seasonal prediction system (every May) for the period of 1900-2010, covering NAE. We compare the SOM-derived modes with circulation patterns derived from EOF analysis, and characterise each class of circulation regime. This analysis is used to distinguish classes of predictions with two different sets of MPI-ESM initialised simulations with 10 and 30 members, covering the period of 1902-2008 and 1982-2016, respectively. We then discuss the differences and advantages of performing a neural interpretation of the skill of an ensemble prediction, over traditional skill analysis.

How to cite: Carvalho Oliveira, J., Zorita, E., Baehr, J., and Ludwig, T.: Neural interpretation of European summer climate ensemble predictions, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-13849, https://doi.org/10.5194/egusphere-egu2020-13849, 2020.

The decadal climate prediction (DCP) is one of the major challenges addressed by the research community focused on climate studies during the last years. DCPs try to fill the gap between seasonal-to-interannual predictions and multidecadal-to-centennial climate change projections by taking advance of not only the forced climate change signal provided by boundary information, but also the initialization of the climate system components which exhibit longer memory, such as the ocean.

Climate modelling for DCP is a very expensive activity in terms of computing resources since many initialized experiments are needed to properly assess the predictive skill at such time scales. In the context of dynamical downscaling (DS), this problem becomes even more important. Hence, the aim of this study is to evaluate some output variables from the Decadal Climate Prediction Large Ensemble (DPLE) and to explore the issue of reducing the number of ensemble members in consideration to make DS more affordable. The DPLE is a set of decadal simulations carried out at NCAR by using the Community Earth System Model (CESM). The DPLE encompasses 62 decadal experiments initialized every year (from 1954 to 2015) for each of the 40 members of the ensemble. Despite the ensemble size, only 10 members provide an adequate set of variables with the proper time aggregation to run a regional model.

Results obtained from this study could be helpful for those researchers who decide to address the regional DCP through a DS approach. Because of high computing resources, conducting DS simulations is restricted to a small number of research groups or institutes which can afford that large investment. It potentially limits the progress on this important and relatively recent branch of the climate science.

ACKNOWLEDGEMENTS: JJRC acknowledges the Spanish Ministry of Science, Innovation and Universities for the predoctoral fellowship (grant code: PRE2018-083921). This research has been carried out in the framework of the project CGL2017-89836-R, funded by the Spanish Ministry of Economy and Competitiveness with additional FEDER funds.

How to cite: Rosa Cánovas, J. J., García Valdecasas-Ojeda, M., Yeste, P., Romero Jiménez, E., Gámiz Fortís, S. R., Esteban Parra, M. J., and Castro Díez, Y.: An evaluation of the CESM Decadal Climate Prediction Large Ensemble with application to regional studies, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-20801, https://doi.org/10.5194/egusphere-egu2020-20801, 2020.

The PARAMOUR project (Decadal Predictability and vAriability of polar climate: the Role of AtMosphere-Ocean-cryosphere mUltiscale inteRactions) is a new project funded in the framework of the Belgian program EOS - The excellence of Science. It aims at revealing fundamental drivers of climate variability and assessing the predictability in high-latitudes by using coupled regional climate models in both hemispheres. In this communication, we will present the ongoing contribution of the Earth and Life Institute in Louvain-la-Neuve (ELI, Belgium) and the Barcelona Supercomputing Center (BSC, Spain) to the PARAMOUR project, specifically in the Austral regions. The ELI and BSC efforts centre around two main objectives. The first one is improving our understanding of key processes that control the variability of the ice-ocean-atmosphere system at decadal timescales. The focus will initially be on the interactions between the components at regional scale and, later on, on the links with larger spatial scales. The second one is to determine how those interactions will lead to some predictability of the full ice-ocean-atmosphere system at decadal timescales or of some specific components only. Achieving our goals will require the development of coupled regional models including the atmosphere, ocean, sea ice and ice sheets, driven at their boundaries by the results of global models. Three configurations are proposed in the PARAMOUR project, covering 1/ Greenland, the Arctic and the North Atlantic sector, 2/ Antarctica and the Southern Ocean, 3/ The Totten glacier region. We will focus here on the latter two configurations, for the Austral regions. Retrospective (1980-2015) and prospective (2015-2045) climate simulations at high resolution will be conducted to evaluate the respective roles of initial conditions, some specific physical processes, teleconnections and couplings in the recent trends and to appreciate the potential fluctuations of key climate indicators within the next decades. A specific aspect will also be to determine the added-value of the regional models compared to the global ones.

