i. advancing the climate forecasts with new initialization and ensemble strategies as well as improved model physics of the earth climate system,
ii. post-processing raw model output (e.g., bias correction, (re)calibration, or downscaling with classic or machine-learning-based statistical methods),
iii. translating physical knowledge on local and remote physical drivers of predictability into tools to detect and indicate “windows of forecast opportunity” (e.g., subsampling or weighting of ensemble members or models),
iv. coupling raw model forecasts to impact models to support early warning systems and adaptation strategies (related to extreme events and hazards in the atmosphere, biosphere, and lithosphere, to health, or to energy).
Stakeholders in all socio-economic sectors require reliable forecasts at multiple timescales as part of their decision-making processes. Although basing decisions mostly on a particular timescale (e.g., weather, subseasonal, seasonal) is the present status quo, this approach tends to lead to missing opportunities for more comprehensive risk-management systems (Goddard et al. 2014).
Seamless operational climate prediction --ranging from hours to multiple decades-- is a sound theoretical possibility that in practice has not been fully realised yet. While today a variety of forecasts are produced targeting distinct timescales in a routine way, these products are generally presented to the users in different websites and bulletins, often without an assessment of how consistent the predictions are across timescales.
Because different models and strategies are used at different timescales by both national and international decadal, seasonal and subseasonal forecasting centers (e.g. Kirtman et al. 2014, Kirtman et al. 2017, Vitart et al. 2017,Mahmood et al., 2021), and skill is different at those timescales, it is key to guarantee that a physically consistent "bridging" between the forecasts exists, and that the cross-timescale predictions are overall skilful and actionable, so decision makers can conduct their work. As recent research suggests, forecasts at different timescales can be merged to mimic the idea of seamless forecasts (e.g. Befort et al., 2022).
Here, a method for merging forecasts across timescales (Muñoz et al., 2023) is discussed and illustrated. The approach is based on the identification of "bridges of opportunity" via a cross-wavelet spectral analysis of causal information flows between prediction systems. The new method enables to formally calculate the timing and length of the windows of opportunity, and the optimal temporal aggregation to conduct the bridging of the forecast systems. The approach is illustrated for operational (a) subseasonal-to-seasonal, and (b) seasonal-to-interannual forecast systems.
How to cite: Muñoz, Á. G.: Forecasting across timescales by crossing “bridges ofopportunity”, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-925, https://doi.org/10.5194/ems2024-925, 2024.
The El Niño–Southern Oscillation (ENSO) is a prominent interannual signal in the global climate system with widespread climatic influence. Our current understanding of ENSO predictability is based mainly on long-term retrospective forecasts obtained from intermediate complexity and hybrid coupled models. Compared with those models, complicated coupled general circulation models (CGCMs) include more realistic physical processes and have the potential to reproduce the ENSO complexity. However, hindcast studies based on CGCMs have only focused on the last 20–60 yrs. In this study, we conducted an ensemble retrospective prediction from 1881 to 2017 using the Community Earth System Model to evaluate El Niño–Southern Oscillation (ENSO) predictability and its variability on different timescales. To our knowledge, this is the first assessment of ENSO predictability using a long-term ensemble hindcast with a complicated coupled general circulation model (CGCM). Our results indicate that both the dispersion component (DC) and signal component (SC) contribute to the interannual variation of ENSO predictability (measured by relative entropy, RE). In detail, the SC is more important for ENSO events, whereas the DC is of comparable important for short lead times and in weak ENSO signal years. The SC dominates the seasonal variation of ENSO predictability, and an abrupt decrease in signal intensity results in the spring predictability barrier feature of ENSO. At the interdecadal scale, the SC controls the variability of ENSO predictability, while the magnitude of ENSO predictability is determined by the DC. The seasonal and interdecadal variations of ENSO predictability in the CGCM are generally consistent with results based on intermediate complexity and hybrid coupled models. However, the DC has a greater contribution in the CGCM than that in the intermediate complexity and hybrid coupled models.
How to cite: Liu, T.: ENSO predictability in the CESM model from 1880 to 2017, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-155, https://doi.org/10.5194/ems2024-155, 2024.
Long-lead climate predictions are increasingly in demand due to climate change. Through its well-known atmospheric teleconnections El Niño Southern Oscillation (ENSO) is a leading source of seasonal and interannual climate predictability, but currently operational ENSO forecasts are limited to about two seasons in advance. At the same time the scientific literature pointing to the feasibility of ENSO forecasts one year and even more in advance is increasing. The early anticipation of ENSO could prepare vulnerable communities around the world and help mitigate its most devastating impacts such as droughts, floods, poor harvests, and the spread of infectious diseases. Here we showcase real-time forecasts of the recent 2023/24 El Niño at lead times up to 1.5 years in advance of an expected peak in December 2023. We use a previously validated statistical ENSO model that relies on surface and subsurface ocean temperatures and zonal wind stress as predictors, and includes various dynamic components in the form of time-varying cyclical components. Real-time early forecasts with the same model were also issued for the 2015/16 and 2018/19 El Niños, when the long-lead forecasts were also coupled to an impact dengue fever model for the city of Machala in Ecuador to predict the probability of a dengue outbreak in the region up to 11 months in advance. In both cases the dengue predictions were successful indicating an outbreak in 2016 and a low dengue season in 2019. Therefore, such longer-lead ENSO forecasts could directly facilitate decision-making in the health sector, but also other key sectors of society.
How to cite: Petrova, D., Rodó, X., Koopman, S. J., Tzanov, V., and Cvijanovic, I.: Long-Lead ENSO Predictability: the 2023/24 El Niño, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-868, https://doi.org/10.5194/ems2024-868, 2024.
