This paper presents the argument that existing polar prediction systems do not yet meet users’ needs and outlines possible ways forward in advancing prediction capacity in polar regions and beyond.
The climate of the Arctic has been changing more rapidly in recent decades than any other region of this planet. The rapid rise in near-surface Arctic air temperatures, about twice as fast as the global increase (Hansen et al. 2010), is called the Arctic amplification (e.g., Holland and Bitz 2003). Its manifestation in terms of decrease in sea ice coverage provides opportunities, but at the same time new risks are emerging. Using the Northern Sea Route, for example, ships can reduce the distance of their journey between Europe and the North Pacific region by more than 40%. In fact, journeys through the Arctic, which are projected to become increasingly feasible as climate change continues (Smith and Stephenson 2013), could provide an opportunity for cutting greenhouse gas emissions. At the same time, the environmental consequences of disasters in the Arctic, such as oil spills, are likely to be worse than in other regions (Emmerson and Lahn 2012). To effectively manage the opportunities and risks associated with climate change, therefore, it is argued that skillful prediction systems tailored to the particularities of the polar regions are needed.
The mounting interest in the polar regions from the general public has also become evident for example from increased levels of tourism in both hemispheres (Hall and Saarinen 2010). The ongoing and projected changes in polar regions and increases in economic activity also lead to concerns for indigenous societies and northern communities. Traditional means of predicting environmental conditions, for example, may become invalid in a changing climate with changing predictor relationships (Holland and Stroeve 2011) and all northern communities are at an increasing risk from accidents such as oil or cargo spills associated with increased economic and transportation activities.
Even though climate change in Antarctica is less apparent than in the Arctic, with the exception of the Antarctic Peninsula and West Antarctica, demand for skillful prediction systems is increasing there too. In the southern polar regions the main stakeholders are the logistics community, which provides essential services to the research community such as flights to and from Antarctica, and tourists and research expeditions, which can encounter extremely harsh conditions (Fig. 1) (Powers et al. 2012). It is through the effective running of essential logistical activities, which in turn depend on skillful environmental predictions, that important scientific challenges such as issuing trustworthy projections of future global sea level rise can be addressed.
Research icebreaker Polarstern on a nocturnal ice station during its winter expedition to Antarctica in 2013. The harsh environmental conditions of the polar regions pose substantial logistical challenges, which call for a concerted international effort to ensure scientific progress. [Photo courtesy of S. Hendricks, Alfred Wegener Institute (AWI).]
Citation: Bulletin of the American Meteorological Society 97, 9; 10.1175/BAMS-D-14-00246.1
In the following we will argue that the science of polar environmental prediction is still in its infancy and that significant progress can be achieved through a concerted international prediction effort, putting the polar regions into focus [see also Eicken (2013)].
HOW TO IMPROVE POLAR PREDICTION CAPACITY?
First, let us turn our attention to the questions of how well existing polar prediction capacity is developed and how progress can be ensured over the coming years. The following discussion will be centered around three research pillars—namely, service-oriented research, forecasting system research, and underpinning research (see Fig. 2). A more comprehensive list of research priorities related to polar prediction is given by PPP Steering Group (2013, 2014).
Research areas that will need to be addressed to advance polar predictive capacity [adapted from PPP Steering Group (2013)].
Citation: Bulletin of the American Meteorological Society 97, 9; 10.1175/BAMS-D-14-00246.1
Service-oriented research.
User applications.
While there is great merit in conducting basic scientific research to better explain fundamental atmosphere–ocean–ice–land processes, the societal value of such knowledge depends on its relevance and application to social, economic, and environmental problems and issues in polar regions. Value accrues through the provision of services, such as weather warnings and ice forecasts, to various users or actors—the individuals, businesses, communities, and agencies that are sensitive to environment-related risks or that manage its effects and consequences. Service-oriented research, rooted in the social and interdisciplinary sciences, is conducted to understand the decision-making context in which these individuals live and organizations operate, appreciating that exposure, vulnerability, and the capacity to respond to weather and ice hazards are largely driven by many interrelated nonweather factors (e.g., cultural and social practices, international demand and pricing of resource commodities, health status of residents). Such research can inform and direct the design and implementation of weather-related services to enhance their effectiveness leading to improved material outcomes (e.g., safety, mobility, productivity, etc.).
Preparatory research should include reviewing existing and planned research to better define and prioritize potential benefit areas and develop a baseline of current experience, use, and perception of services. While presently there is a dearth of social scientific research that explicitly treats the use and value of weather information in polar regions, established programs of study examining adaptation to anthropogenic climate change offer potential opportunities for collaboration on research at the temporal scale of weather-related hazards (e.g., ACIA 2004; Dawson et al. 2014; Lamers et al. 2011; Victoria Team and Manderson 2011). This research has identified several unique pressures that contribute to the rationale for making the polar regions a target for the application of improved environmental prediction science and services and points to several benefit areas—ideas that are also reflected in recent work by the World Meteorological Organization (WMO) Executive Council Panel on Polar Observations, Research and Services (EC PORS) Task Team (available from www.wmo.int/pages/prog/www/WIGOS_6_EC_PORS/EC-PORS-3.html).
Among the challenges for service-oriented research is achieving the necessary balance between depth and breadth. For example, intensive community-based research involving interviews and ethnographic techniques is often required to unpack the intricacies of decision-making among residents and leaders. However, the generalizability of findings can be left unaddressed given limited resources (time as much as funding) to conduct parallel work in several communities over multiple years. Other challenges include the limited availability and accessibility to secondary social and economic data; facilitating actor and stakeholder participation, engagement, and partnership within research projects; and securing the involvement and coordination of expertise across multiple social science and other disciplines.
Verification.
Another important aspect of service-oriented research involves forecast verification. Verification can provide users with information about forecast quality to guide their decision-making procedures, as well as useful feedback to the forecasting community to improve their own systems. Traditionally, forecast verification has focused on weather variables that are of little direct value for most users of weather information, such as the 500-hPa geopotential height. Increasingly though, surface weather parameters like temperature at 2-m height, wind speed at 10-m height, and precipitation are part of standard verification. The diversity of verification measures has been relatively limited with a strong emphasis on basic statistical measures like root-mean-square error and correlation metrics. Standard verification has moreover mostly concentrated on midlatitude and tropical regions. Only very recently has the skill of current operational forecasting systems in the polar regions been considered (Bromwich et al. 2005; Jung and Leutbecher 2007; Jung and Matsueda 2016; Bauer et al. 2016). More work will be needed, especially on the verification of near-surface parameters as well as snow and sea ice characteristics (especially drift and deformation).
Some of the biggest challenges in forecast verification relate to the quality and quantity of observations. In fact, representative observational data are the cornerstone of all successful verification activities. Given the notorious sparseness or even complete lack of conventional observations in the polar regions (Fig. 3), progress in quantifying and monitoring the skill of weather and environmental forecasts will hinge on the availability of additional observations or better usage of satellite data.
Conventional observations that were assimilated by the operational forecasting system at ECMWF on 15 Apr 2015. Different colors are used for different observation types (see legend).
Citation: Bulletin of the American Meteorological Society 97, 9; 10.1175/BAMS-D-14-00246.1
Forecast verification against analyses (which are influenced by the model itself during the data assimilation process) is common practice, because the model introduces spatial and temporal consistency to sparse data and analysis errors are usually much smaller than forecast errors at medium and extended range. This approach can have shortcomings in parts of the world, including the polar regions, where the sparseness of high-quality observations and the difficulty of assimilating satellite observations leads to a very strong influence of the models’ first guess on the analysis. Enhanced verification in observation space (e.g., satellite data simulators) and increasing analysis quality need high priority.