How to cite: Moreno-Chamarro, E., Verfaillie, D., Goosse, H., Ortega, P., Fichefet, T., Massonnet, F., Klein, F., Pelletier, C., and Van Achter, G.: Investigating the climate predictability in the Southern Ocean using global and regional coupled models, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-3677, https://doi.org/10.5194/egusphere-egu2020-3677, 2020.

Seasonal forecasts are medium-range climate predictions that, used for calculating agroclimatic indicators, might potentially help land managers for best decision making. To assess their reliability seasonal forecasts are commonly contrasted against observed datasets, e.g. gridded data coming from reanalysis, classifying yearly pixel conditions in into/out of the norm events (i.e. using the 33^th and 66^th percentiles along a time series to define the occurrence of out of the norm events). Potential differences in the shape of the probability distribution across observed climate datasets might influence the results in the validation procedure of seasonal forecasting since the definition of out of the norm events depends on the properties of the statistical distribution. Here, we assess for different agroclimatic indicators related with water availability, vegetation thermal needs and fire risk, the spatial patterns of skewness using a range of climate datasets, i.e. ERA5, E-OBS and WFDEI along a 30 year period. Skewness represents the degree of asymmetry of the probability distribution evidencing locations in which out of the norm events highly differ from mean conditions which might suggest a potentially higher detectability. Common spatial patterns of great skewness (either positive or negative) across observed dataset might suggest areas with high and consistent detectability whereas contrasting patterns might suggest higher uncertainty for the validation procedure.

How to cite: Costa Saura, J. M., Bacciu, V., Mereu, V., Trabucco, A., and Spano, D.: Assessing consistency across climate datasets for the potential detectability of extreme events in seasonal forecasting using agroclimatic indicators, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-20631, https://doi.org/10.5194/egusphere-egu2020-20631, 2020.

The skill of decadal predictions in the North Atlantic region changed over time in the 20^th century. Recent work based on a single model argued that times of high skill – so-called windows of opportunity – could be identified for average North Atlantic SST by knowing the strength of meridional ocean heat transport in the subpolar North Atlantic at the start of a prediction.

Here, we verify these previous findings for the period 1970-2015 in several prediction systems of the Decadal Climate Prediction Project (DCPP) based on models used in the Coupled Model Intercomparison Project Phase 6 (CMIP6). We find windows of opportunity for decadal predictions of average North Atlantic SST in all examined prediction systems. The timing of these windows of opportunity generally agrees with the published estimate, indicating their robustness around the end of the twentieth century.

Decadal SST prediction skill in the North Atlantic Subpolar Gyre (SPG) shows much less consistent windows of opportunity between prediction systems than average North Atlantic SST. We explore model differences that explain these inconsistencies, discussing the spatial and temporal representation of North Atlantic ocean circulation and heat redistribution in the different prediction systems. We then show that connecting windows of opportunity to observable climatic variables such as sea surface height anomalies in the subpolar North Atlantic can constrain future skill estimates.

How to cite: Mignot, J., Borchert, L., Menary, M., Swingedouw, D., Sgubin, G., and Yeager, S.: Windows of Opportunity in Decadal Predictions of North Atlantic SST, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-3309, https://doi.org/10.5194/egusphere-egu2020-3309, 2020.

How to cite: Hellmich, L., Rautenhaus, M., Athanasiadis, P., Dobrynin, M., Düsterhus, A., Ruggieri, P., and Baehr, J.: The Northern Hemisphere Winter Polar Jet Stream and its Connection to the Seasonal Prediction Skill of Weather Regimes over Europe, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-8669, https://doi.org/10.5194/egusphere-egu2020-8669, 2020.