The implementation of adaptation policies requires seamless and relevant information on the evolution of the climate over the next decades. Decadal climate predictions are subject to drift because of intrinsic model errors and their skill may be limited after a few years or even months depending on the region. Non-initialized ensembles of climate projections have large uncertainties over the next decades, encompassing the full range of uncertainty attributed to internal climate variability. Providing the best climate information at regional scale over the next decades is therefore challenging. Recent studies have started to address this challenge by constraining uninitialized projections of sea surface temperature using decadal predictions or using a storyline approach to constrain uninitialized projections of the Atlantic Meridional Overturning Circulation using observations. Here, we develop a new method combining both observational and forecast information to provide the best climate information over the next decades. First, we select a sub-ensemble of non-initialized climate simulations based on their similarity to observed predictors with multi-decadal signal potential over Europe, such as Atlantic multi-decadal variability (AMV). Then, we try to further refine this sub-ensemble of trajectories by selecting a subset based on its consistency with decadal predictions. This study presents a comparison of these different methods for constraining surface temperatures in the Europe regions as defined in the IPCC over the next decades, focusing on CMIP6 non-initialized simulations. A retrospective evaluation is carried out over the historical period to assess the ability of this method to provide a good climate information in the near-term climate.
How to cite: Bonnet, R., Sanchez-Gomez, E., Boé, J., and Cassou, C.: Constraining near-term climate projections by combining observations with decadal predictions, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-125, https://doi.org/10.5194/ems2024-125, 2024.
We present the development of the initialisation strategy for seasonal to decadal climate predictions based on the ICON-XPP model, within DWD’s Innovation in the Applied Research and Development (IAFE) program. The ICON-XPP model is based on several well-established model components: the ICON-NWP, operational weather forecast model at the DWD, as atmospheric model, ICON-O as ocean model, JSBACH as land model and uses a hydrological discharge model. To develop a weakly coupled data assimilation system for ICON-XPP, we use the experience build on a former ICON-ESM version (Pohlmann et al 2023). In our weakly coupled data assimilation framework, we use two different assimilation methods for atmosphere and ocean. We initialize the ocean component of the climate system through a monthly assimilation of salinity and temperature profiles from the EN4 dataset. For this we use a localised singular evolutive interpolated Kalman filter implemented via the Parallel Data Assimilation Framework (PDAF, Nerger 2020). The atmosphere component is initialised by nudging temperature, pressure and horizontal wind fields of the ERA5 reanalysis. We conduct our experiments with a 25-member ensemble, which we put together from three different historical runs started from different states of our piControl run. In the atmosphere ICON-XPP is run as R2B5 (~80km resolution) with 130 vertical levels and in the ocean we use a resolution of ~40km (R2B6) with 72 vertical levels. Our assimilation experiments start from 1990 after a ten-year assimilation spin-up of the ocean. We show the results of our experiments with the weakly coupled data assimilation system and discuss its challenges.
How to cite: Sgoff, C., Pohlmann, H., Brune, S., Pham, T. V., Schneidereit, A., Steinert, T., and Fröhlich, K.: Weakly coupled data assimilation for climate predictions with ICON-XPP, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-950, https://doi.org/10.5194/ems2024-950, 2024.
Past observational studies and numerical experimentation suggest that spring soil moisture anomalies in the northern hemisphere can induce significant anomalies in local summer temperature and precipitation as well as in large-scale planetary wave patterns. This is particularly the case in regions where soil-moisture memory is long and there are strong soil-moisture-atmosphere interactions. However, recent studies, where soil-moisture initial conditions of dynamical forecasting systems are scrambled, have found more modest impacts suggesting that current seasonal forecasting systems may not accurately capture this source of potential predictability.
Here we present an evaluation of two important aspects in the dynamical models that contribute to the multi-model seasonal forecasting system of the Copernicus Climate Change Service (C3S). Firstly, assessing the skill of soil-moisture in seasonal hindcasts and secondly analysing the realism of soil-moisture-atmosphere coupling strength. Both of which need to be well represented to achieve accurate predictions. These links between soil-moisture, evapotranspiration and temperature are investigated by way of correlation metrics to identify and compare locations where evaporation is soil-moisture limited or energy limited in seasonal hindcasts and observations.
A key finding of this analysis is that in the summer hindcasts, initialised in May, tend to misrepresent the precise location of the so-called North America coupling “hotspot”, an important region for land-atmosphere coupling. The soil-moisture atmosphere coupling tends to be too weak in the models in the western USA and too strong in the east. Both the reasons for this misplacement and the impact of this on forecasts of temperature, precipitation and large-scale planetary waves will be investigated.
How to cite: Day, J., Stockdale, T., and Vitart, F.: How well do dynamical seasonal forecasts capture soil-moisture atmosphere coupling?, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-738, https://doi.org/10.5194/ems2024-738, 2024.
Certain atmospheric conditions provide windows of opportunity for sub-seasonal prediction, such that forecasts initialized at these times maintain good predictive skill further out into the future than usual. Such windows of opportunity are for example given by certain phases of the Madden-Julian oscillation and the state of the stratospheric polar vortex at forecast initialization.
These findings come from retrospective analyses of forecast skill that often investigate only one or two drivers of sub-seasonal predictability at a time. What is lacking is investigating multiple drivers and comparing their relative importance, which we propose to achieve by means of explainable machine learning (ML) techniques. Furthermore, whereas previous ML-based studies of atmospheric predictability quantified uncertainty in terms of deterministic error or ensemble spread as a proxy for forecast skill, we here aim to predict the probabilistic forecast skill itself. The result is an operational decision support tool that can inform users, at forecast initialization time, of the skill expected at sub-seasonal lead times.