In recent years, there has been a shift in how verification is perceived. It has been widely recognized that verification activities should focus more strongly on user-relevant forecast aspects, that more advanced diagnostic verification techniques are required, and that the usefulness of verification depends on the availability of sufficient high-quality observational data. These developments need to be strengthened and promoted in the coming years to advance forecast verification in polar regions.
Forecasting system research.
The elements of forecasting system research—namely, observations, modeling, data assimilation, and ensemble forecasting (Fig. 2)—are no different to those required at lower latitudes. What is important to point out, however, is that there are certain polar-specific aspects that need special consideration in order to enhance our predictive capacity—some of these aspects will be highlighted below.
Observations.
The polar regions are among the most sparsely observed parts of the globe by conventional observing systems such as surface meteorological stations, radiosonde stations, and aircraft reports. Figure 3, which shows conventional observations of different types that were assimilated by ECMWF on 15 April 2015, illustrates the situation: contrast the dense network of surface stations (blue dots: “SYNOPS”) over Scandinavia with the sparse network over the rest of the Arctic, or compare the coarse but arguably adequate network of radiosonde stations (yellow triangles: “TEMP”) over Eurasia with the handful of stations over Antarctica. The polar oceans are also sparsely observed by the Argo array of automated profiling floats (e.g., Roemmich and Gilson 2009), implying challenges in coupled model initialization.
The polar regions are barely sampled by geostationary satellites but generally have a denser sampling by polar-orbiting satellites, providing the potential for improvements in satellite sounding such as the Infrared Atmospheric Sounding Interferometer (IASI) sounder, or sea ice thickness from Cryosphere Satellite-2 (CryoSat-2) (Laxon et al. 2013), Soil Moisture and Ocean Salinity (SMOS) (Kaleschke et al. 2012; Tian-Kunze et al. 2014), and Sentinel-1 and the planned Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) (Kwok 2010; Kern and Spreen 2015). Using satellite-based observations of the polar surface is challenging owing to the presence of snow-covered sea ice, which makes it difficult to determine parameters such as ocean surface temperature, surface winds, and precipitation. Differentiating between snow- and ice-covered surfaces and clouds in the atmosphere has also been a long-running challenge. Making better use of existing and new satellite-based observations is a must for improving forecast initialization and verification.
Given that observations are key to producing accurate initial conditions and hence forecasts, relatively sparse observational coverage in polar regions may be one explanation as to why the skill of weather forecasts in polar regions is relatively low (see also Jung and Leutbecher 2007; Jung and Matsueda 2016; Bauer et al. 2016). In addition, data assimilation systems are not adequate to optimally exploit the information provided by existing observations, as will be discussed below.
The relative remoteness and harsh environmental conditions of the polar regions are always going to provide a barrier to enhanced observations. With improved technology and power systems, the barrier is becoming more of a financial one than a logistical one: improved observations of the polar regions are possible, but are they worth the cost? To answer this, observing system experiments (OSEs) are required [see, e.g., Boullot et al. (2016)], in which specific observations are withheld (denied) during the data assimilation process, with a particular focus on user requirements for these regions. To carry out these experiments, a sustained observing period is required with significantly enhanced spatial and temporal coverage—a Year of Polar Prediction (see below). In this respect, increasing the frequency of observations from existing stations and vessels (e.g., Inoue et al. 2013; Yamazaki et al. 2015; Inoue et al. 2015) and adding additional mobile observing systems such as buoys (Inoue et al. 2009; Meredith et al. 2013) would be excellent options. In addition, periods of intense process-focused field campaigns are required to provide comprehensive observations of processes that are known to be currently poorly represented in coupled models (e.g., Holtslag et al. 2013; Pithan et al. 2014). Furthermore, increased levels of activity in polar regions suggests that additional observations from new voluntary observing platforms may become available in the future. Effectively engaging with stakeholders, therefore, becomes a key element for improving the polar observing system.
Modeling.
Numerical models of the atmosphere, ocean, sea ice, snow, and land play an increasingly important role in prediction. For example, models are used to carry out short- to seasonal-range weather and environmental forecasts, they form an important element in every data assimilation scheme, they serve as a virtual laboratory to carry out experiments devised to understand the functioning of the coupled atmosphere–ocean–sea ice–land system, and they can aid the design of future observing systems (e.g., for satellite missions) through so-called observing system simulation experiments (OSSEs; e.g., Masutani et al. 2010).
Although numerical models have come a long way, even state-of-the-art systems show substantial shortcomings in the representation of certain key processes. For example, skillful model simulations of stable planetary boundary layers and tenuous polar clouds remain elusive (e.g., Sandu et al. 2013; Bromwich et al. 2013). The shallowness of stable planetary boundary layers, layering of low-level clouds, the smaller spatial scale of rotational systems (e.g., polar cyclones) due to the relatively small Rossby radius of deformation along with the presence of steep topographic features in Greenland and Antarctica all suggest that polar predictions will benefit from increased horizontal and vertical resolution (Jung and Rhines 2007; Renfrew et al. 2009; Elvidge et al. 2015). However, while some of the existing problems may be overcome by increased resolution accessible via the projected availability of supercomputing resources during the coming years, it is certain that the parameterizations of polar subgrid-scale processes will remain an important area of research for the foreseeable future (e.g., Holtslag et al. 2013; Vihma et al. 2014).
It is interesting, in this context, to compare the relative importance of different atmospheric processes for different regions [see Bourassa et al. (2013) for a related discussion on turbulent surface fluxes]. Vertical profiles of mean initial temperature tendencies due to various dynamical and physical processes obtained from 1-day forecasts with the European Centre for Medium-Weather Forecasts (ECMWF) model are shown in Fig. 4 for four different regions during boreal winter: the sea ice–free and sea ice–covered Arctic as well as oceanic regions in the Northern Hemisphere midlatitudes and tropics. Initial temperature tendencies are temporal changes in temperature arising from the governing equations solved by the model directly after initializing the forecasts. Note that the mean total initial temperature tendency should be close to zero in the absence of model drift (Rodwell and Jung 2008) if averaging is done over a sufficiently large number of cases (Klinker and Sardeshmukh 1992). In the tropics, for example, strong incoming solar radiation together with boundary layer turbulence leads to a heating of lower-atmospheric levels, while longwave radiation cools away from the surface. This radiative tendency profile is largely balanced by deep convection, which contributes to effectively removing instability. A similar balance can be found in oceanic regions of middle and high latitudes (Figs. 4a,c). However, away from the tropics the importance of dynamical cooling (cold air advection) and boundary layer heating is more pronounced. Radically different heating profiles can be found during boreal winter in ice-covered parts of the Arctic Ocean (Fig. 4b): in the free atmosphere, dynamical heating due to the inflow of relatively warm air from lower latitudes is balanced by longwave radiative cooling; in the polar boundary layer, the situation is more complex with vertical diffusion playing a significant role as well. The modeled tendencies are the largest in the case of Arctic open ocean and the smallest values are found in the sea ice–covered ocean.
Vertical profiles of mean 1-day initial tendencies of temperature (K day−1) averaged over different regions: (a) sea ice–free Arctic Ocean, (b) sea ice–covered Arctic Ocean, (c) Northern Hemisphere midlatitude oceans, and (d) tropical oceans. Tendencies from the dominant dynamics (black) and physical processes are shown: radiation in blue, vertical diffusion in green, convection in red, and large-scale precipitation in yellow. Results are based on weather forecasts during boreal winter with the ECMWF model started every 6 h during the period Dec through Feb from 1979 to 2013.
