Alicia Karspeck

Decadal climate prediction: An update from the trenches

This paper provides an update on research in the relatively new and fast-moving field of decadal climate prediction, and addresses the use of decadal climate predictions not only for potential users of such information but also for improving our understanding of processes in the climate system. External forcing influences the predictions throughout, but their contributions to predictive skill become dominant after most of the improved skill from initialization with observations vanishes after about 6–9 years. Recent multimodel results suggest that there is relatively more decadal predictive skill in the North Atlantic, western Pacific, and Indian Oceans than in other regions of the world oceans. Aspects of decadal variability of SSTs, like the mid-1970s shift in the Pacific, the mid-1990s shift in the northern North Atlantic and western Pacific, and the early-2000s hiatus, are better represented in initialized hindcasts compared to uninitialized simulations. There is evidence of higher skill in initialize...

Forecasting seasonal outbreaks of influenza

Influenza recurs seasonally in temperate regions of the world; however, our ability to predict the timing, duration, and magnitude of local seasonal outbreaks of influenza remains limited. Here we develop a framework for initializing real-time forecasts of seasonal influenza outbreaks, using a data assimilation technique commonly applied in numerical weather prediction. The availability of real-time, web-based estimates of local influenza infection rates makes this type of quantitative forecasting possible. Retrospective ensemble forecasts are generated on a weekly basis following assimilation of these web-based estimates for the 2003–2008 influenza seasons in New York City. The findings indicate that real-time skillful predictions of peak timing can be made more than 7 wk in advance of the actual peak. In addition, confidence in those predictions can be inferred from the spread of the forecast ensemble. This work represents an initial step in the development of a statistically rigorous system for real-time forecast of seasonal influenza.

A Decadal Prediction Case Study: Late Twentieth-Century North Atlantic Ocean Heat Content

An ensemble of initialized decadal prediction (DP) experiments using the Community Climate System Model, version 4 (CCSM4) shows considerable skill at forecasting changes in North Atlantic upper-ocean heat content and surface temperature up to a decade in advance. Coupled model ensembles were integrated forward from each of 10 different start dates spanning from 1961 to 2006 with ocean and sea ice initial conditions obtained from a forced historical experiment, a Coordinated Ocean-Ice Reference Experiment with Interannual forcing (CORE-IA), which exhibits good correspondence with late twentieth-century ocean observations from the North Atlantic subpolar gyre (SPG) region. North Atlantic heat content anomalies from the DP ensemble correlate highly with those from the CORE-IA simulation after correcting for a drift bias.In particular, theobservedlarge,rapid rise in SPGheatcontentin themid-1990sis successfullypredicted in the ensemble initialized in January of 1991. A budget of SPG heat content from the CORE-IA experiment sheds light on the origins of the 1990s regime shift, and it demonstrates the extent to which low-frequency changes in ocean heat advection related to the Atlantic meridional overturning circulation dominate temperature tendencies in this region. Similar budgets from the DP ensembles reveal varying degrees of predictive skill in the individual heat budget terms, with large advective heat flux anomalies from the south exhibiting the highest correlation with CORE-IA. The skill of the DP in this region is thus tied to correct initialization of ocean circulation anomalies, while external forcing is found to contribute negligibly (and for incorrect reasons) to predictive skill in this region over this time period.

Real-time influenza forecasts during the 2012–2013 season

Recently, we developed a seasonal influenza prediction system that uses an advanced data assimilation technique and real-time estimates of influenza incidence to optimize and initialize a population-based mathematical model of influenza transmission dynamics. This system was used to generate and evaluate retrospective forecasts of influenza peak timing in New York City. Here we present weekly forecasts of seasonal influenza developed and run in real time for 108 cites in the United States during the recent 2012–2013 season. Reliable ensemble forecasts of influenza outbreak peak timing with leads of up to 9 weeks were produced. Forecast accuracy increased as the season progressed, and the forecasts significantly outperformed alternate, analog prediction methods. By Week 52, prior to peak for the majority of cities, 63% of all ensemble forecasts were accurate. To our knowledge, this is the first time predictions of seasonal influenza have been made in real time and with demonstrated accuracy.

Comparison of filtering methods for the modeling and retrospective forecasting of influenza epidemics.

