AS1.20

This study proposes an objective methodology to highlight windows of opportunity related to a precursor phenomenon in a numerical subseasonal forecasting system. The methodology is based on a contingency table and is illustrated with the relationship between the Madden-Julian oscillation (MJO) and heavy rainfall in the tropical band. As a slowly propagating signal of enhanced convection, the MJO may indicate favorable conditions for heavy precipitation a few weeks ahead in some tropical areas. The combined knowledge of these climatological impacts and the current phase of the MJO at initialization defines observation-based "climatological windows of opportunity". In a second step, we analyze whether S2S forecasts are indeed more performant when there is increased climatological likelihood of heavy rainfall, i.e whether the forecasts convert "climatological windows of opportunity" into "model windows of opportunity".

The methodology is implemented to the prediction of the upper quintile of weekly precipitation in 20 years of ECMWF S2S reforecasts in the November-to-April season. The ability of the ECMWF forecasts to convert periods with more predictable events into periods of actual forecast skill is only verified for a limited number of small areas, while failures to seize the opportunities lie in misplaced MJO impacts, signal loss or too many false alarms.

How to cite: Specq, D. and Batté, L.: Do subseasonal forecasts take advantage of windows of opportunity related to a precursor phenomenon?, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-2609, https://doi.org/10.5194/egusphere-egu22-2609, 2022.

Sub-seasonal forecasts are challenging for numerical weather prediction (NWP) and machine learning models alike. Predicting temperature with a lead-time of two or more weeks requires a forward model to integrate multiple complex interactions, like oceanic and land surface conditions that might lead to recurrent or persistent weather patterns. The representation of the relevant interactions is imperfect in NWP models, just as our physical understanding of them. Model predictability can therefore deviate from real predictability for poorly understood reasons, hindering future progress.

This paper combines NWP with machine learning to detect and resolve such imperfect representations. We post-process ECMWF extended range forecasts of high summer temperatures in Europe with a shallow artificial neural network (ANN). Predictors are objectively selected from a large set of atmospheric, oceanic and terrestrial sources of predictability from ERA-5 and ECMWF re-forecast output. In the proposed architecture, the ANN learns to ‘update’ a prior ECMWF-given probability of two-meter temperature exceeding a given threshold. Due to the architecture of the network the magnitude of each correction, like increasing underestimated probabilities, can be attributed to specific predictors at initialization- or forecast-time. We interpret the circumstances in which substantial corrections are made. This reveals, e.g., that a tropical west Pacific sea surface temperature pattern is connected to high monthly average European temperature at a two-week lead-time. This teleconnection pattern is underestimated by the dynamical model and by correcting for this bias the ANN-based post-processing can thus improve forecast skill. We further find that the method does not readily increase skill when applied to other combinations of lead-time, averaging period and threshold, possibly due to non-stationarity in the data, lack of real predictability or lack of a re-forecast set of sufficient length.

How to cite: van Straaten, C., Whan, K., Coumou, D., van den Hurk, B., and Schmeits, M.: Improving sub-seasonal forecasts by correcting missing teleconnections using ANN-based post-processing, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-1686, https://doi.org/10.5194/egusphere-egu22-1686, 2022.

Across East Africa, sub-seasonal rainfall variability predominately depends on the phase of the Madden Julian Oscillation (MJO). Rainfall is enhanced during MJO phases 2 to 4, and suppressed during phases 6 to 8. Given that MJO-induced anomalous precipitation can persist beyond several days, a significant surface response is expected. Using earth observations and reanalysis data, this work illustrates how MJO-induced precipitation anomalies promote a surface response which feeds back onto local and regional atmospheric conditions.