Citation: Bulletin of the American Meteorological Society 97, 9; 10.1175/BAMS-D-14-00246.1
Another interesting perspective arises when vertical profiles of the standard deviation of initial temperature tendencies are considered (Fig. 5). Large day-to-day changes in dynamical temperature tendencies can be found everywhere. However, it is only in the tropics that the variability associated with the dynamics is matched by that linked to fast convective processes. In middle and high latitudes the situation is different with both convection and large-scale precipitation (microphysics) and, to a lesser extent, radiation playing a role. Again, the ice-covered Arctic Ocean stands out owing to the relative lack of fast processes in the free atmosphere. As models have problems properly representing the low-level mixed-phase clouds and shallow boundary layers, there are likely to be larger uncertainties in Figs. 4b and 5b than for the other areas. Nevertheless, the above tendency diagnostics highlight the fact that atmospheric regimes in the polar regions can be quite different (ice covered versus ice free) and unique (ice-covered parts) as well as radically different to lower latitudes.
As in Fig. 4, but for the standard deviation of daily initial temperature tendencies.
Citation: Bulletin of the American Meteorological Society 97, 9; 10.1175/BAMS-D-14-00246.1
A survey of the global forecasting systems used for short-range and medium-range predictions, such as the ones that contribute to The Observing System Research and Predictability Experiment (THORPEX) Interactive Grand Global Ensemble (TIGGE; Bougeault et al. 2010), suggests that many aspects relevant to the polar regions are still missing in existing systems. For example, many centers still use atmospheric–land models; in these forecasting systems, sea ice is persisted throughout the forecast. Obviously these “weather” forecasting systems are not tailored to provide predictive information on sea ice characteristics and their future evolution. The expected increase in shipping traffic in the Arctic will require new kinds of forecast products that provide information about sea ice leads, velocity, and pressure; these needs can only be met by incorporating dynamic–thermodynamic sea ice models into forecasting systems. Interestingly, existing sea ice models, which were developed with relatively coarse-resolution climate applications in mind, start to show deformation characteristics such as leads when their horizontal resolution is increased (Fig. 6). It will be important to assess the realism of these features and explore their predictability. Furthermore, persisting sea ice throughout the forecast may lead to sizeable errors in near-surface variables such as air temperature during periods of strong advances and retreats of the sea ice edge such as in autumn and spring. An example of the mean near-surface temperature difference for October 2011 between forecasting experiments with observed and persistent sea ice field is shown in Fig. 7. Evidently, mean differences of up to 4 K after 5 days into the forecast can be found close to the ice edge. Not including coupling between sea ice and atmosphere can result in missing dynamical responses that have consequences beyond the sea ice region and not just near surface (Bhatt et al. 2008). While it may be justified for shorter-term prediction in middle latitudes to use atmosphere-only systems, the cryosphere and the ocean need to be explicitly incorporated when it comes to polar prediction [see also Smith et al. (2013)].
Sea ice thickness (m) on 30 Mar 2001 as simulated by the MITgcm (sea ice–ocean model forced with reanalysis data) at a horizontal resolution of about 4 km. The simulation is very similar to the one described in Nguyen et al. (2012).
Citation: Bulletin of the American Meteorological Society 97, 9; 10.1175/BAMS-D-14-00246.1
Mean 2-m temperature difference (in K) between hindcast experiments using observed and persisted sea ice and sea surface temperature for Oct 2011: (a) day-2, (b) day-5, (c) day-7, and (b) day-10 forecasts with the ECMWF forecasting system.
Citation: Bulletin of the American Meteorological Society 97, 9; 10.1175/BAMS-D-14-00246.1
Furthermore, there is clearly scope for using regional weather prediction systems in polar regions as they offer some advantages compared to global forecast models. For example, polar-optimized physics can be used such as for mixed-phase clouds and for more comprehensive sea ice specifications (Hines et al. 2015). Very large contrasts in turbulent fluxes of sensible and latent heat are frequently encountered along the sea ice edges, which gives rise to characteristic mesoscale phenomena such as low-level jets, vigorous convection, and occasionally polar lows (e.g., Kristjánsson et al. 2011), which require high spatial resolution. Coupling to models for the upper ocean is potentially important since strong low-level winds can invigorate upper-ocean mixing and thus positive feedbacks when warm subsurface water is brought to the surface (Linders and Saetra 2010). Moreover, the use of very high spatial resolution (1 km or so) where nonhydrostatic dynamics becomes important better captures the topographic forcing upon near-surface winds in regions of complex terrain (e.g., Steinhoff et al. 2013). One of the better known regional polar NWP efforts is the Antarctic Mesoscale Prediction System (AMPS; Powers et al. 2012), which telescopes from a 30-km grid covering the Southern Ocean to a 1.1-km nested grid focused on the rugged terrain near Ross Island to support terminal airport forecasts for aircraft coming from New Zealand.
Data assimilation.
In numerical weather prediction, data assimilation systems are used to produce the initial conditions for forecasts. These so-called analyses are based on the numerical model (also used for forecasting) and observations with an optimization algorithm that combines the two such that a physically plausible estimate is derived that matches the model prediction and observations within their respective error margins (Kalnay 2003). The quality of the analysis is of fundamental importance for forecast skill since forecasting on the time scales considered here is, to a large extent, an initial condition problem. Generally, the sensitivity of forecasts to the analysis changes between short, medium, and extended range from smaller-scale and fast processes (e.g., turbulence, clouds, convection) to larger-scale and slow processes (e.g., planetary waves, ocean, snow, and sea ice dynamics).
Modern global weather forecasting employs data assimilation systems that use time integrations of the three-dimensional model at 15–25-km resolution and 50–100 vertical levels [O(109) grid cells] together with O(107) observations resulting in very large numerical optimization problems (e.g., Rabier et al. 2000; Kalnay 2003). Ensemble analysis systems (e.g., Houtekamer and Mitchell 1998) aim at additionally specifying the uncertainty of the analysis that is required for deriving the above-mentioned model error margins but also serve as initializations for ensemble forecasts.
Over polar areas, shortcomings in all three main data assimilation components (models, observations, and assimilation algorithms) contribute to suboptimal state estimates (e.g., Jung and Leutbecher 2007; Bauer et al. 2016) leading to a detrimental impact on forecast skill across all time scales. In the atmosphere in which boundary layer processes and atmosphere–surface interaction—particularly with variable sea ice coverage—are shallow and dominant, the small scale of cyclonic systems (e.g., polar lows) and the interaction of the flow with extremely steep orography are currently not well resolved in global models (and observations) and even less so in data assimilation systems (Tilinina et al. 2014). Observations are sparse and mostly lacking over sea ice and the Antarctic continent. Satellite data are more difficult to interpret owing to, for example, little radiative contrast between the surface and atmosphere. The specification of model and observation uncertainty, required to balance the contributions from observations and model in the analysis, is complex because other processes dominate the error budget and spatial error structures are different from those at lower latitudes.
It will be important to address model improvement, observations, and data assimilation methods together. In doing so, polar-specific aspects such as the atmosphere–sea ice–ocean interaction and spatial resolution, enhanced surface-based observational networks and satellite data exploitation, assimilation methods more optimally tuned to high-latitude conditions, and coupled atmosphere–ocean–sea ice data assimilation at regional and global scales need to be emphasized
Ensemble forecasting.
