Annual Seminar 2019
ECMWF
#AS2019
Subseasonal and seasonal forecasting: recent progress and future prospects
More than two decades ago, seasonal forecasting started as a pilot project at ECMWF. Now it is a corner-stone of climate services. Originally a spin-off from seasonal forecasting, the sub-seasonal range covered the then so-called predictability desert; now it is a key building block of the seamless forecasting strategy, with clear prospects for useful skill gains at week 3 and 4.
Although the predictable processes and predictability drivers differ between the subseasonal and seasonal scales, prediction at these time ranges is a combination of initial and boundary problems. Sub-seasonal and seasonal predictions act as a bridge between weather and climate.
This seminar reviewed recent advances in our understanding of the predictability at these time scales. It presented current forecasting capabilities, and summarised recent but consolidated findings from numerical experimentation and exploitation of public data bases. It also provided a perspective of incipient developments related with data exploration, forecast products and predictability drivers, which will shape the future of seamless forecasting systems.
-
-
12:30
→
13:15
Registration and coffee
- 13:15 → 13:30
-
13:30
→
16:50
Session 1: Basis for predictability at the extended and seasonal rangeConvener: Tim Stockdale (ECMWF)
- 13:30
-
13:45
A personal perspective on predictability on sub-seasonal to seasonal time-scales 1h
The talk will discuss the development of ideas and capabilities over the past 50 years, and the phenomena and dynamics that give the potential for prediction.
The notion of chaos by Lorenz was originally conceived in the context of the long-range forecasting project in which he was employed at MIT. He was essentially showing that the project that employed him was hopeless! However, over the years, researchers using observational data and running state-of-the-art atmospheric models have been able to find hints that some skill may be possible on some occasions.
For me, the potential for predictability beyond the synoptic time-scale is based on the phenomena that occur on these time-scales and the dynamics and physics that underlies them. The talk will expand on this perspective, giving some examples.Speaker: Brian Hoskins (University of Reading & Imperial College London) -
14:45
Taming the butterfly effect to reach subseasonal and seasonal predictability 50m
Operational forecasts valid beyond the two weeks period that was thought to be the limit of predictability in the 1960s/1970s, have been available since the 1990s. At ECMWF, for example, operational, ensemble-based seasonal forecasts have been issued since 1997, and a dedicated monthly ensemble have been in operation since 2004.
Progress in the past two decades in predictability gains have been stunning. For the seasonal forecast range, probabilistic forecasts of sea-surface-temperature anomalies linked to El-Nino have been improving by almost 1 month every decade. In the subseasonal forecast range, probabilistic forecasts for large-scale phenomena such as the Madden-Julian Oscillation have been improving by about one week every decade.
These exceptional skill extensions (exceptional compared to the 1 day/decade gains in predictability detected in single, deterministic forecasts) have been achieved by working on many key aspects of ensemble system design. Gains have been obtained by contrasting error propagation from the smaller to the larger scales, with the propagation of predictable signals from the larger to the smaller scales. Improvements in the model design and in the simulation of atmospheric and land-surface processes, the introduction of coupling to the ocean and the sea-ice, advances in the definition of the initial conditions, the adoption of more reliable ensemble systems, have all contributed to the extension of the forecast skill horizon.
In this lecture, I will present an overview of the key aspects that have led to the operational production of skilful subseasonal and seasonal forecasts, and will make some considerations on the future prospects for subseasonal and seasonal forecasts.Speaker: Roberto Buizza (Scuola Superiore Sant'Anna) -
15:35
Coffee break 25m Weather Room, Lobby
Weather Room, Lobby
-
16:00
Predictability associated with teleconnections from tropical phenomena 50m
Teleconnections associated with ENSO variability have long been recognised as the major source of seasonal predictability for many regions in both the tropics and extratropics. However, in recent years the importance of teleconnections originated outside the "core" ENSO region have been increasingly recognised. On the seasonal scale, the ENSO teleconnections appear to be modulated by anomalies in the tropical Indian and Atlantic oceans; on the subseasonal scale, teleconnections from the Madden-Julian Oscillation have been shown to affect the occurrence of flow regimes in the North Atlantic.
The talk will review evidence on the impact of these teleconnections, with particular attention to anomalies originated from the Indian Ocean. It will be shown that Indian Ocean - North Atlantic teleconnections have common features when studied at sub-seasonal, seasonal and decadal time scale; a possible dynamical mechanism will be discussed, and results on the ability of European state-of-the-art coupled models to simulate these teleconnections will be presented.