                MJO-induced rainfall suppression across East Africa decreases surface soil moisture across semi-arid regions including southern South Sudan, western Kenya and northern Uganda. In regions predominately covered in grass and cropland, reduced soil moisture increases surface sensible heat fluxes and elevates land surface temperatures. A drier and warmer surface promotes an increased boundary-layer height and reduces surface pressure. We identify that spatial variations in the surface response to MJO-induced anomalous precipitation, impacts the intensity of the Turkana jet. Across southern South Sudan and in the exit region of the Turkana jet, reduced soil moisture increases land and near-surface temperatures, whilst in north-east Kenya and in the entrance region of the jet, no land surface temperature response is observed. The difference in surface response between the jet entrance and exit regions increases the pressure gradient along the Turkana channel, and thus intensifies the jet. Since the intensity of the Turkana jet controls the transportation of moisture from low-lying regions of East Africa into Central Africa, we highlight that surface-induced variations in jet intensity impacts rainfall totals across East Africa. Furthermore, due to the Turkana jet response to spatial variations in surface warming, we also identify that the magnitude of MJO-induced anomalous precipitation is influenced by surface conditions prior an MJO event. For example, when the surface over southern South Sudan is anomalously dry, MJO-induced precipitation suppression is greater. This presentation will highlight that to fully exploit predictability from the MJO, forecast models must correctly represent surface processes and land-atmosphere interactions. Future work evaluating sub-seasonal forecast models and improving the representation of land-atmosphere interactions will enhance the lead-time of early warning systems.

How to cite: Talib, J., Taylor, C., Wainwright, C., and Harris, B.: MJO-induced land-atmosphere feedbacks across East Africa, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-1790, https://doi.org/10.5194/egusphere-egu22-1790, 2022.

How to cite: Williams, N., Scaife, A., and Screen, J.: Weak ENSO teleconnections contribute to the signal-to-noise paradox, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-3880, https://doi.org/10.5194/egusphere-egu22-3880, 2022.

Tropical Cyclones (TCs) from the Eastern Pacific (EP) and North Atlantic (NA) Ocean commonly make landfall in the continental landmass of Middle America. These meteorological phenomena not only can cause floods and socioeconomic impacts, but they can also transport such heavy amounts of water that one event refill lakes, rivers, aquifers, and dams up to 100% during periods of prolonged droughts in arid and semiarid regions. This water resource can be used for agricultural and livestock activities, which are essential for Mexico and Central America’s countries. That is why local decision-makers are interested in having seasonal forecasts of tropical cyclone activity for the region. Current seasonal forecasts of tropical cyclone activity only focus on providing a number of TCs for the whole basin. However, local decision makers need information about possible affected regions at least 2 months in advance of the TC season peak (July-August-September for EP and August-September-October for the NA). This work is aimed at exploring a statistical-dynamical method for creating a seasonal forecast of TCs for Middle America.  We track TC-like vortices in five Coupled Global Models:  ECMWF, Météo-France, UKMO, DWD and CMCC during the 1993-2015 period (climatology period) and using two initial conditions: 1st July and 1st August for a three-month forecast. Our preliminary results show that three of the five models have the skill to adequately forecast the standardized track density anomaly and the TC activity per tercile (above-normal, normal, and below-normal) over the EP and NA basins. However, most of the models overestimate the activity. as indicated by the Brier Score (BS) and the Ranked Probability Skill Score (RPSS). Additionally, we present a statistical analysis of the type of tracks that are more important for the region and discuss how these types of tracks can be predicted depending on ENSO phase. We conclude that some models are useful to predict the TC activity 3 months in advance (dynamical approach), which can be combined with a statistical approach to provide more information about the type of TC track and possible affected regions.

How to cite: Vargas Hernandez, F. and Dominguez Sarmiento, C.: Seasonal Tropical Cyclone Forecasts for the Middle America, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-1398, https://doi.org/10.5194/egusphere-egu22-1398, 2022.

The last few years have seen an ever growing interest in weather predictions on sub-seasonal time scales ranging from 2 weeks to about 2 months. By forecasting aggregated weather statistics, such as weekly precipitation, it has indeed become possible to overcome the theoretical predictability limit of 2 weeks, bringing life to time scales which historically have been known as the “predictability desert”. The growing success at these time scales is largely due to the identification of weather and climate processes providing sub-seasonal predictability, such as the Madden-Julian Oscillation (MJO) and anomaly patterns of global sea surface temperature (SST), sea surface salinity, soil moisture and snow cover. Although much has been gained by these studies, a comprehensive analysis of potential predictors and their relative relevance to forecast sub-seasonal rainfall is still missing.