The predictability of spring (March–May) precipitation over East China is investigated in the present study based on the February-start hindcasts of eight coupled models from DEMETER and ENSEMBLES during 1960–2001. Five out of the eight models exhibit significantly increased predictability of central East China spring precipitation (CECSP) after the late 1970s. The mechanism analysis indicates that CECSP variability is closely related to a meridional dipole vorticity pattern at 200 hPa and southerly wind at 850 hPa over East Asia, whose prediction skill also increased significantly around the late 1970s, consistent with the changes in CECSP predictability. Observational analysis indicates that the sea surface temperature (SST) over the tropical Pacific and Indian Ocean experienced a notable decadal change around the late 1970s. After the decadal change, the tropical SST has an enhanced impact on the CECSP-related East Asian dipole vorticity pattern at the upper level and on the western North Pacific anticyclone at the lower level. The five models can adequately reproduce the observed enhanced connection between the tropical SST and East Asian atmospheric circulation after the late 1970s, consequently showing higher predictability of East Asian atmospheric circulation and CECSP. However, the other three models cannot reproduce the relationship between the tropical SST and East Asian atmospheric circulation; therefore, CECSP predictability in these models remains low during the entire period. The increased predictability is valuable for current dynamical seasonal prediction for central East China.

How to cite: Zhang, M. and Sun, J.: Increased predictability of spring precipitation over central East China around the late 1970s, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-8329, https://doi.org/10.5194/egusphere-egu2020-8329, 2020.

With seasonal forecast models we investigate whether it is possible to give the people in Tanzania, Peru and India time to adapt and prepare to different weather conditions. In recent years, these countries have repeatedly experienced devastating droughts or floods, such as in East Africa in November 2019.

Under the framework of the research project EPICC (East Africa Peru India Climate Capacities) supported by the BMU (Federal Ministry for the Environment, Nature Conservation and Nuclear Safety), we aim to set up a seasonal forecast system. The goal is to make the data useful for the hydrologists at the project partner from PIK (Potsdam Institute for Climate Impact Research) for integration in a tool for adaption in local agriculture in the affected countries (India, Peru and Tanzania). In this study, we validate a number of variables of predicted anomalies in seasonal forecast models as well as of a multimodel product.

There are different methods of seasonal predictability, based on slow variations of boundary conditions, coupled ocean-atmosphere model simulations as well as the concept of ensembles, multi-model ensembles and uncertainties. The focus in this study is on the intercomparison of the single models and the multimodel in a forecast range between 1 and 6 months. In particular, we investigate three-month mean deviation from the long-term mean. It is important for the population (especially for the agriculture industry) in the focus region to know whether in a certain period (rainy season, dry season, El Nino etc.) the next 3 months will be colder, warmer, drier or even wetter compared to the long-term mean.

Due to the fact, that various seasonal forecasting models perform differently, it is the challenge, to find the best fitting seasonal forecast model for each of the affected countries.

How to cite: Möller, T. and Gates, L.: Seasonal forecasting and the predictability of the rainy and dry seasons for Peru, Tanzania and India, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-8355, https://doi.org/10.5194/egusphere-egu2020-8355, 2020.

Intense precipitations events are associated with impacts like damages to infrastructures, economic activities, agricultural crops, power production and society in general. The ability to predict extreme precipitation events months in advance is therefore of great value in densely populated areas like the Mediterranean and may be achieved using seasonal prediction systems like the Copernicus Climate Change Services (C3S) suite of models. Using weather regimes (WRs) from 500 hPa geopotential heights over the Mediterranean the two main objectives of this study are: first to identify how these regimes are linked to extreme precipitation events over the region using reanalysis data; and second to assess the ability of the C3S models in reproducing/predicting these extreme events. We identify four weather regimes for the winter season (DJF) describing the atmospheric circulation in the Mediterranean using the 1993-2016 period as reference, i.e. maximum availability of C3S hindcasts. We thus provide an assessment of the models’s ability in predicting extreme precipitation over the Mediterranean having quantified how daily precipitation anomalies are associated to each WR.

How to cite: Giuntoli, I., Fabiano, F., and Corti, S.: Predictability of precipitation extremes over the Mediterranean, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-17020, https://doi.org/10.5194/egusphere-egu2020-17020, 2020.