Ensemble forecasting is an approach to quantifying the uncertainty of weather or climate forecasts (e.g., Leutbecher and Palmer 2008). The main challenge when designing ensemble prediction systems (EPSs) lies in the proper representation of initial conditions (and their errors) and of model uncertainty to obtain reliable estimates of prediction error and forecast probabilities. Most operational EPSs employ optimal perturbations to represent initial condition uncertainty. Here, optimality refers to perturbations that are designed to ensure their growth and, hence, the increase of the ensemble spread throughout the early stages of the forecasts. In the atmospheric midlatitudes, baroclinic instability dominates the early stage of forecast error growth (e.g., Buizza and Palmer 1995; Toth and Kalnay 1993); in the tropical atmosphere, on the other hand, convective instability plays the dominant role (e.g., Buizza et al. 1999; Toth and Kalnay 1993). Although it can be anticipated that baroclinic instability has some role to play in the polar regions, research needs to be carried out to identify other more polar-specific sources of perturbation growth—for the atmosphere as well as for other components of the polar climate system such as the ocean and the sea ice.
Given the limitations of existing models in representing some of the key processes in the polar regions, it will be imperative to properly represent model inaccuracy in operational ensemble forecasts from hourly to seasonal time scales and beyond. Different approaches have been suggested including multimodel ensembles and stochastic parameterizations (e.g., Palmer et al. 2005). Most of the existing schemes were developed with nonpolar regions in mind, so that it will be important to assess their performance in polar regions taking into account polar-specific aspects, such as the absence of convection in ice-covered regions and the need to describe uncertainty for coupled processes at the interface between the atmosphere and land/snow/sea ice. Furthermore, given that routine weather forecasts are likely to be carried out with coupled models by the end of this decade, as they are already used for subseasonal and seasonal forecasting, the representation of model uncertainty in sea ice, ocean, land surface, and land-based hydrology will also need to be addressed [see, e.g., Juricke et al. (2014) for first steps in this direction].
In short, it can be argued that with a few exceptions (e.g., Aspelien et al. 2011; Kristiansen et al. 2011) existing work on operational EPSs has focused on nonpolar regions. Because of this, relatively little is known about the quality of ensemble forecasts, including the associated probability forecasts, in polar regions. In fact, a lot of progress in the provision of environmental information can be made by raising awareness of the importance of polar ensemble forecasting by improving polar-specific aspects in EPSs (e.g., the presence of sea ice) and by applying existing ensemble verification techniques to the polar regions.
Underpinning research.
Predictability.
Predictability research is primarily concerned with the mechanisms that potentially influence forecast skill at different time scales. The predictability of a system is determined by its instabilities and nonlinearities and by the structure of the imperfections (analysis and model error) in the system (e.g., Palmer et al. 2005). Because of its relative persistence or stability, sea ice anomalies are usually considered a potential source of predictability, especially on subseasonal and seasonal time scales (Chevallier and Salas-Mélia 2012; Tietsche et al. 2014; Day et al. 2014). In fact, predictability of Arctic sea ice has attracted considerable attention in recent years, especially when it comes to predicting sea ice extent anomalies in late summer. Interestingly, there is a large gap between potential predictability estimates of late summer Arctic sea ice extent (e.g., Guemas et al. 2016; Juricke et al. 2014), which provide a relatively optimistic view, and actual skill, which is rather modest (Wang et al. 2013; Stroeve et al. 2014). This highlights the fact that the potential of seasonal to interannual sea ice prediction has not been fully exploited yet and/or potential predictability estimates are overly optimistic owing to insufficient representation of the underlying initial and model uncertainties [see Day et al. (2014) for points on the importance of sea ice thickness initialization].
Perhaps because of these shortcomings, statistical forecasts of Arctic sea ice cover currently perform just as well as those performed with dynamical models (Stroeve et al. 2014). This is reminiscent of the case of ENSO forecasting, where even after years of development, dynamical models are only marginally more skillful than statistical models at seasonal time scales (Barnston et al. 2012). However, climate change in the Arctic is happening more rapidly than any other region on Earth and there is evidence that these changes could fundamentally affect predictor–predictand relationships in the region, making it difficult to both train and trust such models (Holland and Stroeve 2011). It is therefore imperative for seasonal polar prediction that coupled models improve.
The presence of sea ice, land ice, and snow in the polar regions in conjunction with midtropospheric inflows of relatively warm air from the midlatitudes (Fig. 4) leads, at times, to the development of shallow and stably stratified planetary boundary layers (PBLs) in the interior of the Arctic and Antarctic during wintertime (Holtslag et al. 2013). The resulting decoupling of the boundary layer from the free atmosphere may have implications for the predictability of the system. On the other hand, extreme temperature contrasts across the ice edge can lead to very unstable PBLs and to turbulent surface heat fluxes in excess of 1,000 W m−2 over the adjacent open ocean regions (Papritz et al. 2015). Depending on the dynamical conditions associated with the free-tropospheric outflowing air masses, very strong, hurricane-like vortices with diameters typically of a few hundred kilometers may develop within a period of a few hours, under the influence of sensible and latent heating from the open ocean (e.g., Rasmussen and Turner 2003; Kristjánsson et al. 2011). These polar lows are responsible for some of the most dangerous weather in the Arctic, owing to strong winds, heavy snowfall, and icing on ships and installations. Furthermore, their predictability is highly variable (while some polar lows are very well forecasted, some still come “out of the blue”), because of the fast development over areas with sparse observations and their small scales. It is also likely that some aspects of model formulations in terms of spatial resolution and parameterized processes are inadequate. Finally, the regions where polar lows strike may change as the Arctic sea ice continues to decline. It is to be expected that the regional vulnerability to polar lows will be much higher because of these changes, as necessary preparedness may be neglected over areas such as the Kara and Laptev Seas.
From the above discussion, it can be argued that our existing knowledge on predictability, which is primarily obtained from studies in lower latitudes, is not easily transferable owing to particular characteristics of the polar regions. Predictability research that focuses on polar regions is therefore urgently needed.
Diagnostics. Forecast error diagnosis is a means to identifying possible weaknesses in the different components of operational forecasting systems. Proper diagnosis, therefore, can help to prioritize research activities in relation to their relative importance.
Substantial progress could be achieved by employing diagnostic methods that have been successfully used in lower latitudes [see Rodwell and Jung (2010) for a more comprehensive discussion]. It would be desirable, for example, to identify situations where existing prediction systems have difficulties; backtracking of forecast busts (unusually large forecast errors) throughout the forecast would be one promising approach (Rodwell et al. 2013).
Another promising way forward would be to employ initial tendency diagnostics in polar regions using output from data assimilation systems. By evaluating the initial drift of the model in an NWP context it will be possible to identify possible model weaknesses that result in systematic model error (Rodwell and Palmer 2007; Rodwell and Jung 2008).
Global linkages.
Teleconnections between the polar regions and lower latitudes have attracted considerable attention in recent years. In particular, the possible influence of “Arctic amplification” on the frequency of occurrence of high-impact events over the Northern Hemisphere has been a matter of intensive discussion and controversy (Cohen et al. 2014; Barnes and Screen 2015; Jung et al. 2015). Compared to tropical–extratropical interactions, for which a vast body of literature is available, relatively little is known about the dynamics of polar–lower-latitude linkages, especially for the atmosphere. In fact, it could be argued that at present we are at a preconsensus state (Cohen et al. 2014), not unlike where ENSO research was in the 1970s and early 1980s (Overland et al. 2015; Jung et al. 2015). To further our understanding of polar–lower-latitude linkages—from their source regions, via atmospheric teleconnections to the places where related changes in weather and climate impact society—it will be important that experts on polar atmospheric processes (i.e., the polar research community) join forces with atmospheric dynamicists traditionally working more on middle-latitude phenomena.