At the same time, data-driven machine learning (ML) models have proved to be excellent candidates to tackle two common challenges in weather forecasting: (i) resolving the non-linear relationships inherent to the chaotic climate system and (ii) handling the steadily growing amounts of Earth observational data. Not surprisingly, a variety of studies have already displayed the potential of ML models to improve the state-of-the-art dynamical weather prediction models currently in use for sub-seasonal predictions, in particular for temperatures, precipitation and the MJO. It seems therefore inevitable that the future of sub-seasonal prediction lies in the combination of both the dynamical, process-based and the statistical, data-driven approach. 

In the advent of this new age of combined Neural Earth System Modeling, we want to provide insight and guidance for future studies (i) to what extent large-scale teleconnections on the sub-seasonal scale can be resolved by purely data-driven models and (ii) what the relative contributions of the individual large-scale predictors are to make a skillful forecast. To this end, we build neural networks to predict sub-seasonal precipitation based on a variety of large-scale predictors derived from oceanic, atmospheric and terrestrial sources. As a second step, we apply layer-wise relevance propagation to examine the relative importance of different climate modes and processes in skillful forecasts.

Preliminary results show that the skill of our data-driven ML approach is comparable to state-of-the-art dynamical models suggesting that current operational models are able to correctly model large-scale teleconnections within the climate system. The ML model achieves highest skills over the tropical Pacific, the Maritime Continent and the Caribbean Sea, in agreement with dynamical models. By investigating the relative importance of those large-scale predictors for skillful predictions, we find that the MJO and processes associated with SST anomalies like the El Niño-Southern Oscillation, the Pacific decadal oscillation and the Atlantic meridional mode all play an important role for individual regions along the tropics.

How to cite: Rieger, N., Corral, Á., Olmedo, E., Magnusson, L., Ferranti, L., Pappenberger, F., and Turiel, A.: Identifying relevant large-scale predictors for sub-seasonal precipitation forecast using explainable neural networks, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-7530, https://doi.org/10.5194/egusphere-egu22-7530, 2022.

The simulated Northern Hemisphere stationary wave (SW) field is investigated in 11 subseasonal-to-seasonal (S2S) models. It is shown that while most models considered can well-simulate the stationary wavenumbers 1 and 2 during the first two weeks of integration, they diverge from observations following week 3. Those models with a poor resolution in the stratosphere struggle to simulate the waves, both in the troposphere and the stratosphere, even during the first two weeks, and biases extend from the troposphere all the way up to the stratosphere. Focusing on the tropospheric regions where SWs peak in amplitude reveals that the models generally do a better job in simulating the Northwest Pacific stationary trough, while certain models struggle to simulate the stationary ridges both in Western North America and the North Atlantic. In addition, a strong relationship is found between regional biases in the stationary height field and model errors in simulated upward propagation of planetary waves into the stratosphere. In the stratosphere, biases mostly are in wave-2 in those models with high stratospheric resolution, whereas in those models with low resolution in the stratosphere, a wave-1 bias is evident, which leads to a strong bias in the stratospheric mean zonal circulation due to the predominance of wave-1 there. Finally, biases in both amplitude and location of mean tropical convection and the subsequent subtropical downwelling, are identified as possible contributors to biases in the regional SW field in the troposphere.

How to cite: Schwartz, C., Garfinkel, C., Domeisen, D., Yadav, P., and Chen, W.: Stationary Wave Biases and Their Effect on Upward Troposphere - Stratosphere Coupling in Sub-seasonal Prediction Models, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-10898, https://doi.org/10.5194/egusphere-egu22-10898, 2022.