Speaker: Franco Molteni (ECMWF)
-
16:50
→
17:40
Session 2: Physical processes, modelling and initialization requirementsConvener: Gianpaolo Balsamo (ECMWF)
-
16:50
The complexity of ENSO and its impacts: lessons learnt from initialized predictions 50m
Over the past decade, substantial efforts have been devoted to elucidate how a changing climate will impact ENSO. At the same time, there have been continuous advances in ENSO monitoring and predictive capabilities. In this talk, we bring together both lines of work by discussing the performance of initialized ENSO predictions. Improved reanalyses data sets, real-time seasonal forecasts, and the hand of nature providing additional case studies, have contributed to the realization of ENSO as a varied and complex phenomenon. The variety of ENSO flavours results from different balances among a handful of thermodynamical processes operating at a multiplicity of time scales. ENSO is a clear example of spatial and temporal scale interaction, being influenced by tropical weather as well as by slow decadal variability modes and trends. This scale interaction poses considerable challenges for ENSO forecasting systems, as it will be illustrated during this lecture.
Speaker: Magdalena Alonso Balmaseda (ECMWF)
-
16:50
-
17:40
→
18:00
Express poster presentationsConvener: Anca Brookshaw (ECMWF)
-
18:00
→
19:00
Poster session 1 and drinks reception
-
12:30
→
13:15
-
-
09:00
→
16:50
Session 2: Physical processes, modelling and initialization requirementsConvener: Gianpaolo Balsamo (ECMWF)
-
09:10
The Madden-Julian Oscillation 50m
The Madden-Julian Oscillation is the dominant mode of sub-seasonal variability in the tropical climate system and a major potential source of predictability on sub-seasonal timescales. This presentation will review our current understanding of the MJO; the impacts of the MJO in the tropics and extra-tropics; the skill of weather and climate models in simulating the MJO; and conclude with some of the challenges in realizing the potential predictability associated with the MJO.
Speaker: Steven Woolnough (National Centre for Atmospheric Science, University of Reading) -
10:00
Multi-Scale Impacts of Extratropical Ocean on the Atmosphere 50m
Unlike in vast areas of midlatitude ocean basins, the warm midlatitude/subtropical western boundary currents and associated oceanic frontal zones can potentially impact the overlying atmosphere. Multi-scale aspects of the impacts are overviewed in this presentation. Effective moisture supply from a warm current to individual cyclones and efficient restoration of near-surface atmospheric baroclinicity across frontal sea-surface temperature (SST) gradient allow recurrent development of cyclones, contributing to the formation of a stormtrack and associated eddy-driven polar-front jet (PFJ) along the frontal zone. The presence of midlatitude oceanic fronts is thus essential for the annular mode, as a manifestation of wobble of a PFJ, especially over the Southern Ocean, where oceanic fronts are more or less circumpolar. In the North Pacific, variability of oceanic fronts accompanies persistent SST anomalies, manifested as centers of action of decadal SST variability. Caused by ocean dynamics, decadal displacement of the subarctic frontal zone can force a basin-scale atmospheric anomaly into midwinter through modulating stormtrack activity.
Furthermore, intense heat and moisture release from the warm western boundary currents leaves distinct meso-scale imprints in the atmospheric boundary layer. The most notable imprint is locally enhanced convective precipitation, as observed along the Gulf Stream and Kuroshio, in association with enhanced surface wind convergence. Frequent cold-air outbreak associated with traveling weather systems along the nearby stormtrack enhances sensible heat flux from the warm current and thereby favors formation of shallow convective clouds. In summer, high SST along the Kuroshio and East China Sea helps organize deep convective precipitation under the moist monsoonal southwesterlies toward the Baiu/Meiyu rain front. Under the monsoonal northerlies in winter, the warm Kuroshio also organizes shallow convective stratocumulus within well-developed unstable mixed layer, where strong ascent acts to augment the super-saturation level, thus leading to a substantial increase in cloud droplet density.Speaker: Hisashi Nakamura (RCAST, University of Tokyo) -
10:50
Coffee break 30m Weather Room, Lobby
Weather Room, Lobby
-
11:20
Land surface as a predictability driver in Subseasonal and seasonal Forecasts 50m
Processes occurring at the land surface impact weather and climate variability in a wide range of timescales from days to millennia, making land surface models a required component of both weather and climate prediction systems. Of special relevance is the role of land surface processes involving snow, soil water and vegetation in the amplification of extreme weather and climate anomalies, such as the extreme hot summers in Europe in 2003 and 2010. Soil moisture has a long residence time, about 60 days, when compared with 10 days in the atmosphere. This leads to a considerable potential predictability induced by land surface conditions. However, this memory effect is hampered by (i) difficulties to accurately monitor surface conditions (e.g soil moisture); (ii) forecast errors (e.g. bias in precipitation) and (iii) representation of surface turbulent fluxes and soil-plant water transport. This talk will provide an overview of the current understanding of the land surface role in sub-seasonal to seasonal predictability. Particular focus will be given to observational proxies, land data assimilation and land-surface modelling. Finally, current and future challenges in the initialization and representation of surface process relevant for sub-seasoal to seasonal prediction will be presented.