It could be argued that further insight could be gained by studying polar–lower-latitude linkages also from a prediction perspective. In fact, while teleconnection patterns are well-studied phenomena, there is little quantitative knowledge about their role in transferring forecast skill (or uncertainty) from the polar regions into the midlatitudes and vice versa. Given the relatively poor observational coverage in polar regions (Fig. 3), for example, it seems plausible that enhanced observational capacity in polar regions would lead to improved midlatitude predictions if polar–lower-latitude linkages were sufficiently strong. In fact, recent research indicates that better Arctic predictions will lead to better medium-range and subseasonal forecasts in Northern Hemisphere middle latitudes, especially over Eurasia and North America (Jung et al. 2014; Hines et al. 2015). Second, by considering the interplay between polar and nonpolar regions from a prediction perspective on time scales from daily to seasonal, polar–lower-latitude linkages involving relatively fast atmospheric processes could actually be verified. The underlying premise is that the atmospheric processes involved are actually the same across a wide range of time scales [see Palmer et al. (2008) for a more detailed discussion].
In short, it is expected that research on global linkages will enhance our understanding of the role of the polar regions in the global climate system, both in terms of the underlying dynamics and in terms of predictability on time scales from days to seasons and beyond.
INTERNATIONAL COOPERATION.
To advance predictive capacity in polar regions, a strong element of coordination will be required. In the following, we introduce two (related) initiatives that provide an international framework through which collaboration between natural and social scientists, operational prediction centers, and stakeholders from different nations can be effectively facilitated.
Polar Prediction Project.
The growing need for reliable polar prediction capabilities has been recognized by the WMO when its World Weather Research Programme (WWRP) established the Polar Prediction Project (PPP) as one of three legacy activities of THORPEX. The aim of PPP, a 10-year endeavor (2013–22), is to “Promote cooperative international research enabling development of improved weather and environmental prediction services for the polar regions, on time scales from hours to seasonal.” To achieve its goals, PPP enhances international and interdisciplinary collaboration through the development of strong linkages with related initiatives; strengthens linkages between academia, research institutions, and operational forecasting centers; promotes interactions and communication between research and stakeholders; and fosters education and outreach.
Flagship research activities of PPP include i) advancing sea ice prediction, ii) understanding polar–lower-latitude linkages along with their role in weather and climate prediction, and iii) the Year of Polar Prediction—an intensive observational and modeling period planned for mid-2017 to mid-2019 (see below for details).
PPP is supported through the International Coordination Office (ICO) for Polar Prediction, which is hosted by the Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research, in Germany, and informs about, promotes, and coordinates PPP-related activities. Further details, including the PPP Implementation Plan (PPP Steering Group 2013), are available from the ICO’s website (http://polarprediction.net).
Year of Polar Prediction.
One particularly important international initiative is the Year of Polar Prediction (YOPP). YOPP is a key element of PPP and provides an extended period of coordinated intensive observational and modeling activities in order to improve prediction capabilities for the Arctic, the Antarctic, and beyond on a wide range of time scales from hours to seasons, supporting improved weather and climate services, including the Global Framework for Climate Services (GFCS). This concerted effort will be augmented by research into forecast–stakeholder interaction, verification, and a strong educational component. Being focused on polar prediction rather than a very broad range of activities, YOPP is quite different from the International Polar Year (IPY; 2007/08). Prediction of sea ice and other key variables such as visibility, wind, and precipitation will be central to YOPP.
Extra observations will be crucial to YOPP in order to test an augmented polar observing system, generate the knowledge necessary to improve the representation of key polar processes in models, and provide ground truthing that is so important to exploit the full potential of the spaceborne satellite network. YOPP will also encourage research, development, and employment of innovative systems.
Following the success of the virtual field campaign during the Year of Tropical Convection (YOTC; Moncrieff et al. 2012), YOPP will also have a strong virtual component through support from the numerical modeling community, encompassing high-resolution model simulations that include important polar-specific aspects. Operational model runs will cover time scales from hours to seasons, with a particular focus on sea ice, since for polar regions sea ice is both a critically important environmental variable to be predicted and a strong modulator of other weather-related predictands across a wide range of time scales.
Output from operational models, including specific additional diagnostics, and dedicated numerical experiments during YOPP will be archived and made available for researchers to better understand strengths and shortcomings of existing prediction systems. The new archive will be valuable in itself, even without the planned additional observations that will be assimilated into models. It will certainly help improve process understanding at a detailed level.
Regarding the data strategy, YOPP will take into account lessons learned from the IPY. This includes developing a YOPP data portal that builds on the experience of the Global Cryosphere Watch (GCW), including the use of consistent metadata and pointers to other online locations where data can be retrieved. A small number of data centers willing to archive YOPP data (and to support the process) and able to provide digital object identifiers (DOIs) will be identified. Datasets must be open access and, where observations are suited for real-time operational use, submission through the Global Telecommunication System (GTS)/WMO Information System (WIS) should be mandatory. Special attention will be given to WMO standards including the Binary Universal Form for the Representation of meteorological data (BUFR). Finally, all datasets should be published in data journals such as Earth System Science Data (ESSD), and a YOPP special issue in ESSD is desirable.
YOPP will also explore largely uncharted territory in the area of polar forecast verification, it will contribute to our understanding of the value of improved polar prediction capabilities, and it will help to educate the next generation of scientists. YOPP will be carried out in three stages (Fig. 8): the ongoing YOPP Preparation Phase that started in 2013, the YOPP Phase from mid-2017 to mid-2019, and the YOPP Consolidation Phase from mid-2019 to 2023. A more detailed description is available from the YOPP Implementation Plan (PPP Steering Group 2014) and in a meeting report from a high-level planning event—the YOPP Summit—that was held at WMO headquarters from 13 to 15 July 2015 (Goessling et al. 2016).
Three stages of the YOPP, including main activities [adapted from PPP Steering Group (2014)].
Citation: Bulletin of the American Meteorological Society 97, 9; 10.1175/BAMS-D-14-00246.1
DISCUSSION.
Given the increasing interest in polar regions, it has been argued that existing prediction capacity needs to be urgently enhanced to effectively manage the risks and opportunities associated with growing human activities and to support local communities in a rapidly changing climate. Research areas with specific activities that have been identified here will need particular attention from the international community of scientists, operational prediction centers, and stakeholders to ensure timely progress.
While the focus of the discussion in this paper has been primarily on environmental prediction on daily to seasonal time scales, it is important to point out that by moving polar prediction into the focus of the international community, much-needed progress in many areas of climate research and prediction can also be anticipated. In fact, we would argue that the polar regions are ideally suited to a seamless prediction approach (Palmer et al. 2008; Brunet et al. 2010). First, there is no clear distinction between the weather and climate research community in polar regions, with the latter, for example, providing substantial contributions to developing and running the observing system. Second, coupled models and coupled data assimilation systems will need to be used, even for short-term predictions traditionally addressed by atmosphere-only systems. While clearly challenging, eventually using coupled models in short-term predictions will provide a unique opportunity for diagnosing the origins of model error and hence improving climate models and climate projections. Furthermore, the high resolution needed for short-term predictions will allow new insights into the climate relevance of small-scale features such as leads in sea ice or orographic jets.