The Madden-Julian Oscillation (MJO) is one of the leading sources of tropical and extra-tropical predictability on subseasonal-to-seasonal timescales, but numerical models often suffer from systematic errors in capturing the MJO dynamics. Large-scale convection associated with the MJO is initiated over the Indian Ocean and propagates eastward across the Maritime Continent (MC) and into the western Pacific. As an MJO event enters the MC, it often weakens or completely dissipates due to complex interactions between the large-scale MJO and the MC landmass and its topography. This MC barrier effect is responsible for the dissipation of 40-50% of observed MJO events, though the exact nature of the barrier effect is unclear. Common mechanisms include the physical barrier of the islands of the MC, and the dynamical barrier of strong diurnally driven circulations that exist around those islands. The MC barrier effect is often exaggerated in when it comes to MJO prediction.

In this study, we examine convection-permitting, atmosphere-ocean coupled model simulations of an MJO event to determine how the MJO responds to physical and dynamical changes implemented over the MC region. In addition to the control simulation with real topography, we introduce two idealized simulations – (1) where we flatten the topography of the MC to sea level, but leave the land-sea distribution as is, and (2) where we entirely remove the MC islands and replace them with a 50-m deep ocean. How the MJO responds to the implemented changes can help us determine whether some physical processes that occur over the MC are more detrimental to MJO propagation than others. The differences between the control simulation and the first scenario can tell us about the physical barrier effect of the MC on MJO propagation. The complete removal of land in the second scenario also removes the diurnal changes associated with air-sea boundaries (e.g., land-sea breezes and convergence zones between islands), exploring whether the barrier effect of the MC on the MJO is more dynamically driven.

Results show that flattening the MC terrain only has a small impact on large-scale MJO characteristics. However, as expected, removing the land, and diurnal cycle associated with it, drastically smooths the MJO’s propagation and the produced MJO shows no sign of dissipation over the MC region. We examine the model simulations to gain insight on what physical processes are behind the changes among model simulations and to expose some modeling difficulties that could contribute to numerical models’ exaggerating the effects of the MC barrier effect.

How to cite: Savarin, A. and Chen, S.: Understanding the Barrier Effects of the Maritime Continent on MJO Prediction, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-10984, https://doi.org/10.5194/egusphere-egu22-10984, 2022.

Air-sea coupling is critical in influencing atmospheric temperature and precipitation. The effect of greenhouse gases has influenced atmospheric variability and extreme events. Understanding and quantifying the effect of air-sea feedback on atmospheric variability and extremes remains unknown.

In this work we show results obtained from two numerical experiments. We use the HadSM4 configuration that couples the HadAM4 model at N144 resolution with a Slab Ocean to generate a large ensemble (~1000 members) of realizations of the 2013-14 October-March winter season, forced with a calibrated ocean heat convergence flux.

A twin experiment is performed by forcing HadAM4 with the diagnosed SST and sea ice from the ensemble, yielding a new ensemble with identical realizations of SST and sea ice. The only difference between the two ensembles is the enabling or disabling of the feedback of air-sea heat fluxes on SST.

While the impact of the feedback on the mean climate is relatively small, we show that its influence has important consequences for the variability of many important quantities, including air-sea fluxes and return periods of extreme events.

How to cite: Aengenheyster, M., Sparrow, S., Watson, P., Wallom, D., Zanna, L., and Allen, M.: Impact of sub-seasonal atmosphere-ocean interactions in a large ensemble, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-12033, https://doi.org/10.5194/egusphere-egu22-12033, 2022.

Extreme weather events have devastating impacts on human health, economic activities, ecosystems, and infrastructure. It is therefore crucial to anticipate extremes and their impacts to allow for preparedness and emergency measures. There is indeed potential for probabilistic subseasonal prediction on timescales of several weeks for selected cases of extreme events that are linked to remote drivers and large-scale teleconnections. We here present a range of case studies, including heatwaves, cold spells, and tropical cyclones, where precursors and global linkages may have improved sub-seasonal predictability. These linkages include teleconnections from the tropics as well as the stratosphere, in addition to circumglobal teleconnections. The considered heatwaves exhibit predictability on timescales of 3-4 weeks, while this timescale is 2-3 weeks for cold spells. Precipitation extremes are the least predictable among the considered extremes. Tropical cyclones, on the other hand, can exhibit probabilistic forecast skill on timescales of  up to 3 weeks,  which tends to be favored by remote precursors such as the Madden-Julian Oscillation. These case studies clearly illustrate the potential for event – dependent advance warnings for a wide range of extreme events globally. The subseasonal predictability of extreme events allows for an extension of warning horizons, can provide advance information to impact modelers, and informs communities and stakeholders affected by the impacts of extreme weather events.