Sahelian rainfall presents variability from internannual to interdecadal timescales, which is influenced by the sea surface temperature anomalies (SSTa) in different basins. At interannual times scales it has been shown that this variability depends on the SSTa over the equatorial Pacific, Atlantic and eastern Mediterranean. In this work we consider the set of models from the North American Multi-model ensemble (NMME) in order to analyze their skill in reproducing the Sahelian precipitation variability and relate it to their skill in reproducing the variability of the SSTa over the equatorial Pacific, equatorial Atlantic and eastern Mediterranean as well as their ability to simulate their teleconnections with Sahel rainfall.

Results show that the skill in predicting Sahel rainfall is low, decreases rapidly with lead time and is highly model dependent. Skill is improved for those models that are able to correctly simulate the Pacific SST - Sahel rainfall teleconnection.  Models present a good ability to reproduce the Mediterranean SST – Sahel teleconnection, and skill in Sahel rainfall prediction is more dependent on the correct prediction of the Mediterranean SST anomalies. These results suggest a path to increase skill in Sahel rainfall prediction.

How to cite: Martin-Gomez, V., Mohino, E., and Rodriguez-Fonseca, B.: Understanding Sahelian rainfall skill in the NMME seasonal forecast, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-19710, https://doi.org/10.5194/egusphere-egu2020-19710, 2020.

We analyse central equatorial Pacific inter-annual prediction skill of sea surface temperature (SST) and net primary productivity (NPP) using initialized retrospective forecasts with the Max Planck Institute Earth system model over the time period 1998-2014. We find significant NPP predictability for up to 5 lead years, which is far beyond the SST predictability of less than 1 year in this area. While El-Nino-Southern-Oscillation (ENSO) limits SST predictability, we find the origin of the high NPP prediction skill to be in the tropical upwelling zones of the eastern Pacific, i.e., the Peru-Chile current system offshore South America. Off-equatorial Rossby waves are initiated off the coast of Chile and travel towards the central tropical Pacific on a time scale of 4 to 5 years. On their arrival, the Rossby waves modify the depth of the nutricline, which is fundamental to the availability of nutrients in the euphotic layer in the central tropical Pacific.

We further demonstrate that the seasonal upwelling in the central equatorial Pacific, which is mainly driven by ENSO, transports nutrients, i.e. nitrate and phosphate, from below the nutricline into the euphotic zone, effectively transferring the Rossby wave signal from depth to the near-surface ocean. A shallower than normal nutricline leads to larger primary production, and vice versa, a deeper than normal nutricline to smaller primary production. The Rossby waves also modulate the SST, however, these changes are damped on the daily to weekly time scale due to surface heat fluxes at the atmosphere-ocean boundary. Therefore, the off-equatorial Rossby waves maintain the high predictability of NPP but not the SST. We conclude that NPP predictions in the central equatorial Pacific benefit from the memory contained in properly simulated off-equatorial Rossby waves.

How to cite: Brune, S., Caballero Espejo, M., Li, H., Ilyina, T., and Baehr, J.: Inter-annual predictability of net primary productivity in the central equatorial Pacific, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-11256, https://doi.org/10.5194/egusphere-egu2020-11256, 2020.

Anthropogenic fossil fuel emissions are increasing, and about a half of these emissions is absorbed by land and ocean. The CO2 fraction remaining in the atmosphere, the airborne fraction, is varying from year to year. Most of this variability can be explained by the land-atmosphere carbon fluxes. This variability is strongly affected by the El Niño – Southern Oscillation (ENSO); however, it is difficult to determine the cause of the flux anomalies due to the complex interactions between the climatic effects of the ENSO cycle. Here, we use MPI Earth System Model, MPI-ESM, to study the mechanisms of post El Niño carbon fluxes and assess their predictability. 10-member ensemble simulations with small perturbations are initialized at six El Niño events of a 1000-year control run. After removing the long-term mean from the ensemble simulations, a density-based clustering algorithm is applied to the carbon fluxes due to primary productivity, respiration and fires. This allows to identify and delimit the individual hotspots of ENSO-related carbon flux anomalies that contribute most to the atmospheric CO2 change.
We found that the main carbon sources are due to a reduction of primary production in the tropics, while the carbon sinks are due to reduced respiration or increased primary production in the extratropics. The potential predictability of the carbon fluxes from these clusters was assessed by using the perfect model approach. In accordance with this method, the predictive horizon is estimated as the time, when the variability within the ensemble members exceeds the long-term variability. As climate change will likely modify the frequency of El Niño events, this decomposition of the ENSO carbon flux anomalies could be used to improve our understanding of the future trends of land carbon sinks.