Coupled data assimilation systems will also be important for optimizing the observing system in polar regions. In the past, much emphasis has been put on climate monitoring. With the increasing demand for predictive information, more is asked of the polar observing system, and well-tested coupled data assimilation systems provide a good opportunity to redesign the polar observing system to meet the different competing demands in a cost-effective manner. The work will also pave the way for improved reanalysis of the polar regions.
In summary, the growing demand for polar predictive capacity, along with a community ready to take on the challenge through international collaboration, means that significant future advances can be expected that go well beyond the polar regions and time scales considered in this paper.
ACKNOWLEDGMENTS
Mohamed Dahoui from ECMWF is acknowledged for providing Fig. 3. Soumia Serrar contributed to producing Figs. 4 and 5. Gunnar Spreen and Dimitris Menemenlis provided the original configuration for the 4-km simulations with the MITgcm that formed the basis for Fig. 6. Contributions to the PPP Trust Fund are acknowledged as well as AWI’s financial contribution to hosting the ICO.
REFERENCES
ACIA, 2004: Impacts of a warming Arctic—Arctic climate impact assessment. ACIA Overview Rep., 140 pp. [Available online at www.amap.no/documents/doc/impacts-of-a-warming-arctic-2004/786.]
Aspelien, T., T. Iversen, J. B. Bremnes, and I.-L. Frogner, 2011: Short-range probabilistic forecasts from the Norwegian limited-area EPS: Long-term validation and a polar low study. Tellus, 63A, 564–584, doi:10.1111/j.1600-0870.2010.00502.x.
Barnes, E. A., and J. A. Screen, 2015: The impact of Arctic warming on the midlatitude jet-stream: Can it? Has it? Will it? WIREs Climatic Change, 6, 277–286, doi:10.1002/wcc.337.
Barnston, A. G., M. K. Tippett, M. L. L’Heureux, S. Li, and D. G. DeWitt, 2012: Skill of real-time seasonal ENSO model predictions during 2002–11: Is our capability increasing? Bull. Amer. Meteor. Soc., 93, 631–651, doi:10.1175/BAMS-D-11-00111.1.
Bauer, P., L. Magnusson, J.-N. Thépaut, and T. M. Hamill, 2016: Aspects of ECMWF model performance in polar areas. Quart. J. Roy. Meteor. Soc., 142, 583–596, doi:10.1002/qj.2449.
Bhatt, U. S., M. A. Alexander, C. Deser, J. E. Walsh, J. S. Miller, M. S. Timlin, J. Scott, and R. A. Tomas, 2008: The atmospheric response to realistic reduced summer arctic sea ice anomalies. Arctic Sea Ice Decline: Observations, Projections, Mechanisms, and Implications, E. T. DeWeaver, C. M. Bitz, and L.-B. Tremblay, Eds., Amer. Geophys. Union, 91–110.
Bougeault, P., and Coauthors, 2010: The THORPEX Interactive Grand Global Ensemble. Bull. Amer. Meteor. Soc., 91, 1059–1072, doi:10.1175/2010BAMS2853.1.
Boullot, N., F. Rabier, R. Langland, R. Gelaro, C. Cardinali, V. Guidard, P. Bauer, and A. Doerenbecher, 2016: Observation impact over the southern polar area during the Concordiasi field campaign. Quart. J. Roy. Meteor. Soc., 142, 597–610, doi:10.1002/qj.2470.
Bourassa, M. A., and Coauthors, 2013: High-latitude ocean and sea ice surface fluxes: Challenges for climate research. Bull. Amer. Meteor. Soc., 94, 403–423, doi:10.1175/BAMS-D-11-00244.1.
Bromwich, D. H., A. J. Monaghan, K. W. Manning, and J. G. Powers, 2005: Real-time forecasting for the Antarctic: An evaluation of the Antarctic Mesoscale Prediction System (AMPS). Mon. Wea. Rev., 133, 579–603, doi:10.1175/MWR-2881.1.
Bromwich, D. H., F. O. Otieno, K. M. Hines, K. W. Manning, and E. Shilo, 2013: Comprehensive evaluation of polar weather research and forecasting model performance in the Antarctic. J. Geophys. Res. Atmos., 118, 274–292, doi:10.1029/2012JD018139.
Brunet, G., and Coauthors, 2010: Collaboration of the weather and climate communities to advance subseasonal-to-seasonal prediction. Bull. Amer. Meteor. Soc., 91, 1397–1406, doi:10.1175/2010BAMS3013.1.
Buizza, R., and T. Palmer, 1995: The singular-vector structure of the atmospheric global circulation. J. Atmos. Sci., 52, 1434–1456, doi:10.1175/1520-0469(1995)052<1434:TSVSOT>2.0.CO;2.
Buizza, R., M. Miller, and T. N. Palmer, 1999: Stochastic representation of model uncertainities in the ECWMF Ensemble Prediction System. Quart. J. Roy. Meteor. Soc., 125, 2887–2908, doi:10.1002/qj.49712556006.
Chevallier, M., and D. Salas-Mélia, 2012: The role of sea ice thickness distribution in the Arctic sea ice potential predictability: A diagnostic approach with a coupled GCM. J. Climate, 25, 3025–3038, doi:10.1175/JCLI-D-11-00209.1.
Cohen, J., and Coauthors, 2014: Recent Arctic amplification and extreme mid-latitude weather. Nat. Geosci., 7, 627–637, doi:10.1038/ngeo2234.
Dawson, J., M. Johnston, and E. Stewart, 2014: Governance of Arctic expedition cruise ships in a time of rapid environmental and economic change. Ocean Coastal Manage., 89, 88–99, doi:10.1016/j.ocecoaman.2013.12.005.
Day, J., E. Hawkins, and S. Tietsche, 2014: Will Arctic sea ice thickness initialization improve seasonal forecast skill? Geophys. Res. Lett., 41, 7566–7575, doi:10.1002/2014GL061694.
Eicken, H., 2013: Ocean science: Arctic sea ice needs better forecasts. Nature, 497, 431–433, doi:10.1038/497431a.
Elvidge, A. D., I. A. Renfrew, J. C. King, A. Orr, T. A. Lachlan-Cope, M. Weeks, and S. L. Gray, 2015: Foehn jets over the Larsen C Ice Shelf, Antarctica. Quart. J. Roy. Meteor. Soc., 141, 698–713, doi:10.1002/qj.2382.
Emmerson, C., and G. Lahn, 2012: Arctic opening: Opportunity and risk in the high north. Lloyds Rep., 59 pp. [Available online at www.chathamhouse.org/sites/files/chathamhouse/public/Research/Energy,%20Environment%20and%20Development/0412arctic.pdf.]
Goessling, H. F., and Coauthors, 2016: Paving the way for the Year of Polar Prediction. Bull. Amer. Meteor. Soc., 97, ES85–ES88, doi:10.1175/BAMS-D-15-00270.1.
Guemas, V., and Coauthors, 2016: A review on Arctic sea-ice predictability and prediction on seasonal to decadal time-scales. Quart. J. Roy. Meteor. Soc., 142, 546–561, doi:10.1002/qj.2401.
Hall, C. M., and J. Saarinen, 2010: Polar tourism: Definitions and dimensions. Scand. J. Hospitality Tourism, 10, 448–467, doi:10.1080/15022250.2010.521686.
Hansen, J., R. Ruedy, M. Sato, and K. Lo, 2010: Global surface temperature change. Rev. Geophys., 48, RG4004, doi:10.1029/2010RG000345.