How to cite: Domeisen, D. I. V., White, C. J., Afargan-Gerstman, H., Antoine, S., Ardilouze, C., Batté, L., Camargo, S. J., Collins, D., Ferranti, L., Infanti, J. M., Janiga, M. A., Kolstad, E. W., LaJoie, E., Magnusson, L., Strazzo, S., Vitart, F., and Wulff, C. O.: Teleconnection-driven sub-seasonal predictability of extreme events: Relevant case studies, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-4948, https://doi.org/10.5194/egusphere-egu22-4948, 2022.

One of the great concerns of the scientific community in the last decade concerns climate change and its consequences for humanity. The Brazilian hydric planning has faced constant challenges to guarantee supply and energy. In the various regions of the country, serious water shortages have been observed in recent years, due to the scarcity of rainfall, which has become more frequent and intense over the years. Due to its vast territorial extension and topographic complexity, Brazil has different precipitation regimes. Therefore, the study of changes in hydro-meteorological data time series is of extreme importance for the management of water resources. In this context, an analysis of the variability of the historical series of inflowing water flows of the main reservoirs in the country is fundamental for the understanding of the processes involved in drought episodes, in view of the significant impact that these oscillations can produce on Brazilian hydric planning. It is known that the variability of the flow in these reservoirs is closely related to the rainfall regime of each region, which in turn is influenced by climatic variability. Among the climatic variability stands out the Madden-Julian Oscillation (MJO) or 30-60-day Oscillation, which is a mode of intraseasonal climate variability that plays a key role in precipitation over much of South America. Based on this theme, the present work aims to analyze, quantify and predict the influence of MJO on precipitation and consequent hydro-energy variability in Brazil, considering the significant impact that these oscillations can produce on the economy of the country. In this way, we sought to determine what is the contribution of this oscillation to the variability of tributary flows, in the period from 1990 to 2016, data obtained through the National Water Agency (ANA), in order to contribute substantially to the improvement of hydroclimatic forecasts. After the identification of the extemos events of the historical series (positive and negative anomalies of affluent flow) the following analyses were carried out through the method of Wavelets, with the objective of identifying the intensity and the temporal scale of the most expressive phenomena acting in each reservoir of this study. Then a filter was applied on the Wavelets in order to highlight the intraseasonal scale (MJO) and smooth the interannual scale to identify what is the contribution of this phenomenon for each of the regions studied. Finally, in possession of these results, a comparison was made between them and the precipitation simulated by the Brazilian Global Atmospheric Model (BAM) for the same period, with the objective of analyzing the precipitation simulations and their influence on the levels of the country's water reservoirs. In view of this information, the analysis of BAM simulations will be of extreme importance in the comparison between simulations of precipitation and the levels of the reservoirs studied, which can contribute to the forecast and decision-making regarding the management of water resources and thus efficiently improve administration and investments throughout the sector, reducing the degree of vulnerability currently faced by the country.

How to cite: Kunzler, S., Boiaski, N., Ferraz, S., Herdies, D., and Bresciani, C.: Variability of the Inflow at the Intraseasonal Scale and Relationship with Precipitation from the BAM Model, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-8807, https://doi.org/10.5194/egusphere-egu22-8807, 2022.