Therefore we are presenting, for the first time to our knowledge, an explainable ML model (based on random forests) that learns to predict directly the skill of the ECMWF extended-range forecasts at several weeks’ lead time, given the atmospheric state at initialization. We use reanalysis data as ground truth and focus on the skill of weekly aggregated statistics of geopotential height at 500 hPa, a variable of particular interest to the renewable energy industry since it tracks the synoptic-scale weather patterns controlling European energy production and demand.
We expect that our forecast skill prediction model will enable enhanced risk management for sub-seasonal forecast users, such as stakeholders in the renewable energy industry, and thereby accelerate the transition to a decarbonized energy system. From a scientific perspective, interrogating our ML model with explainability techniques should yield new insights into the sources of sub-seasonal predictability.
How to cite: Mârza, A.-C., Domeisen, D. I. V., and Meyer, A.: Machine learning-driven assessment and prediction of sub-seasonal forecast skill, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-267, https://doi.org/10.5194/ems2024-267, 2024.
The North Atlantic Oscillation (NAO) represents the dominant mode of atmospheric circulation variability over the North Atlantic, driving winter weather conditions over a large part of the Euro-Mediterranean sector. Seasonal forecast systems have demonstrated some predictive skill for wintertime NAO, related to the enhanced ability of dynamical models to correctly represent possible sources of NAO predictability. However, increasing the predictive skill at the seasonal timescale over the European domain is still considered a major challenge.
In this work the aim is to extract the potential hidden skill in a dynamical seasonal forecast ensemble by properly selecting relevant realizations. The idea is to define a reduced ensemble better performing in terms of NAO predictions and to assess the performance of this subsample compared to the full ensemble.
Different subsampling criteria have been tested and verified. On the one hand, under the assumption that the ensemble average represents the most predictable path, the members simulating at the beginning of the forecast a NAO state closest to the ensemble mean NAO are selected. On the other hand, the realizations that resemble a reliable and independent estimate of the winter NAO, derived from the autumn conditions of a set of established dynamical predictors, are taken into consideration.
The comparison between the results obtained with the full ensemble and these subsampled ensembles reveals the potential for a significant improvement in the prediction of 2m temperature and precipitation anomalies, therefore representing a valuable strategy for possible real-time operational applications both in a multi-model and in a single model framework.
How to cite: Benassi, M., Athanasiadis, P., Borrelli, A., Cavicchia, L., Gualdi, S., Hashemi Devin, M., Sanna, A., and Tibaldi, S.: A NAO-based subsampling approach for winter seasonal prediction , EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-529, https://doi.org/10.5194/ems2024-529, 2024.
Accurate seasonal forecasts can be very useful for taking adaptation and prevention measures to unfavourable weather conditions, such as droughts, heatwaves, extreme precipitation events or high fire risk conditions. In a region with a very marked seasonal cycle and a high inter-annual variability of atmospheric conditions such as the Iberian Peninsula, the demand for improved seasonal forecasting becomes even more compelling.
Currently operational seasonal forecasting systems have considerable scope for improvement in their ability to predict mid-latitude temperature or precipitation. However, some of these systems have shown some skill in predicting, months in advance, the phase of some modes of variability such as the North Atlantic Oscillation.
Here we show that seasonal forecast of surface variables (such as 2m air temperature or total precipitation) can be improved by taking advantage of the models' skill in predicting the main modes of variability in Europe. First, a quantification of the theoretical potential for improving the forecasts is carried out using an ensemble-member weighting technique that increase the statistical weight of those members whose variability mode configuration is closer to the observed configuration. Then, we assess the actual potential for improving the forecasts by using a two steps prediction: (1) a multi-system forecast of the corresponding variability mode configuration and (2) a seasonal forecast of the surface variables in which we use the first step forecast of the variability mode configuration to weight the ensemble members. Different verification metrics were used to quantify the skill of the forecasts, both deterministic and probabilistic, all defined as in the WMO forecast guidance.
The results show that ensemble-member weighting method with information on variability modes is a window of forecast opportunity that is able to improve seasonal forecasts if further research is conducted.
How to cite: Senande-Rivera, M., Domínguez-Alonso, M., and Rodríguez-Guisado, E.: Improving seasonal forecasts for the Iberian Peninsula through climate variability modes, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-1011, https://doi.org/10.5194/ems2024-1011, 2024.
In the context of climate change adaptation, the use of climate predictions is steadily gaining importance. One critical challenge to consider when dealing with climate simulation outputs is the systematic bias affecting the modelled data. While bias correction methods are commonly employed in impact models to assess the effect of climate events on human activities, their effectiveness is often reduced in the case of extreme events, due to the scarcity of data for these low-probability and high-impact phenomena.
This study, conducted as part of the European project FOCUS-Africa, is dedicated to advancing innovative climate services in the southern regions of Africa. Our primary objective is to respond to the needs of risk assessment studies, focusing on the impact of extreme events and their implications for climate change adaptation. To this end, we designed a novel bias correction method to consistently correct extreme events of temperature and precipitation, but is adaptable to other climate variables, such as wind speed. Our approach conceptually extends one of the classic Quantile Mapping (QM) methods by improving the description of the tail ends of the distribution through a generalised extreme value distribution (GEV) fitting. Our methodology also incorporates a downscaling component. QM is indeed frequently both as a bias correction method and for downscaling simulations to finer observed scales. Therefore, our method not only corrects the climate data but also enhances the raw resolution of the model outputs (typically around 100 km) to match the 9 km grid of the observational reference.