Speaker: Emanuel Dutra (Instituto Dom Luiz, Faculdade de Ciências, Universidade de Lisboa) -
12:10
High-latitude processes in sub-seasonal to seasonal predictions 50m
At high latitudes, the dominant physical processes are radically different from the mid-latitudes and the tropics. At the surface, the large-scale presence of snow and sea ice provides a potential source of predictability on sub-seasonal to seasonal time scales, but poses a difficult challenge for reliable observation, model representation and initialization. Likewise, the atmosphere over snow and sea ice often exhibits unique boundary layer and cloud regimes that are not well represented in global models. Combined with an extreme sparsity of conventional observations, this raises the question whether sub-seasonal to seasonal predictions could benefit from increased investments in observation networks and modelling capacities, especially in the Northern Hemisphere, where teleconnections between Arctic and mid-latitude phenomena are actively being discussed in the research community. Furthermore, the Arctic environment has changed dramatically in the last decades, with important socio-economic, environmental and security implications. To reflect on these issues, this talk is formed of two parts. The first part gives an overview of current state of the art and prospects in observing, modelling and initializing sea ice and other high-latitude processes to improve subseasonal to seasonal predictions. In the second part, recent progress in understanding Arctic-midlatitude linkages will be reviewed and implications for subseasonal to seasonal predictions will be discussed.
Speaker: Steffen Tietsche (ECMWF) -
13:00
Lunch break 1h ECMWF
ECMWF
Reading -
14:00
The role of atmospheric composition in the predictability at the S2S scale 50m
In recent years, user demand for forecasts that fill the gap between medium-range weather (up to 15 days) and long-range or seasonal
(3–6 months) forecasts has increased. Skillful subseasonal to seasonal prediction can support decision-making and help
optimizing resource management decisions.
Prediction at the S2S scale, however, is particularly challenging because it is both an initial value problems, much like
the standard Numerical Weather Prediction (NWP) medium-range, and a boundary value problem like seasonal prediction. Over the
years, researchers have been trying to find source of predictability at this scale, looking at natural-occurring patterns or
processes both in the troposphere and the stratosphere that have a periodicity of weeks up to a few months.
The Madden-Julian oscillation has been identified as one of the most important of
such patterns. However, much debate is still ongoing as far as what the triggers of the MJO and what the important
feedbacks connected to it are.Recent work has shown that the atmospheric constituents such as aerosols, ozone and other trace gases can
be important modulators of the radiative processes at the S2S scale.
For example, the direct effect of aerosols may influence
predictability via the MJO modulation of the
aerosol fields. In clear-sky, the cumulative aerosol forcing can modify the radiative balance of the atmospheric column
and introduce temperature perturbations which depend on the dominant aerosol types and
their optical properties. Wind-emitted aerosols such as dust appear to be the main
contributors. However, sensitivity studies performed with the
ECMWF’s coupled Ensemble Prediction System have shown that biomass
burning aerosols may also play an important part, in particular for areas
where extensive seasonal biomass burning takes place such as central Africa
and Indonesia.
Aeorols of volcanic origin have also been shown to affect stratospheric processes and to have also a large impact on the S2S prediction.