How to cite: Dunkl, I. and Brovkin, V.: Decomposing terrestrial carbon flux anomalies after El Niño: process-based predictability of land carbon sinks and sources , EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-10438, https://doi.org/10.5194/egusphere-egu2020-10438, 2020.

How to cite: Hempel, T., Düsterhus, A., and Baehr, J.: Seasonal prediction of the austral summer Southern Annular Mode, and investigation of its connection to the Southern Ocean, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-8885, https://doi.org/10.5194/egusphere-egu2020-8885, 2020.

Seasonal prediction systems based on comprehensive Earth System Models are capable of skillfully predicting the winter North Atlantic Oscillation. However, the predictive skill reported for these systems is accompanied by a potential inconsistency: The quality of the predictions measured over a set of retrospective forecasts and quantified by the correlation coefficient between prediction and observation exceeds expectations based exclusively on model properties. This discrepancy is commonly referred to as the signal-to-noise paradox (SNP)

Current investigations of the SNP are predominantly looking at seasonal predictions systems based on comprehensive Earth System Models, focusing the uncertainties in the model formulation. In the present contribution, we investigate the SNP in a simple conceptual framework of an ensemble prediction system based on the simple three dimensional Lorenz 1963 Model (L63). This framework enables us to separate the influence of uncertainties in the model initialization and uncertainties in the model formulation on the occurrence of the SNP.

We show that in the absence of uncertainties in the model formulation the SNP is not apparent in L63, if the uncertainty assumed for the initialization of the ensemble is equal to the observational uncertainty. However, if we assume that the uncertainty in the initialization systematically overestimates the observational uncertainty, the SNP is also apparent in L63 - even if there are no uncertainties in the model formulation itself.

While these results obtained in the conceptual framework cannot directly translated to the SNP in comprehensive Earth System Models, we suggest to include in further investigations of the SNP in Earth System Models also a comparison of the magnitude of the initial ensemble spread and the observational uncertainty.

How to cite: Mayer, B., Düsterhus, A., and Bahr, J.: The signal-to-noise paradox in a conceptual framework based on the 1963 Lorenz model, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-11613, https://doi.org/10.5194/egusphere-egu2020-11613, 2020.

The North Atlantic has a major influence on the climate of Europe. In the past, decadal prediction systems have shown consistent prediction skill in the North Atlantic for initialised models, indicating the potential to exploit this skill for better predictions on the continent. One prime area of potential for this approach is Ireland, due to its proximity to the Atlantic.

Until now, the prediction skill for the island of Ireland is limited, leading to the conclusion that dynamical models alone are not able to transfer the prediction skill from the North Atlantic to the surrounding land masses. Therefore, the project Aigéin, Aeráid, agus athrú Atlantaigh (A4) aims to establish new physical and statistical approaches to enhance the skill. This includes a better understanding on the oceanographic processes leading to the prediction skill in the North Atlantic as well as usage of statistical-dynamical predictions.

This contribution will give an overview of the approaches and a first look on the factors we anticipate to use in our analysis. Special attention will be given to the statistical approaches for the statistical dynamical prediction as well new verification procedures to evaluate them.

How to cite: McCarthy, G., Düsterhus, A., O'Beirne, C., Ogungbenro, S., Diabate, S. T., Caesar, L., Upton, M. C., Cahill, N., and Parnell, A. C.: New approaches to decadal predictions on the regional scale, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-13560, https://doi.org/10.5194/egusphere-egu2020-13560, 2020.