    This study examines the impacts of the Madden-Julian Oscillation (MJO) on extratropical prediction skill in the Northern Hemisphere during the extended winter, using subseasonal-to-seasonal reforecasts. All four models examined in this study showed a sensitivity of the prediction skill in the North Pacific basin to the initial MJO amplitude at lead weeks 4 and 5. In the ECMWF model, for example, pattern anomaly correlation coefficient (PACC) skill of 300-hPa geopotential height at week 4 becomes higher when the model is initialized with strong MJO than weak MJO. An improved PACC skill with strong MJO is also found in surface air temperature prediction, primarily over the United States. Although not always statistically significant, the similar results are also found in other three reforecasts. The changes in extratropical prediction skill seem to be linked with the MJO prediction skill and the amplitude of the predicted MJO in the first 3 weeks of forecast lead, both of which are higher with strong MJO at initial state than those without active MJO. In addition, the impact of the MJO initial condition on the extratropical prediction is different for each MJO phase. The prediction skill mainly changes in the region where observed teleconnection pattern is consistent across events for each phase, verifying the causality between the MJO and the extratropical prediction skills.

How to cite: Kim, H., Kim, H., and Son, S.-W.: The influence of MJO initial condition on the extratropical prediction skills in subseasonal-to-seasonal prediction model, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-13017, https://doi.org/10.5194/egusphere-egu22-13017, 2022.

The aim of the Subseasonal to Seasonal (S2S) Prediction Project Real Time Pilot (RTP) Initiative is to identify best practices for the development of useful and usable, user-orientated S2S forecasts. Typically, S2S forecasts are only available to researchers with a 3-week lag, but this can represent a barrier to the development of user-orientated applications as it prevents users from being able to understand the utility of this information in real time within their decision-making frameworks. To accomplish the aims of the RTP, the initiative looked to engage with existing user-orientated projects and offer them the opportunity to access real time S2S forecast information to enable better end-to-end development and evaluation of applications. To ensure that sufficient time was available for projects and users to use and become familiar with the real time S2S forecasts, an agreement was reached where S2S forecasts would be available in real time, to a small set of projects for a 2-year period (November 2019 up to November 2021). This has since extended up to November 2022.

To address the aims of the S2S RTP a series of feedback activities have been undertaken with the 16 projects involved. This has included the dissemination and analysis of 2 sets of questionnaires, followed by more detailed semi-structured interviews and subsequent synthesis. All feedback activities were inclusive of researchers and users participating in the initiative. This presentation will describe the different approaches projects have taken in the development of S2S forecast applications, focussing on co-production and user engagement activities across the value chain. Benefits, opportunities and challenges to using co-production methods in the development of user-orientated forecasts are identified through the feedback activities and wider literature. These findings suggest that the application of co-production methods remains novel in the S2S time range, with time and resource availability for stakeholder engagement posing a challenge. However, the feedback indicates that where bi-directional interaction is sustained, positive feedback mechanisms can develop, which build trust, strengthen collaborative working arrangements and enhance forecast product development specific to user requirements.

How to cite: Robbins, J. and Simmonds, R.: Using the Subseasonal-to-Seasonal (S2S) Real Time Pilot (RTP) Initiative to understand the challenges and opportunities of co-production in S2S forecast application development, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-8008, https://doi.org/10.5194/egusphere-egu22-8008, 2022.

The PISSARO project focuses on atmospheric and oceanic forecasting at the subseasonal scale for applications over the South West Indian Ocean basin (SWIO). It is a collaborative academic research project, developed and conducted in partnership with stakeholders from Reunion and Seychelles and a panel of scientific experts in subseasonal forecasting. The aim of this project is to evaluate, improve and valorize subseasonal forecasting data. For this purpose, we mainly use the data archived into the the S2S (Subseasonal-to-Seasonal prediction project) data base in order to 1) evaluate the quality of subseasonal forecasts for tropical cyclones and weather patterns, and 2) develop forecast products suitable for potential users. This project focuses on the SWIO, which has been little studied by the S2S community until now. The different territories of the SWIO are subject to extreme events and a significant cyclonic activity. It is important to take into account the specificities of this region in order to improve their warning systems.