In this study, we applied our technique to daily mean temperature and total precipitation data from three seasonal forecasting systems: SEAS5, System7, and GCFS2.1, developed respectively by ECMWF, Météo-France, and DWD. The bias correction efficiency was tested over the Southern African Development Community (SADC) region, which includes 15 Southern African countries. The performance was verified by comparing each of the three models with a reference dataset, the ECMWF reanalysis ERA5-Land. The results reveal that this novel technique significantly reduces the systematic biases in the forecasting models, yielding further improvements over the classic QM. For both the mean temperature and total precipitation, the bias correction produces a decrease in the Root Mean Squared Error (RMSE) and in the bias between the simulated and the reference data. After bias correcting the data, the ensemble forecasts members that correctly predict the temperature extreme increases. On the other hand, the number of members identifying precipitation extremes decreases after the bias correction, highlighting the challenge of obtaining robust statistics due to the lack of information about extreme events.
How to cite: Trentini, L., Venturini, M., Guerrini, F., Dal Gesso, S., Calmanti, S., and Petitta, M.: An advanced bias correction technique for improved Seasonal Forecasts: a focus on extreme events in Southern Africa, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-604, https://doi.org/10.5194/ems2024-604, 2024.
Climate change has a noticeable impact on the Mediterranean region, causing changes in the seasonal patterns and yearly variations of important meteorological variables along with the occurrence, strength, duration, and timing of unusual and extreme events. While the Mediterranean is recognized as one of the prominent global hot spot for climate impacts, it's imperative to dissect its sub-regions more meticulously. These areas exhibit significant meteorological shifts due to climate change and merit classification as unique hotspots. The Mediterranean Hotspot Index (MED-HOT index) is utilized to identify hotspot sub-regions in the Mediterranean region, followed by a detailed assessment of the seasonal data on these regions.
MED-HOT index is designed to assess regional climate extremes by simultaneously analyzing changes in frequency and intensity of key climatic variables, specifically precipitation and temperature. The MED-HOT index offers a comprehensive assessment of the major climate challenges in the Mediterranean, including heatwaves, extreme rainfall, and drought events, pinpointing regions requiring necessitating immediate attention and intervention . It calculated utilizing ERA5 daily maximum and minimum temperatures as well as precipitation data spanning into two time periods between 1981 to 2020 across 30 Mediterranean subregions. The analysis of the MED-HOT index reveals that Mediterranean hotspot region identification primarily relied on changes in the frequency of extremes, as the contribution of intensity changes was less important. Notably the change in extreme maximum temperature appears to play a crucial role in the most subregions. On an annual basis, the primary hot spots in the Mediterranean are identified as northern Greece, southern Italy, and the southeastern Mediterranean.
Moreover, this study examined the performance of ECMWF-SEAS5 in predicting extreme temperature and precipitation in the Mediterranean hot spot regions. It assessed the quality of seasonal temperature and precipitation forecasts over the subregions by analyzing predictions from all members of the hindcasts and forecasts ensemble across the two seasons (winter and summer). The analyses included seasonal mean fields, anomaly correlations, statistical indicators, and seasonality index. ECMWF-SEAS5 successfully replicated the extreme precipitation anomalies observed in recent decades, with effectiveness varying based on lead time and subregion. Nevertheless, the system exhibited only moderate predictive ability in regions with medium and low predictability.
Acknowledgments
The work was supported by PREVENT project. This project has received funding from Horizon Europe programme under Grant Agreement No: 101081276.
How to cite: Anagnostopoulou, C., Lazoglou, G., Papadopoulos-Zachos, A., Georgiades, P., Velikou, K., Manios, E. M., and Zittis, G.: Seasonal Data Evaluation in MED-HOT Index Hotspots: A Climatological Perspective on Mediterranean Subregions, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-580, https://doi.org/10.5194/ems2024-580, 2024.
The evolution of the atmosphere on weekly to sub-seasonal time scales is driven by a combination of factors from different compartments of the Earth system. At these time scales, the atmospheric predictability can arise from the initial state of the atmosphere, the land surface and the ocean. The weight of these different sources and their joint contribution can change quickly depending on the season and the time scale considered. As an illustration, while subseasonal forecasts often exhibit limited skill across mid-latitudes, occasional improvements are observed in specific locations during certain periods, known as "windows of opportunity." Understanding these windows is complex due to the diverse and interdependent nature of predictors, their spatial and temporal variability, and the challenges in establishing causality relationships. A typical strategy could consist of assessing the time-lagged correlation or regression with potential predictors. The main caveat of this approach is that even a lagged relationship between two variables X and Y does not ensure causality, because of the many confounding factors involved in complex earth system interactions. Furthermore, a high correlation between X(t) and Y(t+dt) may just be the consequence of Y(t) causing X(t). In this study, we introduce a novel approach based on Liang-Kleeman information flow, allowing the assessment of causal links across various lead times. Applied to reforecast and reanalysis data, our method identifies significant predictability drivers, revealing their evolving patterns and prevalence from seasonal to subseasonal scales. Additionally, the comparison between reanalysis and reforecast results aids in assessing the capability of models to capture these causality features. We will illustrate the theoretical background by showcasing the causal factors influencing a window of opportunity identified from a multimodel subseasonal reforecast.
How to cite: Ardilouze, C., Muñoz, Á., and Materia, S.: A causality framework to decipher prediction windows of opportunity, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-156, https://doi.org/10.5194/ems2024-156, 2024.