Interactions between stratosphere and troposhere are also considered to play
an important role. A correct definition of the stratosphere in models is connected to a correct representation
of ozone. While this has been explored at the seasonal scale, investigation in the role of interactive ozone at the S2S scales is ongoing.In this talk a review of current efforts to understand the impact of atmospheric constituents on the S2S predcition will be presented and
discussed.Speaker: Angela Benedetti (ECMWF) -
14:50
The role of the stratosphere for sub-seasonal to seasonal forecasting 50m
Sub-seasonal to seasonal (S2S) predictions of surface climate are crucial for a wide range of sectors. One of the promising areas that adds predictability on these timescales is the stratosphere, which has been found to play an important role in the predictability of surface weather on S2S timescales. The downward influence of the stratosphere onto the surface can lead to increased persistence and predictability of surface weather. For example, the SSW event in February 2018 led to persistent cold weather over large parts of Europe in late February and early March after an otherwise mild winter. Like the 2018 event, up to two thirds of SSW events are followed by anomalous tropospheric weather patterns that can remain persistent for several weeks. Other types of polar stratospheric extreme events such as strong vortex events or wave reflection events can also have impacts on surface weather. However, skill for extratropical stratospheric events generally only exists at the deterministic level for one- to two- week lead times. Probabilistic skill exists for stratospheric events when including precursor events and teleconnections, though these are often poorly captured by models. These remote effects include El Nino Southern Oscillation and the Quasi-Biennial Oscillation, which themselves hold predictability of several months and can therefore induce persistent forcings of the extratropical stratosphere. This lecture will cover the current ability of models to predict stratospheric events, their precursors, and their surface impacts, and provide an outlook on future efforts to increase the skill arising from stratospheric events.
Speaker: Daniela Domeisen (ETH Zurich) -
15:40
Coffee break 20m Weather Room, Lobby
Weather Room, Lobby
-
16:00
Statistical methods for verification of probabilistic forecasts at the extended and seasonal range 50m
This seminar will review statistical methods for verification of probabilistic forecast at the extended and seasonal range. Transparent statistical modelling frameworks will be presented for modelling the distributional properties of forecasts and observations. It will be shown how such frameworks can provide complete verification summaries and be used to diagnose predictability issues such as over-dispersion in the forecasts (the so-called signal-to-noise paradox). The use of the statistical frameworks for quantifying uncertainty in skill and recalibrating forecasts will also be discussed. The use of Bayesian estimation will be presented as a means for addressing the issue of small sample sizes.
References
Siegert S, Stephenson D. (2018) Forecast recalibration and multi-model combination, Sub-seasonal to Seasonal Prediction: The Gap Between Weather and Climate Forecasting, Elsevier. Editors: Andrew Robertson & Frederic Vitart, 585 pages.
Siegert S, Stephenson DB, Sansom PG, Scaife AA, Eade R, Arribas A. (2016) A Bayesian framework for verification and recalibration of ensemble forecasts: How uncertain is NAO predictability?, Journal of Climate, volume 29, no. 3, pages 995-1012.
Speaker: David Stephenson (University of Exeter)
-
09:10
-
16:50
→
17:10
Express poster presentationsConvener: Anca Brookshaw (ECMWF)
-
17:10
→
18:00
Poster session 2 and drinks Weather Room
Weather Room
-
09:00
→
16:50
-
-
09:10
→
14:00
Session 3: System designConvener: Frederic Vitart (ECMWF)
-
09:10
Atmospheric teleconnections, the North Atlantic Oscillation and long range forecasts of European winters 50m
We present an update on long range predictability of the winter NAO and hence European winters. Recent years continue to support hindcast results that the winter NAO is predictable at seasonal lead times and here we investigate mechanisms for this predictability from the tropics and the stratosphere.
High predictability of tropical rainfall is first demonstrated for current prediction systems and this is shown to lead to predictable changes in vorticity sources. These are associated with clear stationary Rossby waves that propagate into the extratropics and the Atlantic sector. We estimate that this mechanism can explain around half of the forecast variance in the NAO.