The ambition to deploy early warning tools cannot be achieved without discussions between potential users and S2S experts. The users specify the characteristics of the products to be developed so that they offer an asset for decision-making, and the experts assess the feasibility of these products. In the presentation, we will first discuss the importance of collaboration between the users and the experts within the project using two concrete actions: the animation of a monthly experimental forecasting briefing with operational forecasters and the participation in conferences in the humanitarian field. Then, we will present the subseasonal forecasting products which are under development for the anticipation of cyclonic risk at monthly scale in the SWIO basin. To address the urgency of the need of the disaster risk reduction, we first made a basic adaptation of already existing tropical cyclone occurrence probability and rainfall forecasting products into products interpretable by non-meteorological users.

We consider that a crucial information from the S2S data base to provide to users is the level of uncertainty. However, estimating the quality of S2S forecasts is not straighforward. It is actually difficult to match a forecasted cyclone to an actual observed cyclone, let alone detect a false alarm. To this end, we are working on the classification of S2S tropical cyclone trajectories with clustering methods and we will show the first results. We aim to exploit the ensemble character of the subseasonal forecast for the development of future S2S-derived forecast products that would provide probabilities of scenarios of potential trajectories (based on these clusters).

How to cite: Vérèmes, H., Malardel, S., Bonnardot, F., Labbé, L., Langlade, S., Peyrillé, P., Charpigny, S., Lefort, T., and Mékiès, D.: The PISSARO project: subseasonal impact-based forecasts of the cyclonic activity in the South West Indian Ocean basin, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-41, https://doi.org/10.5194/egusphere-egu22-41, 2022.

In the agricultural sector there is a high interest for forecasts that predict relevant agroclimatic indicators related to heat accumulation and frost characteristics. The forecasts can simplify agricultural decisions related to planting and harvest timing. Motivated by this, we propose a probabilistic forecasting framework for predicting the end of the freeze-free season, or the time to a mean daily near-surface air temperature below 0 ^°C (here referred to as hard freeze). The forecasts are constructed based on a multi-model seasonal temperature forecast ensemble provided by the Copernicus Climate Data Store. The raw temperature forecast is statistically post-processed through a mean and variance correction. The resulting ensemble is next used as input to a survival analysis model. Survival analysis is a broad statistical field that is commonly used in the field of biostatistics, but rarely used in meteorology.

The forecasting framework is evaluated by predicting the time to hard freeze from October 1 for 1993-2020 for a region in Fennoscandia that covers Norway and parts of Sweden, Finland and Russia. We find that the proposed forecast outperforms a climatology forecast from an observation-based data product at locations where the average predicted time to hard freeze is less than 40 days after the initialization date.

Our work also forms an entry point showing how survival models can be used in general to construct seasonal forecasts for other meteorological events, e.g. the onset of the rainy season or the time to the next drought.

How to cite: Roksvåg, T., Lenkoski, A., Sheuerer, M., Heinrich-Mertsching, C., and L. Thorarinsdottir, T.: Probabilistic prediction of the time to hard freeze using seasonal weather forecasts and survival time methods, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-7005, https://doi.org/10.5194/egusphere-egu22-7005, 2022.

Reliable sub-seasonal forecasts for precipitation and temperature are crucial to many sectors including agriculture, public health and renewable energy production. Since the forecast skill of numerical weather forecasts for lead times beyond two weeks is limited, the World Meteorological Organization launched a Challenge to improve Sub-seasonal to Seasonal Predictions using Artificial Intelligence, which was held from June to October 2021. Within the framework of this challenge, we have developed a hybrid forecasting model based on a convolutional neural network (CNN) that combines post-processing ideas with meteorological process understanding to improve sub-seasonal forecasts from the European Centre for Medium-Range Weather Forecasts (ECMWF).

Here, we present a refined version of our model that predicts tercile probabilities for biweekly averaged temperature and accumulated precipitation for weeks 3 – 4 and 5 – 6. Our model is trained on limited-area patches that are sampled from global predictor fields. It uses anomalies of large-scale predictors and features derived from the target variable forecasts as inputs. Spatial probabilistic forecasts are obtained by estimating coefficient values for local, spatially smooth basis functions as outputs of the CNN. Our CNN model provides calibrated and skillful probabilistic predictions, and clearly improves over climatology and the respective ECMWF baseline forecast in terms of the ranked probability score for weeks 3 – 4 and 5 - 6.