The subseasonal predictability of heatwaves in Europe is relatively well understood regarding prediction skill horizon and physical drivers of predictability. Despite this progress, few studies have translated subseasonal model output into skillful operational heatwave forecast products and end-user-tailored impact forecasts. These are substantial challenges, given the relatively high uncertainties and the flow-dependent skill inherent in subseasonal prediction. In this project, we aim to translate subseasonal model output from the European Centre for Medium-Range Weather Forecasts (ECMWF) into end-user-tailored heatwave forecast products for Switzerland. We first perform a detailed verification of average subseasonal (hindcast) prediction skill for temperature and heatwaves in Switzerland on different spatial and temporal aggregation scales. This analysis demonstrates a significant increase in subseasonal forecast skill with increasing temporal aggregation scales. We then analyze to what extent previously-studied local and remote drivers (such as dry soils, lower-frequency atmospheric modes, or sea surface temperature anomalies) manifest as “windows of forecast opportunity” for heatwave prediction in Switzerland and its subregions. These steps are performed with two sets of hindcasts – one with the native grid resolution and one that has been downscaled (and bias-corrected) to a higher resolution using quantile mapping. This postprocessing helps to quantify the added value of downscaling at subseasonal lead times. Finally, we present some ideas on how the gained knowledge on spatio-temporal and flow-dependent characteristics of skill could be translated into an operational subseasonal heatwave prediction system for Switzerland – a step that is closely linked to the challenging question of how much skill is enough skill for specific end-user applications.
How to cite: Büeler, D., Pyrina, M., Imamovic, A., Moret, L., Spirig, C., Lehning, M., and Domeisen, D. I. V.: Towards operational subseasonal heatwave prediction in Switzerland, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-657, https://doi.org/10.5194/ems2024-657, 2024.
Seasonal prediction is a field of research attracting growing interest beyond the scientific community due to its strong potential to guide decision-making in many sectors (e.g. agriculture and food security, health, energy production, water management, disaster risk reduction) in the face of the pressing dangers of climate change.
Among the various techniques being considered to improve the predictive skill of seasonal prediction systems, increasing the horizontal resolution of the climate models is a promising avenue. There are several indications that higher resolution versions of the current generation of climate models might improve key air-sea teleconnections, decreasing common biases of global models and improving the skill to predict certain regions at seasonal scales, e.g. in tropical sea surface temperature.
In this study, we analyze the differences in the predictive skill of two different seasonal prediction systems, based on the same climate model EC-Earth3 and initialized in the same way but using two different horizontal resolutions. The standard (SR) and high resolution (HR) configurations are based on an atmospheric component, IFS, of ~100 km and ~40 km of resolution respectively and on an ocean component, NEMO3.6, of ~100 km and ~25 km respectively. We focus in particular on the Tropical Pacific region where statistically significant improvements are found in HR with respect to SR for predicting ENSO and its associated climate teleconnections. We explore some processes that can explain these differences, such as the simulation of the tropical ocean mean state and atmospheric teleconnections between the Atlantic and Pacific tropical oceans.
A weaker mean-state bias in the HR configuration, with less westward extension of ENSO-related SST anomalies, leads to better skill in ENSO regions, which can also be linked to better localization of the atmospheric teleconnection with the equatorial Atlantic Ocean. It remains to be assessed if similar improvements are consistently identified for HR versions in other forecast systems, which would prompt their routine use in seasonal climate prediction.
How to cite: Carreric, A.: Comparing the seasonal predictability of ENSO and the Tropical Pacific variability in EC-Earth3 at two different horizontal resolutions, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-752, https://doi.org/10.5194/ems2024-752, 2024.
Quality and high-resolution seasonal forecast is a key aspect for planning and adapting strategies of different socio-economic sectors, providing knowledge of seasonal anomalies a few months ahead (Buontempo et al., 2018). These downscaled forecast can be used as input to impact models (e.g. crop or hydrology), developing climate services applications. However, the skill of seasonal forecasts is limited over mid-latitudes, as a consequence of the limited predictability at seasonal scale (Doblas-Reyes et al., 2013). The potential showed in postprocessing methods to improve the skill of seasonal forecast and its high case-dependent on region, season or application (Manzanas et al., 2020) have been the motivation to carry out this work.
A statistical downscaling method developed by AEMET (Petisco, 2008a; Petisco 2008b; Amblar, 2017) have been applied to different runs of ECMWF- SEAS5, covering winter, spring and summer periods, on a domain centered over the Iberian Peninsula. The twenty-years (1997-2016) hindcast (25-members) have been considered as model-climatological reference. A very deep evaluation process of the method had been published with satisfactory results (Hernanz et al. 2022a, d, e). The algorithm makes successive use of an analogue technique -based on a Euclidean distance- and multivariate regression to downscale maximum and minimum temperature and precipitation at daily timescale, through a selection of large-scale model circulation variables (predictors) linked to the local observed variable of interest (predictand). The method uses a high resolution observational gridded dataset developed by AEMET (Peral et al., 2017) (0.05º), covering the peninsular Spain and the Balearic Islands.
We have obtained results for mean seasonal temperature and accumulated seasonal precipitation for the follow metrics: the forecasted anomaly, the lower and upper forecasted probabilities and the AreaROC for lower and upper terciles. They show improved spatial detail of the probability of occurrence compared to raw SEAS5 and high values of ROC area (spatially and in percentage), allowing to conclude that at least in certain seasons and over the Iberian Peninsula, the downscaling algorithm developed by AEMET provides added value to ECMWF-S5 seasonal forecasts.
How to cite: Domínguez-Alonso, M., Senande-Rivera, M., Pérez-Pérez, F. J., and Rodríguez-Guisado, E.: Can statistical downscaling improve the skill of Seasonal Forecast over Iberian Peninsula? , EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-1042, https://doi.org/10.5194/ems2024-1042, 2024.