Hines, K., D. Bromwich, L. Bai, C. Bitz, J. Powers, and K. Manning, 2015: Sea ice enhancements to Polar WRF. Mon. Wea. Rev., 143, 2363–2385, doi:10.1175/MWR-D-14-00344.1.
Holland, M., and C. Bitz, 2003: Polar amplification of climate change in coupled models. Climate Dyn., 21, 221–232, doi:10.1007/s00382-003-0332-6.
Holland, M., and J. Stroeve, 2011: Changing seasonal sea ice predictor relationships in a changing Arctic climate. Geophys. Res. Lett., 38, L18501, doi:10.1029/2011GL049303.
Holtslag, A., and Coauthors, 2013: Stable atmospheric boundary layers and diurnal cycles: Challenges for weather and climate models. Bull. Amer. Meteor. Soc., 94, 1691–1706, doi:10.1175/BAMS-D-11-00187.1.
Houtekamer, P. L., and H. L. Mitchell, 1998: Data assimilation using an ensemble Kalman filter technique. Mon. Wea. Rev., 126, 796–811, doi:10.1175/1520-0493(1998)126<0796:DAUAEK>2.0.CO;2.
Inoue, J., T. Enomoto, T. Miyoshi, and S. Yamane, 2009: Impact of observations from Arctic drifting buoys on the reanalysis of surface fields. Geophys. Res. Lett., 36, L08501, doi:10.1029/2009GL037380.
Inoue, J., T. Enomoto, and M. E. Hori, 2013: The impact of radiosonde data over the ice-free Arctic Ocean on the atmospheric circulation in the Northern Hemisphere. Geophys. Res. Lett., 40, 864–869, doi:10.1002/grl.50207.
Inoue, J., A. Yamazaki, J. Ono, K. Dethloff, M. Maturilli, R. Neuber, P. Edwards, and H. Yamaguchi, 2015: Additional Arctic observations improve weather and sea-ice forecasts for the Northern Sea Route. Sci. Rep., 5, 16868; doi:10.1038/srep16868.
Jung, T., and M. Leutbecher, 2007: Performance of the ECMWF forecasting system in the Arctic during winter. Quart. J. Roy. Meteor. Soc., 133, 1327–1340, doi:10.1002/qj.99.
Jung, T., and P. B. Rhines, 2007: Greenland’s pressure drag and the Atlantic storm track. J. Atmos. Sci., 64, 4004–4030, doi:10.1175/2007JAS2216.1.
Jung, T., and M. Matsueda, 2016: Verification of global numerical weather forecasting systems in polar regions using TIGGE data. Quart. J. Roy. Meteor. Soc., 142, 574–582, doi:10.1002/qj.2437.
Jung, T., M. A. Kasper, T. Semmler, and S. Serrar, 2014: Arctic influence on subseasonal midlatitude prediction. Geophys. Res. Lett., 41, 3676–3680, doi:10.1002/2014GL059961.
Jung, T., and Coauthors, 2015: Polar lower-latitude linkages and their role in weather and climate prediction. Bull. Amer. Meteor. Soc., 96, ES197–ES200, doi:10.1175/BAMS-D-15-00121.1.
Juricke, S., H. F. Goessling, and T. Jung, 2014: Potential sea ice predictability and the role of stochastic sea ice strength perturbations. Geophys. Res. Lett., 41, 8396–8403, doi:10.1002/2014GL062081.
Kaleschke, L., X. Tian-Kunze, N. Maaß, M. Mäkynen, and M. Drusch, 2012: Sea ice thickness retrieval from SMOS brightness temperatures during the Arctic freeze-up period. Geophys. Res. Lett., 39, L05501, doi:10.1029/2012GL050916.
Kalnay, E., 2003: Atmospheric Modelling, Data Assimilation and Predictability. Cambridge University Press, 368 pp.
Kern, S., and G. Spreen, 2015: Uncertainties in Antarctic sea-ice thickness retrieval from ICESat. Ann. Glaciol., 56, 107–119, doi:10.3189/2015AoG69A736.
Klinker, E., and P. D. Sardeshmukh, 1992: The diagnosis of mechanical dissipation in the atmosphere from large-scale balance requirements. J. Atmos. Sci., 49, 608–627, doi:10.1175/1520-0469(1992)049<0608:TDOMDI>2.0.CO;2.
Kristiansen, J., S. L. Sørland, T. Iversen, D. Bjørge, and M. Ø. Køltzow, 2011: High-resolution ensemble prediction of a polar low development. Tellus, 63, 585–604, doi:10.1111/j.1600-0870.2010.00498.x.
Kristjánsson, J., and Coauthors, 2011: The Norwegian IPY–THORPEX: Polar lows and Arctic fronts during the 2008 Andøya campaign. Bull. Amer. Meteor. Soc., 92, 1443–1466, doi:10.1175/2011BAMS2901.1.
Kwok, R., 2010: Satellite remote sensing of sea-ice thickness and kinematics: A review. J. Glaciol., 56, 1129–1140, doi:10.3189/002214311796406167.
Lamers, M., E. Eijgelaar, and B. Amelung, 2011: Last chance tourism in Antarctica—Cruising for change? Last-Chance Tourism: Adapting Tourism Opportunities in a Changing World, H. Lemelin, J. Dawson, and E. Stewart, Eds., Routledge, 25–41.
Laxon, S., and Coauthors, 2013: CryoSat-2 estimates of Arctic sea ice thickness and volume. Geophys. Res. Lett., 40, 732–737, doi:10.1002/grl.50193.
Leutbecher, M., and T. Palmer, 2008: Ensemble forecasting. J. Comput. Phys., 227, 3515–3539, doi:10.1016/j.jcp.2007.02.014.
Linders, T., and Saetra. Ø, 2010: Can CAPE maintain polar lows? J. Atmos. Sci., 67, 2559–2571, doi:10.1175/2010JAS3131.1.
Masutani, M., and Coauthors, 2010: Observing system simulation experiments at the National Centers for Environmental Prediction. J. Geophys. Res., 115, D07101, doi:10.1029/2009JD012528.
Meredith, M., O. Schofield, L. Newman, E. Urban, and M. Sparrow, 2013: The vision for a Southern Ocean observing system. Curr. Opin. Environ. Sustainability, 5, 306–313, doi:10.1016/j.cosust.2013.03.002.
Moncrieff, M., D. Waliser, M. Miller, M. Shapiro, G. Asrar, and J. Caughey, 2012: Multiscale convective organization and the YOTC virtual global field campaign. Bull. Amer. Meteor. Soc., 93, 1171–1187, doi:10.1175/BAMS-D-11-00233.1.
Nguyen, A., R. Kwok, and D. Menemenlis, 2012: Source and pathway of the western Arctic upper halocline in a data-constrained coupled ocean and sea ice model. J. Phys. Oceanogr., 42, 802–823, doi:10.1175/JPO-D-11-040.1.
Overland, J., J. A. Francis, R. Hall, E. Hanna, S.-J. Kim, and T. Vihma, 2015: The melting Arctic and midlatitude weather patterns: Are they connected? J. Climate, 28, 7917–7932, doi:10.1175/JCLI-D-14-00822.1.
Palmer, T. N., G. J. Shutts, R. Hagedorn, F. Doblas-Reyes, T. Jung, and M. Leutbecher, 2005: Representing model uncertainty in weather and climate prediction. Annu. Rev. Earth Planet. Sci., 33, 163–193, doi:10.1146/annurev.earth.33.092203.122552.