Secondly, we show that initial atmospheric conditions are also important for seasonal prediction of the NAO. Initial anomalies in stratospheric winds at the start of winter propagate downwards into the troposphere on subseasonal timescales where they lead to anomalies in the winter mean surface conditions. Together, these mechanisms appear to explain the majority of forecast variance in the winter NAO. Finally, we discuss some of the remaining errors in forecasts of tropical rainfall and the unresolved signal-to-noise paradox in ensemble forecasts.Speaker: Adam Scaife (Met Office Hadley Centre) -
10:00
Seasonal forecasting systems: present and future 50m
Seasonal forecasting systems: present and future
Seasonal forecasting systems based on general circulation models have been developed at various operational forecast centres over the last 25 years, and ECMWF is one of the places which has much experience with this type of forecasting system. Some aspects of such forecasting systems are mature, and give moderately reliable information about the future. Other aspects are still considered problematic, with models struggling to capture apparent features of reality, and significant gaps in our knowledge of the processes determining predicatability at seasonal timescales.
The status and development of EMCWF's current seasonal forecast system, SEAS5, will be discussed in this context, together with some of the outstanding challenges and plans for the future. The critical role of model errors and how they can be dealt with will be reviewed, including the important role of multi-model forecasting systems.
Speaker: Tim Stockdale (ECMWF) -
10:50
Coffee break 30m Weather Room, Lobby
Weather Room, Lobby
-
11:20
Multi-week/seasonal prediction for agricultural applications in Australia 50m
Multi-week/seasonal prediction for agricultural applications in Australia
Oscar Alves
Bureau of Meteorology, 700 Collins St, Melbourne, VIC 3008The main driver for the development of multi-week to seasonal prediction systems in Australia is to support decision making in the agricultural sector. The development of a new dynamical seasonal prediction model, ACCESS-S (Australian Community Climate and Earth-System Simulator – Seasonal), is a major step forward in the ability to predict climate variability on multi-week to seasonal time scales in Australia. The ACCESS-S modelling framework provides improved seasonal forecast skill and enables a range of products to be developed for various agricultural sectors.
A range of products have been developed and exposed to agricultural scientist and end-users. These products include: general climate drivers such as ENSO, the Indian Ocean dipole, the Madden-Julian Oscillation and the southern annular mode; forecast of extreme events, including heat waves, and extreme cold and wet periods; forecasts for the northern Australia wet season onset; and a range of tailored products, for example, forecasts of temperature humidity heat index.
To facilitate the development of products a quantile-quantile mapping calibration scheme was used to convert raw ACCESS-S model output at 60km to calibrated daily data at 5km. This calibration technique was applied to a range of variables used in driving agricultural models such as rainfall, temperature, radiation, wind speed and evaporation. This data has in turn been used to drive a range of agricultural models. For example, an agricultural package called Ask Bill has been developed to predict a range of sheep livestock quantities, including: available pasture, risk of fly strike, risk of heat stress and risk of cold sheep.Speaker: Oscar Alves (Bureau of Meteorology) -
12:10
The US Navy’s Extended-range Prediction System with High-resolution Ocean and Ice models 50m
The National Earth System Prediction Capability (National ESPC) is a U.S. multi-agency collaborative effort to leverage resources to develop the next generation environmental forecasting system. As part of this effort, the U. S. Navy is developing a fully coupled global system including the Navy Global Environmental Model (NAVGEM), the HYbrid Coordinate Ocean Model (HYCOM), and the Los Alamos Community Ice CodE (CICE). This system is being developed to meet Navy needs for high-resolution global environmental forecasts on timescales from days to months. The design and implementation of the coupled architecture uses the Earth System Modeling Framework (ESMF) with the National Unified Operational Prediction Capability (NUOPC) standard in order to maximize flexibility in adopting future models. Initial operational capability is planned for 2019-2020 and will include daily high-resolution deterministic forecasts (with 19 km atmospheric resolution, 1/25o ocean and sea ice resolution, and 1/8o wave model resolution) and weekly extended-range ensemble forecasts (with 37 km atmospheric resolution, 1/12o ocean and sea ice resolution, and 1/4o wave model resolution). A 20-year archive of 45-day deterministic forecasts four times per week at the ensemble resolution has been produced as part of the Navy’s participation in the NOAA Subseasonal eXperiment (SubX) project. An aspect that makes the system unusual and will be highlighted is the relatively high resolution of the ocean and ice models, reflecting the Navy’s strategic and tactical interests in these realms. The performance of ocean, atmosphere, and ice components of both the deterministic (high-resolution) and ensemble (lower-resolution) components of the system will be summarized, including comparisons to persistence and climatology. Identification of current short-comings and planned future upgrades will also be summarized.