Being able to predict meteorological droughts several weeks ahead would add value to many sectors including agriculture, river shipping as well as water and energy management. A commonly used meteorological drought index is the standardized precipitation index SPI-N, which puts precipitation anomalies of the past N months into a climatological perspective. The SPI correlates with anomalies of soil-moisture, streamflow or groundwater storage, and thus serves as an inexpensive and attractive hydrological proxy. In this study we quantify how well the SPI-N can be skillfully forecasted in Switzerland. Using ECMWF IFS extended-range forecasts quantile mapped from its native 36 km to a 2 km grid, we produce ensembles of SPI-N forecasts for the Swiss drought warning regions. While previous research has underlined the challenges faced by ensemble forecasting systems in accurately predicting daily precipitation in Europe beyond lead week 1, our analysis reveals that the skill of SPI-1, SPI-3, and SPI-6 forecasts extends into weeks 3 and 4. It generally holds that skill SPI-6 > skill SPI-3 > skill SPI-1. For example, we find that the skill of an SPI-3 forecast for week 4 is comparable to the skill of an SPI-1 forecast for week 2. Overall, the results indicate the potential for skillful prediction of meteorological drought on sub-seasonal timescales. We link the extended predictability horizon to the inherent characteristics of the SPI being a temporal aggregate: the SPI is less sensitive to the exact timing of precipitation events, while also retaining “memory” of past precipitation. The latter manifests in larger skill for longer accumulation time N, in which more observation are weighted into the forecasted SPI. Finally, we show how SPI forecasts and hydrological forecasts are devised as factors for the combined drought indicator, which forms the numerical basis of the new Swiss drought early warning system.
How to cite: Imamovic, A., Büeler, D., Pyrina, M., Humphrey, V., Spirig, C., and Domeisen, D.: Skillful Extended-Range Forecasts of Standardized Precipitation Indices for Drought Early Warning in Switzerland, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-162, https://doi.org/10.5194/ems2024-162, 2024.
Dynamical forecasts use full three-dimensional climate models to simulate potential changes in the atmosphere and ocean over the next few months based on current conditions. The ensembles of simulations provide probabilistic weather scenarios that indicate the likelihood of a given period being wetter, drier, warmer, or colder compared to the seasonal average. The added value of various downscaling approaches for seasonal forecasts is a topic of frequent debate. This work focuses on statistical downscaling, which is based on empirical relationships derived between a local observed predictand of interest (summer temperature in this case) and one or several suitable model predictors from global seasonal forecasting systems.
Unlike tropical regions, seasonal predictability in Europe remains limited. This study analyses the seasonal forecast systems available in the Copernicus Climate Change Service (C3S) archive, which provide near-surface air temperature data at 1° to 1° spatial resolution. This study examines the statistical downscaling of summer air temperature forecasts for Central Europe from two weather forecast systems: the European Centre for Medium-Range Weather Forecasting (ECMWF) SEAS5.1 system and the Météo-France 8 (MF) system.
The study analyses the period 1993-2016, which is the longest hindcast period common to all systems, and the domain of the Czech Republic in Central Europe (47-52°N, 11-20°E). To perform statistical downscaling using the neural network method in STATISTICA software, we tested air temperature and sea surface pressure from global forecast models as predictors. The reference data used in this study are gridded E-OBS observational air temperature data sets. The best neural networks found are tested on forecasts for the period 2020–2023.
How to cite: Kliegrová, S., Metelka, L., Solanská, J., and Štěpánek, P.: Use of Neural Networks for Statistical Downscaling of Long-term Air Temperature Forecasts for the Czech Republic, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-414, https://doi.org/10.5194/ems2024-414, 2024.
Ensemble precipitation forecasts with sub-seasonal lead times provide useful information for decision-makers when they sufficiently sample the possible outcomes of trajectories. In this study, we present a forecasting tool for extreme precipitation ensemble forecast over Europe using a stochastic weather generator (SWG) based on analogs of the atmospheric circulation. This approach is tested for sub-seasonal lead times (from 2 to 4 weeks) to forecast European precipitation and temperature as well as the Madden Jullian Oscillation (Krouma et al, 2022,2023). SWG ensemble forecasts yield promising probabilistic skill scores for shorter and sub-seasonal timescales for precipitation (Krouma et al., 2022,2024) as well as for temperature (Yiou and Déandréis, 2019).
An updated version of the SWG, HC-SWG forecasting tool (HC refers to Hindcast and SWG to the stochastic weather generator) based on a combination of dynamical and stochastic models, was used to forecast European precipitation for the sub-seasonal lead time (Krouma et al., 2024, in review, QJRMS). The HC-SWG is based on analogs of the S2S model of the ECMWF and CNRM ensemble members 5 days ahead. We obtained reasonable forecast skill scores at the station level with respect to climatology. And we found that the HC-SWG shows improvement against the ECMWF precipitation forecast until 25 days.
In this work, we aim to use the HC-SWG to generate an ensemble of 100 members for extreme precipitation over Europe at the station level (Stockholm, Madrid, Paris..). We evaluate the ensemble forecast of the HC-SWG and we compare the HC-SWG forecast with other precipitation extreme forecasts to further confirm the advantage of our method.
How to cite: Krouma, M. and Messori, G.: Ensemble forecast of extreme precipitation in Europe by combining a stochastic weather generator with dynamical models , EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-28, https://doi.org/10.5194/ems2024-28, 2024.
Decadal climate predictions are our most reliable source of climate information for the near-term climate change. They can be very useful to develop climate services to users from sectors such as agriculture, industry, energy, water management, health, etc. However, before developing these climate services, an assessment of the forecast quality for the variables and regions of interest is needed. In this work, we consider the available retrospective decadal predictions (DCPP-A) contributing to phase 6 of the Coupled model Intercomparison Project (CMIP6) and analyze their skill to predict rainfall over Europe.