Palmer, T. N., F. J. Doblas-Reyes, A. Weisheimer, and M. J. Rodwell, 2008: Toward seamless prediction: Calibration of climate change projections using seasonal forecasts. Bull. Amer. Meteor. Soc., 89, 459–470, doi:10.1175/BAMS-89-4-459.
Papritz, L., S. Pfahl, H. Sodemann, and H. Wernli, 2015: A climatology of cold air outbreaks and their impact on air–sea heat fluxes in the high-latitude South Pacific. J. Climate, 28, 342–364, doi:10.1175/JCLI-D-14-00482.1.
Pithan, F., B. Medeiros, and T. Mauritsen, 2014: Mixed-phase clouds cause climate model biases in Arctic wintertime temperature inversions. Climate Dyn., 43, 289–303, doi:10.1007/s00382-013-1964-9.
Powers, J., K. Manning, D. Bromwich, J. Cassano, and A. Cayette, 2012: A decade of Antarctic science support through AMPS. Bull. Amer. Meteor. Soc., 93, 1699–1712, doi:10.1175/BAMS-D-11-00186.1.
PPP Steering Group, 2013: WWRP Polar Prediction Project Implementation Plan. WMO Rep. WWRP/PPP 2-2013, 72 pp. [Available online at www.polarprediction.net/fileadmin/user_upload/redakteur/Home/Documents/WWRP-PPP_IP_Final_12Jan2013_v1_2.pdf.]
PPP Steering Group, 2014: WWRP Polar Prediction Project Implementation Plan for the Year of Polar Prediction (YOPP). WMO Rep. WWRP/PPP 3-2014, 53 pp. [Available online at www.polarprediction.net/fileadmin/user_upload/redakteur/Home/Documents/WWRP-PPP_YOPP_Plan_2014_v1_1.pdf.]
Rabier, F., H. Järvinen, E. Klinker, J.-F. Mahfouf, and A. Simmons, 2000: The ECMWF operational implementation of four-dimensional variational assimilation. I: Experimental results with simplified physics. Quart. J. Roy. Meteor. Soc., 126, 1143–1170, doi:10.1002/qj.49712656415.
Rasmussen, E. A., and J. Turner, Eds., 2003: Polar Lows: Mesoscale Weather Systems in the Polar Regions. Cambridge University Press, 628 pp.
Renfrew, I. A., G. N. Petersen, D. A. J. Sproson, G. W. K. Moore, H. Adiwidjaja, S. Zhang, and R. North, 2009: A comparison of aircraft-based surface-layer observations over Denmark Strait and the Irminger Sea with meteorological analyses and QuikSCAT winds. Quart. J. Roy. Meteor. Soc., 135, 2046–2066, doi:10.1002/qj.444.
Rodwell, M. J., and T. N. Palmer, 2007: Using numerical weather prediction to assess climate models. Quart. J. Roy. Meteor. Soc., 133, 129–146, doi:10.1002/qj.23.
Rodwell, M. J., and T. Jung, 2008: Understanding the local and global impacts of model physics changes: An aerosol example. Quart. J. Roy. Meteor. Soc., 134, 1479–1497, doi:10.1002/qj.298.
Rodwell, M. J., and T. Jung, 2010: Diagnostics at ECMWF. Proc. ECMWF Seminar on Diagnosis of Forecasting and Data Assimilation Systems, Shinfield Park, Reading, ECMWF, 77–94. [Available online at www.ecmwf.int/sites/default/files/elibrary/2010/11981-diagnostics-ecmwf.pdf.]
Rodwell, M. J., and Coauthors, 2013: Characteristics of occasional poor medium-range weather forecasts for Europe. Bull. Amer. Meteor. Soc., 94, 1393–1405, doi:10.1175/BAMS-D-12-00099.1.
Roemmich, D., and J. Gilson, 2009: The 2004–2008 mean and annual cycle of temperature, salinity, and steric height in the global ocean from the Argo Program. Prog. Oceanogr., 82, 81–100, doi:10.1016/j.pocean.2009.03.004.
Sandu, I., A. Beljaars, P. Bechtold, T. Mauritsen, and G. Balsamo, 2013: Why is it so difficult to represent stably stratified conditions in numerical weather prediction (NWP) models? J. Adv. Model. Earth Syst., 5, 117–133, doi:10.1002/jame.20013.
Smith, G. C., F. Roy, and B. Brasnett, 2013: Evaluation of an operational ice-ocean analysis and forecasting system for the Gulf of St Lawrence. Quart. J. Roy. Meteor. Soc., 139, 419–433, doi:10.1002/qj.1982.
Smith, L. C., and S. R. Stephenson, 2013: New trans-Arctic shipping routes navigable by midcentury. Proc. Natl. Acad. Sci. USA, 110, E1191–E1195, doi:10.1073/pnas.1214212110.
Steinhoff, D. F., D. H. Bromwich, and A. Monahan, 2013: Dynamics of the Foehn dynamics of the Foehn mechanism in the McMurdo Dry Valleys of Antarctica from Polar WRF. Quart. J. Roy. Meteor. Soc., 139, 1615–1631, doi:10.1002/qj.2038.
Stroeve, J., L. C. Hamilton, C. M. Bitz, and E. Blanchard-Wrigglesworth, 2014: Predicting September sea ice: Ensemble skill of the SEARCH Sea Ice Outlook 2008–2013. Geophys. Res. Lett., 41, 2411–2418, doi:10.1002/2014GL059388.
Tian-Kunze, X., L. Kaleschke, N. Maaß, M. Mäkynen, N. Serra, M. Drusch, and T. Krumpen, 2014: SMOS derived sea ice thickness: Algorithm baseline, product specifications and initial verification. Cryosphere, 8, 997–1018, doi:10.5194/tc-8-997-2014.
Tietsche, S., and Coauthors, 2014: Seasonal to interannual Arctic sea ice predictability in current global climate models. Geophys. Res. Lett., 41, 1035–1043, doi:10.1002/2013GL058755.
Tilinina, N., S. Gulev, and D. Bromwich, 2014: New view of Arctic cyclone activity from the Arctic system reanalysis. Geophys. Res. Lett., 41, 1766–1772, doi:10.1002/2013GL058924.
Toth, Z., and E. Kalnay, 1993: Ensemble forecasting at NMC: The generation of perturbations. Bull. Amer. Meteor. Soc., 74, 2317–2330, doi:10.1175/1520-0477(1993)074<2317:EFANTG>2.0.CO;2.
Victoria Team, and L. Manderson, 2011: Social and public health effects of climate change in the ‘40 South.’ WIREs Climate Change, 2, 902–918, doi:10.1002/wcc.138.
Vihma, T., and Coauthors, 2014: Advances in understanding and parameterization of small-scale physical processes in the marine Arctic climate system: A review. Atmos. Chem. Phys., 14, 9403–9450, doi:10.5194/acp-14-9403-2014.
Wang, W., M. Chen, and A. Kumar, 2013: Seasonal prediction of Arctic Sea ice extent from a coupled dynamical forecast system. Mon. Wea. Rev., 141, 1375–1394, doi:10.1175/MWR-D-12-00057.1.
Yamazaki, A., J. Inoue, K. Dethloff, M. Maturilli, and G. König-Langlo, 2015: Impact of radiosonde observations on forecasting summertime arctic cyclone formation. J. Geophys. Res. Atmos., 120, 3249–3273, doi:10.1002/2014JD022925.