Speaker: Carolyn Reynolds (US Naval Research Laboratory) -
13:00
Lunch break 1h ECMWF
ECMWF
Reading
-
09:10
-
14:00
→
17:40
Session 4: Detecting and exploiting skillConvener: Antje Weisheimer
-
14:00
How Using NASA’s Observations Affects the Balance Among Spatial Resolution, Ensemble Size, and Physical Complexity in the GEOS-S2S System 50m
NASA’s Global Earth Observing System (GEOS) model and data assimilation system can be configured for several applications in NASA’s Global Modeling and Assimilation Office (GMAO). The GMAO performs model-based studies to enhance NASA’s Earth observations, including (among others) the provision of: auxiliary information to space observations; forecasting support for aircraft field missions; quantitative information to the design of new space missions; demonstrations of the value of NASA’s observations in Earth System analysis and prediction.
This presentation addresses the GEOS S2S (subseasonal to seasonal) prediction system. It focuses first on the decisions made in refining the system from Version 2 to Version 3, which will go into production by early 2020, emphasizing how data used impact the model configuration. Looking forwards, a pathway towards a coupled analysis capability is outlined, that emphasizes NASA’s unique observations. GMAO intends to include NASA’s remote sensing information from altimetry, sea-surface temperature and salinity, land- and sea-ice distributions alongside in-situ data and other satellite observations.
One aspect that impacts the use of research observations, that typically exist for short periods, is to use anomalies computed from multiple years of hindcasts: this approach can be detrimental to assessing the impacts of new data types that may induce physical perturbations in such a system. A second aspect of adding some new data types is that their meaningful implementation necessitates extra computation cost: for instance, atmospheric aerosols and ozone require complex and costly enhancements to the model. This in turn raises new issues about balancing compute resources: choices must be made about the balance between the physical complexity of the model and the “traditional” factors of spatial resolution and ensemble size. These aspects will be discussed in this presentation.
Speaker: Steven Pawson (NASA GSFC) -
14:50
Comparing the Predictability and Skill of Subseasonal Forecasts 50m
In this talk, I discuss three techniques for comparing forecasts. The first technique is to identify the predictable space at a given time scale using predictable component analysis. Predictable component analysis finds the linear combination of variables that maximizes a measure of predictability. On subseasonal time scales, this space appears to be very low dimensional. I also will address whether there exist forms of subseasonal predictability that cannot be captured by predictable component analysis. The second technique is pair-wise comparison of forecasts, which allows differences in skill to be detected that otherwise could not be detected with conventional methods, such as a difference-in-correlation test or a difference-in-MSE test. Third, I will explain a permutation method that allows a rigorous assessment of statistical significance even when serial correlation exists between forecasts, which is the rule rather than the exception in subseasonal forecasting.
Speaker: Timothy DelSole (George Mason University) -
15:40
Coffee break 20m Weather Room, Lobby
Weather Room, Lobby
-
16:00
Predicting high impact weather events beyond the medium range 50m
In line with its long-term strategy, ECMWF has recently developed new diagnostics designed to support the prediction of severe weather in Europe. Predicting severe events presents several challenges. It is difficult to gather a large enough sample to get robust statistics. Each event has its own unique nature. A specific physical process relevant for the development of one cold spell, as for example the stratospheric sudden warming, might not play a relevant role during another cold event. As a consequence, it is difficult to identify weaknesses in the forecasting system that have a significant impact on the predictive skill of all events. We will discuss the key processes for predicting the severe temperatures events at different time scales, with emphasis to those factors relevant at sub-seasonal time scale. We will examine the forecast performance during the most recent severe cold events and heat waves, highlighting the strengths and weaknesses of the current forecasting systems. While at medium range, predictions for severe temperature conditions can be directly based on temperature forecast probabilities, at the extended range, the predictable signal for severe and persistent cold/warm spells is better exploited using large-scale circulation patterns. For example, we will show that reliable extended-range forecasts of flow patterns such as the NAO and blocking are instrumental for early warnings of severe cold events over Europe.
Speaker: Laura Ferranti (ECMWF) -
16:50
Towards process-based narratives for seasonal climate predictions 50m
Skillful seasonal climate predictions for European climate remain a formidable challenge. I will present recent progress in predicting European climate variability using prediction systems based on the Max-Planck-Institute Earth System Model (MPI-ESM). In the presentation, I will challenge the current practice in seasonal climate predictions to focus on the analysis of the ensemble-mean forecast. In addition, I will suggest process-based narratives, which I will illustrate by analyzing seasonal re-forecasts for European summer and winter climate.