Rainfall variability over Europe is strongly linked to variations in the North Atlantic jet stream. At multidecadal time scales, some studies have shown that the variability of the North Atlantic jet stream could be driven by low-frequency changes in the sea surface temperature over the north Atlantic associated with the Atlantic Multidecadal Variability. Retrospective decadal predictions present skill to predict the low-frequency variability of the SST associated with Atlantic Multidecadal Variability. However, they have difficulties in predicting precipitation variability over Europe. Here, we analyze the skill of DCPP-A models to predict precipitation over Europe, exploring the multimodel ensemble to identify the models with enhanced skill, investigating as well the underlying reasons. In particular, we look at the link between the local precipitation skill and the representation of key teleconnection mechanisms that connect European precipitation to other remote highly-predictable regions.
Results from this study will contribute to a better understanding of the multidecadal European precipitation variability and will define benchmarks for improving its prediction in decadal forecast systems.
How to cite: Martin-Gomez, V. and Ortega, P.: Multi-model Physically-based constrained decadal predictions of precipitation over Europe , EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-191, https://doi.org/10.5194/ems2024-191, 2024.
In disaster prevention, water management, agriculture, and hydropower generation, an accurate seasonal streamflow forecast (SSF) is crucial, while global approaches become necessary in regions lacking forecast systems. This study evaluates the Météo-France seasonal prediction system (SYS8) skill for global SSF through hindcasts of river discharges. Contributing to Copernicus Climate Change Services (C3S), the SYS8 employs a fully coupled Atmosphere-Ocean General Circulation Model (AOGCM) with an advanced river routing component (CTRIP) interacting with the ISBA land-surface scheme. This research is part of the European project CERISE, which aims to enhance the C3S seasonal forecast portfolio by improving land initialisation methodologies.
SYS8 derives land initial conditions from a historical coupled initialisation run where land-river is weakly constrained, while atmosphere/ocean is nudged to the ERA5/GLORYS re-analysis. This study improves the initialisation run by relaxing soil moisture to fields reconstructed from an offline land simulation. Daily streamflow ensemble hindcasts of 25 members are generated in a 0.5° grid, with a lead time of up to 4 months initialised on the 1st of May/August/November between 1993-2017, allowing hindcasting summer (JJA), fall (SON) and winter (DJF) seasons. Forecast skill is assessed against discharge observations in 1608 monitored basins worldwide (with areas > 3000 km²) using deterministic and probabilistic metrics. The classical Ensemble Streamflow Prediction approach (ESP) is a benchmark for evaluating the control SYS8 skill and the additional skill of moisture nudging.
Globally, the control SYS8 skill is superior to the ESP, but the bias is higher in dry regions such as northeastern Brazil, western US and some rivers in Spain and Africa. On the other hand, the hindcast with enhanced land surface initial conditions outperformed the control SYS8 and benchmark ESP, especially during summer. Local skill degradation in higher latitudes will be discussed. Still, overall positive results support ongoing efforts to enhance land initialisation through a global land data assimilation system.
How to cite: Narváez Campo, G. F. and Ardilouze, C.: Global Streamflow Seasonal Forecast by a novel two-way AOGCM/Land/River coupling, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-158, https://doi.org/10.5194/ems2024-158, 2024.
The Copernicus Climate Change Service Evolution (CERISE) project, as an EU-HORIZON project, aims to enhance the quality of the C3S (Copernicus Climate Change Service) reanalysis and seasonal forecast portfolio, with a focus on land-atmosphere coupling. CERISE will develop new and innovative ensemble-based coupled land-atmosphere data assimilation approaches and land surface initialisation techniques to pave the way for the next generations of the C3S reanalysis and seasonal prediction systems. Deutscher Wetterdienst is developing its land data assimilation for the initialisation of seasonal forecasts with ICON-XPP. ICON-XPP (ICON eXtended Predictions and Projections) is not only the state-of-art climate modelling system in Germany but also an effort to unify knowledge and experiences from many institutes in one model system that can deliver seamless weather and climate simulations. We aim to include snow analysis, soil moisture analysis and leaf area index assimilation into our climate forecast data assimilation system. Here we present a baseline version of multi-year historical forecasts, so-called hindcasts for a period of 1993 to 2022. The current system to generate initial conditions consists of an Ensemble Kalman Filter for the ocean while nudging is applied in the atmosphere. Further, annually changing ESA-CCI land cover replaces the LUH2 data during this time, which is processed into plant functional types (PFTs) for use with the land model JSBACH. Results from these first seasonal hindcasts using the ICON-XPP model will be shown and discussed. Of special interest are the start months for the boreal spring and autumn forecasts, as the impact of land data assimilation is expected to be strongest.
How to cite: Noll, N., Romanova, V., Sgoff, C., Fröhlich, K., and Geppert, G.: ICON-XPP in the CERISE project: a first set of seasonal hindcasts, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-700, https://doi.org/10.5194/ems2024-700, 2024.
Climate Services based on seasonal forecasts are a powerful tool for adaptation in a changing climate and they attract growing interest from different sectors. However, operational seasonal forecasts have traditionally been issued following a subjective procedure, combining information from different sources, such as observation, empirical and dynamical models. Although it adds value by incorporating expert knowledge, the subjective procedure usually results in graphic products, with limited traceability, and not suitable for objective skill assessment or coupling sectoral applications. Identifying this issue, WMO encourages Regional Climate Centers and RCOFs to develop an objective procedure. The purpose is to increase the reliability of our results and to provide the basis for future climate services. With that aim, we explored ways of developing an objective approach that adds value to raw model forecasts in the Mediterranean region.