How to cite: Horat, N. and Lerch, S.: Convolutional neural networks for skillful global probabilistic predictions of temperature and precipitation on sub-seasonal time-scales, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-920, https://doi.org/10.5194/egusphere-egu22-920, 2022.

Severe weather associated with organized convective systems is becoming more intense globally and is also observed in the Arabian Peninsula (AP). The extreme rainfall-associated flooding in low soil infiltration region like the AP often lead to significant social and economic losses within a very short period. Improving forecast capability at sub-seasonal to seasonal (S2S) timescale can potentially assist disaster risk mitigation, and water resource management.

A series of S2S regional climate model reforecasts were completed using the Weather Research and Forecasting Model (WRF) at convective-permitting resolution (4 km) for the AP. We dynamically downscale 20 years of winter season from the European Centre of Medium-range Weather Forecasts (ECMWF) S2S reforecast product. WRF simulations were initialized weekly with 1-month simulation duration between November and April.  A total of 191,400 hindcast days have been generated to evaluate the predictability of winter rainfall associated with convective activities.

Methods designed to evaluate the S2S forecast skills considers the probability of detection of precipitation at neighboring grids, determining the rate of forecast agreements between ensemble members, and running evaluation of the probability of forecast from 1-week to 4-week lead time. We evaluated all rain gauge measurement and gridded precipitation datasets available for the study period and determined the following datasets as our ground-base reference: satellite based Global Precipitation Mission (GPM) and 4-km reanalysis data produced by the King Abdullah University of Science and Technology (KAUST-RA). The WRF S2S downscaled reforecasts significantly improved from the driving ECMWF reforecast climatology, as evaluated against  the GPM and the KAUST-RA dataset. Our WRF results also produced reasonable winter precipitation climatology over the AP as compared to the satellite observations and high-resolution reanalysis products, at 1-week, 2-week, 3-week and 4-week forecast lead times.

How to cite: Chang, H.-I., Castro, C. L., Luong, T. M., Risanto, C. B., and Hoteit, I.: Analyses of convective event climatology in the Arabian Peninsula and forecast opportunity at S2S time scale , EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-8965, https://doi.org/10.5194/egusphere-egu22-8965, 2022.

The need to build reliable weather forecasting systems for subseasonal to seasonal (S2S) timescales has never been greater as the world continues to experience increased numbers of extreme weather events. This study addresses the skill gap between numerical weather prediction (NWP) and seasonal forecasting by proposing a daily probabilistic forecast model that predicts 2-meter temperature and total precipitation on a global scale. It combines multimodal data (e.g., physics-based ensembles, climate modes, recent climatology) into feature vectors given as inputs to three ML models: Extreme Gradient Boosting, U-Net, and Natural Gradient Boosting. We use Bayesian hyperparameter search and leave-one-year-out RPSS cross-validated scores to accelerate learning and ensure generalizability. Our method consistently outperforms both ECMWF 46-day forecasts and climatology. We find that augmenting physics-based issued forecasts with other sources of predictability greatly improves the performance of the underlying dynamical models. We hope that by improving the physics-based probabilistic forecasts, we will unlock skill in predicting climate extremes-oriented indices. Subsequent representation learning models may be trained to efficiently navigate the ensembles' uncertainty space and estimate the likelihood of extreme events.

How to cite: Zaytar, M. A., Zadrozny, B., Watson, C., Salles Civitarese, D., Eben Vos, E., Michael Mathonsi, T., and Lukhetho Mashinini, T.: ML-based Probabilistic Prediction of 2m Temperature and Total Precipitation, EGU General Assembly 2022, Vienna, Austria, 23–27 May 2022, EGU22-11063, https://doi.org/10.5194/egusphere-egu22-11063, 2022.