Speaker: Johanna Baehr (Institute of Oceanography, CEN, Uni Hamburg)
-
14:00
-
17:40
→
18:00
Express poster presentationsConvener: Anca Brookshaw (ECMWF)
-
18:00
→
19:00
Poster session 3 and pre-dinner drinks Weather Room
Weather Room
-
19:00
→
21:00
Seminar dinner
-
09:10
→
14:00
-
-
09:10
→
15:55
Session 5: Multi-model databases for advancing science and applicationsConvener: Laura Ferranti (ECMWF)
-
09:10
What did we learn from the S2S database? 50m
The S2S database, a key component of the World Weather Research Programme (WWRP)/World Climate Research Programme (WCRP) Sub-seasonal to Seasonal Prediction Project science plan, opened to the public in 2015. It contains sub-seasonal (up to 60 days) forecasts, 3 weeks behind real time, and reforecasts from 11 operational centres, modelled in part on The Observing System Research and Predictability Experiment (THORPEX) Interactive Grand Global Ensemble (TIGGE) database for medium-range forecasts (up to 15 days). It is hosted at ECMWF, CMA and IRI. In this talk, the main lessons learned during the development of the database and future developments will be discussed.
The S2S database represents an important tool to advance our understanding of the sub-seasonal to seasonal time range that has been considered for a long time as a “desert of predictability”. More than 70 articles based on the database have been published so far in the peer-reviewed literature. They provide an extensive assessment and model inter-comparison of the skill of state-of-the-art S2S models to predict a wide range of parameters, including sea-ice cover, the Madden-Julian Oscillation and Sudden Stratospheric Warmings. Other scientific questions which have been investigated include:
• What is the importance of multi-model forecast for sub-seasonal to seasonal prediction?
• What is the predictability of extreme events and how can we identify windows of opportunity for sub-seasonal to seasonal prediction?
• What is the impact of atmospheric horizontal and vertical resolution on sub-seasonal to seasonal forecasts?
• How are state-of-the-art models representing tropical-extra-tropical and stratosphere-troposphere teleconnections?
• What are current S2S forecasting capabilities for daily weather characteristics relevant to agriculture, water resource management and public health, such as heavy rainfall events, dry spells and monsoon onset/cessation dates?This talk will review some of the key results obtained from the S2S database.
Speaker: Frederic Vitart (ECMWF) -
10:00
Bridging the Gap between Weather and Climate Prediction using Multi-model Ensembles 50m
Skillful, useful predictions of the climate system beyond the deterministic limits of weather must be built on a basic scientific understanding of predictability and the physical processes giving rise to predictability. Rigorous evaluation of prediction systems is also necessary to understand our current prediction capabilities and identify avenues for improvement. Finally, prediction capabilities must align with what users need to make decisions regarding risk reduction and disaster preparedness, public health, energy, water management, agriculture, marine fisheries, and many other applications.
In this presentation, I discuss the use of two multi-model ensemble prediction projects to investigate questions about predictability and prediction on subseasonal (2-4 weeks) to seasonal (1-month to a year) timescales. The North American Multi-model Ensemble (NMME) and the Subseasonal Experiment (SubX) provide retrospective and real-time forecasts for monthly-to-seasonal (NMME) and subseasonal (SubX) timescales. For seasonal predictions, the NMME provides insight into how well we understand the upper limit of predictability for the El Niño Southern Oscillation, temperature, and precipitation. Results demonstrate that there are significant challenges to estimating predictability, and that a multi-model approach is needed to get a full picture of the limit of predictability. Our understanding of subseasonal prediction capabilities is less mature than for the seasonal timescale. Despite gains from using the full SubX multi-model ensemble, skill for temperature and precipitation is generally low, emphasizing the need to identify forecasts of opportunity and better understand the phenomena and processes that give rise to predictability at these timescales. The Madden-Julian Oscillation and the North Atlantic Oscillation, two sources of subseasonal predictability, are also evaluated.