As is usually accepted, the starting point is a multimodel ensemble, which in our case combines seven Copernicus seasonal forecast models, hoping to minimize the weaknesses of individual models. The work focuses on looking for ways of subsampling the ensemble data based on comparing observational patterns with the evolution of ensemble members at the beginning of the period. Therefore, we did not use the latest model run, choosing instead earlier initializations and applying techniques such as cluster analysis or subsampling a fixed number of members to select those that were closer to reality.
First, we performed a cluster analysis to the ensemble forecast for winter (DJF) 2023-2024 and chose the cluster which best predicted the values of precipitation in October, which would be the last month with complete data when producing the winter seasonal forecast. However, we did not find a significant increase in skill in the Mediterranean region, possibly due to the great differences in cluster population between each year of the hindcast.
Then, we tested an alternative method by selecting a fixed number of members for the forecast and each year of the hindcast. We subsampled the group of members which best predicted precipitation in October and found a significant increase in skill in certain areas. However, there were not consistent improvements along the whole region, with some areas showing lower skill.
A comparison of the methodology using different model runs was conducted, finding better performance for the September run.
How to cite: Pérez Pérez, F. J. and Rodríguez Guisado, E.: Subsampling members in a seasonal forecast ensemble, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-1101, https://doi.org/10.5194/ems2024-1101, 2024.
Climate services that rely on the provision of climate forecasts at sub-seasonal, seasonal or decadal time scales (S2S2D) are widely exploited these days. To make them helpful for decision-making, state-of-the-art climate forecast model outputs are tailored to user needs. The conjunction of scientific knowledge and scientific exploration to fulfil users' needs determines the post-processing workflow, from selecting the datasets to visualising the product.
Consequently, scientists need to perform different combinations of possible workflows to evaluate the quality of the final products, ensuring the results are meaningful and that best practices (such as cross-validation strategies) are followed. For instance, essential climate variables (ECVs) can be calibrated before computing an indicator, or the indicator based on ECVs could be calibrated instead. Furthermore, several combinations of forecast systems and observation-based datasets can be explored to provide the most convenient product. On the other hand, the required computational resources can be a limitation depending on the total data size involved in the post-processing workflow.
To efficiently and flexibly handle all the requirements when exploring and delivering climate services based on S2S2D predictions, the Earth Sciences department of the Barcelona Supercomputing Center has developed the SUbseasoNal to decadal climate forecast post-processIng and asSEssmenT suite, so-called SUNSET. This suite takes advantage of existing software packages and tools (such as startR, CSTools, s2dv, CSDownscale and Autosubmit) to facilitate the definition of the workflow and its parallelisation on HPC machines when available.
To tailor climate products for each application and sector (e.g. agriculture, energy, water management, or health), the scientist can decide on the post-processing required steps, such as region selection, regridding method and resolution, anomaly calculation, and downscaling and bias-adjustment methods. SUNSET also allows the creation and visualisation of climate forecast products, such as maps for the most likely tercile. It performs the verification of the products using deterministic and probabilistic metrics, which can be visualised with maps and summarised with scorecards. The storage of final numerical outputs is also designed to consider the need to be ingested by third-party applications or impact models.
SUNSET is available in a public repository licensed under the GPLv3 license, and it is under continuous development. The suite is being used in scientific projects such as CERISE, where the next generation of the Copernicus Climate Change Service seasonal climate forecasts are under development.
How to cite: Pérez-Zanón, N., Agudetse, V., Palma, L., Ho, A.-C., Delgado-Torres, C., Milders, N., Duzenli, E., Llabrés-Brustenga, A., de Paula Kinoshita, B., Bretonnière, P.-A., and G. Muñoz, Á.: SUNSET: SUbseasoNal to decadal climate forecast post-processIng and asSEssmenT suite, EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-361, https://doi.org/10.5194/ems2024-361, 2024.
In early 2023, the European project CERISE started, under the auspices of the Copernicus Climate Change Service (C3S), devoted to enhancing the quality of the C3S reanalysis and seasonal forecasts with a focus on land-atmosphere coupling and land surface initialization. Studying snow cover is an important part of the CERISE project as it affects surface energy budget and hydrology, leading to shifts in atmospheric circulation and potential remote climate impacts. The impact of snow initialization on (sub-)seasonal atmospheric forecasts has received renewed attention in recent years.
In this study, a first attempt is made to analyze snow-atmosphere coupling in seasonal forecasts produced by the European Center for Medium-range Weather Forecasts (ECMWF) as the phase 0 demonstrator in the CERISE project. In phase 0, the atmospheric and land initial conditions (including snow) were taken from the ERA5 and the experiments were run with Integrated Forecasting System (IFS) Cycle 48R1.1. The snow initial conditions in ERA5 comprise assimilation of Interactive Multi-Sensor Snow and Ice Mapping System (IMS) satellite observations and in-situ station data. The 4-month forecasts consist of 51 ensemble members, and 4 start dates are available per year (in months 2, 5, 8, and 11).
To assess the snow-atmosphere coupling, we map the temporal correlation between the daily snow depth and near-surface temperature over winters 2000 – 2022. Based on this (and other) metrics we identify regions of snow-atmosphere coupling. These “cold spots” are mainly situated in the snow transition regions at mid-latitudes, where the snow cover is highly variable. We further discuss the role of snow in large-scale land-atmosphere interactions. The accuracy and fidelity of the snow forecasts is also examined against satellite observations and re-analyses. The satellite observations include snow cover fraction and/or snow water equivalent products from the ESA-CCI (European Space Agency – Climate Change Initiative) dataset.
How to cite: Vorobeva, E. and Orsolini, Y.: First results of the snow-atmosphere coupling analysis in the CERISE project , EMS Annual Meeting 2024, Barcelona, Spain, 1–6 Sep 2024, EMS2024-169, https://doi.org/10.5194/ems2024-169, 2024.