Speaker: Kathleen Pegion (George Mason University/COLA) -
10:50
Coffee break 30m Weather Room, Lobby
Weather Room, Lobby
-
11:20
Seasonal forecast data and information: the providers' perspective from the Copernicus Climate Change Service 50m
The Copernicus Climate Change Service (C3S), implemented by ECMWF on behalf of the European Union, has developed a seasonal forecast component which has now reached operational status. Based on a multi-system framework, this component of the service offers data previously not widely available, and has proven very popular with users. In addition to a large volume and variety of seasonal forecast data, C3S also offers users examples of information derived from such data, including multi-system combination forecasts.
This presentation will describe the set up of the C3S multi-system and the scope of the resulting data base, as well as of the set of example products issued as part of the operational service. Challenges related to verification and calibration of products and of combination of data from multiple sources will be described from a producer's perspective, setting the scene for a comparison with the user perspective addressed by the presentation following in the programme of the event.Speaker: Anca Brookshaw (ECMWF) -
12:10
Subseasonal and seasonal climate forecast applications 50m
Substantial progress has taken place in the development of subseasonal and seasonal climate forecast systems. Operational systems now provide a comprehensive set of forecast products, often using the multi-model approach, and a consistent set of forecast quality metrics. However, there are still important gaps in illustrating the applicability of the forecast information. While the advent of climate services has opened an avenue for the participation of social sciences and humanities in the co-production of action-oriented climate information at subseasonal and seasonal time scales, the engagement of climate forecasters and analysts in this exercise has been disappointing. This presentation will describe some of the challenges the co-production of user-oriented climate information using subseasonal and seasonal forecasts needs to address.
The challenges will be considered in an operational context. The definition of relevant indicators, user-oriented verification, the relevant merits of the forecast post-processing and the communication of forecast uncertainty will all be illustrated using some simple user-driven applications. The discussion will focus on how these challenges can be addressed jointly by user and forecasting communities.
The discussion will also deal with the tension appearing between the quality measures communicated by forecast providers and the quality of the actual information reaching the applications. Such tension is the result of the imperfect representation of the quality of real-time forecasts by forecast quality measures based on hindcasts, the skill degradation intrinsically associated with bias adjustment practices and the mismatch between what the forecasters often provide and the actual user needs.Speaker: Francisco Doblas-Reyes (BSC) -
13:00
Lunch break 1h ECMWF
ECMWF
Reading -
14:00
Analyzing causal pathways of the stratospheric polar vortex using machine learning tools 50m
Extremely strong and weak phases of the stratospheric polar vortex are known to affect the North Atlantic Oscillation and therefore mid-latitude weather. Troposphere–Stratosphere coupling thus provides an important source of predictability for winter forecast on subseasonal to seasonal timescales. However, the exact mechanisms are still unclear and their representations in climate models vary.
Extracting the physically relevant teleconnection pathways from observation or model data remains challenging. One major issue is to separate the signal from the noise given large internal atmospheric variability. This is compounded by varying dimensions in space and time and competing effects of different processes.
Here, to overcome these current limitations, we apply a novel data-based method, called causal effect networks (Kretschmer et al. 2016), to analyze causal pathways of the stratospheric polar vortex. This approach allows evaluating troposphere-stratosphere coupling with more confidence than using solely correlation analysis. We focus on winter circulation in the North Atlantic/European region and particularly assess the role of stratospheric polar vortex variability to predict the North Atlantic Oscillation.Speaker: Marlene Kretschmer (PIK Potsdam) -
14:50
Where our science ambitions meet computing and data handling limitations 50m
As sub-seasonal to seasonal predictive skill is determined by all components relevant for weather and climate prediction, namely initial conditions, Earth-system model components and forcings, it can serve as a driver for joint developments across both communities. Such developments consist of both science and technology elements, and they become rather challenging when aiming for future convection resolving capability. Examples for science elements are fully coupled and seamless ensemble data assimilation and forecasting systems, diagnostic methods tracing the roots of model errors, or parameter optimisation to adjust uncertain variables in parameterisations and to define model uncertainty in ensembles. Eventually, advanced science can only be implemented through advanced technology, which is seriously affected by the lack of 'free' computational performance growth in the post-Moore's law era and by the unmanageable volume and diversity of big data. This talk aims to demonstrate which investment in computing, data handling technology and artificial intelligence is needed for realising the seamless weather and climate prediction system of the future.
Speaker: Peter Bauer (ECMWF) -
15:40
Close of seminar 15m
-
09:10
-
09:10
→
15:55