A variety of filtering methods enable the recursive estimation of system state variables and inference of model parameters. These methods have found application in a range of disciplines and settings, including engineering design and forecasting, and, over the last two decades, have been applied to infectious disease epidemiology. For any system of interest, the ideal filter depends on the nonlinearity and complexity of the model to which it is applied, the quality and abundance of observations being entrained, and the ultimate application (e.g. forecast, parameter estimation, etc.). Here, we compare the performance of six state-of-the-art filter methods when used to model and forecast influenza activity. Three particle filters—a basic particle filter (PF) with resampling and regularization, maximum likelihood estimation via iterated filtering (MIF), and particle Markov chain Monte Carlo (pMCMC)—and three ensemble filters—the ensemble Kalman filter (EnKF), the ensemble adjustment Kalman filter (EAKF), and the rank histogram filter (RHF)—were used in conjunction with a humidity-forced susceptible-infectious-recovered-susceptible (SIRS) model and weekly estimates of influenza incidence. The modeling frameworks, first validated with synthetic influenza epidemic data, were then applied to fit and retrospectively forecast the historical incidence time series of seven influenza epidemics during 2003–2012, for 115 cities in the United States. Results suggest that when using the SIRS model the ensemble filters and the basic PF are more capable of faithfully recreating historical influenza incidence time series, while the MIF and pMCMC do not perform as well for multimodal outbreaks. For forecast of the week with the highest influenza activity, the accuracies of the six model-filter frameworks are comparable; the three particle filters perform slightly better predicting peaks 1–5 weeks in the future; the ensemble filters are more accurate predicting peaks in the past.

Decadal upper ocean temperature variability in the tropical Pacific

Decadal variability in upper ocean temperature in the Pacific is studied by using observations and results from model experiments. Especially propagation of upper ocean thermal anomalies from the midlatitudes to the tropics is studied as a possible source for decadal equatorial thermocline variability. In the observations, propagation along the subtropical gyre of the North Pacific is clear. However, no propagation into the equatorial region is found. Model experiments with an ocean model forced with observed monthly wind and wind stress anomalies are performed to study the apparent propagation. Distinct propagation of thermal anomalies in the subtropics is found in the model, although the amplitude of the anomalies is small. The anomalies clearly propagate into the tropics, but they do not reach the equatorial region. The small response at the equator to extratropical variability consists of a change in the mean depth of the thermocline. It appears that most variability in the subtropics and tropics is generated by local wind stress anomalies. The results are discussed by using results from a linear shallow water model in which similar features are found.

Predicted slowdown in the rate of Atlantic sea ice loss

Coupled climate models initialized from historical climate states and subject to anthropogenic forcings can produce skillful decadal predictions of sea surface temperature change in the subpolar North Atlantic. The skill derives largely from initialization, which improves the representation of slow changes in ocean circulation and associated poleward heat transport. We show that skillful predictions of decadal trends in Arctic winter sea ice extent are also possible, particularly in the Atlantic sector. External radiative forcing contributes to the skill of retrospective decadal sea ice predictions, but the spatial and temporal accuracy is greatly enhanced by the more realistic representation of ocean heat transport anomalies afforded by initialization. Recent forecasts indicate that a spin-down of the thermohaline circulation that began near the turn of the century will continue, and this will result in near-neutral decadal trends in Atlantic winter sea ice extent in the coming years, with decadal growth in select regions.

Tropical Pacific 1976-77 Climate Shift in a Linear, Wind-Driven Model*

A number of studies have attempted to explain the cause of decadal variability in the tropical Pacific and explore its possible link to decadal variability in the midlatitude Pacific. To investigate some of the current theories of Pacific decadal variability, a linear, wind-driven model, designed to simulate only baroclinic wave dynamics, was forced with wind stress anomalies in the Pacific Ocean basin from 1945 through 1992. An analysis technique designed to isolate the decadal/interdecadal scale variability from interannual ENSO variability was performed on the model’s thermocline depth anomaly (TDA). It was found that the temporal and spatial patterns of the observed tropical decadal sea surface temperatures are consistent with our modeled TDA. Furthermore, restricting the wind forcing to within 58 of the equator does not substantially alter the decadal/interdecadal variability of the equatorial region. The authors conclude that the observed decadal variability in the low-latitude Pacific is primarily a linear dynamical response to tropical wind forcing and does not directly require an oceanic link to the midlatitudes. The question of how tropical wind anomalies are generated is not addressed. In addition, it is shown that in model scenarios where the wind forcing is restricted to the western equatorial Pacific, the 1976‐77 climate shift is still clearly visible as a dominant feature of tropical decadal variability. The temporal decadal signal of the model-generated TDA is more pronounced during the eastern equatorial upwelling season (July‐September) than in the boreal winter. This is consistent with the observed seasonal bias in tracer and SST data from the eastern equatorial Pacific.

Predicting Pacific Decadal Variability

The case is advanced that decadal variability of climate in the Pacific sector is driven by tropical atmosphere-ocean interactions and communicated to the extratropics. It is shown that tropical decadal variations in the last century could arise as a consequence of the regional subset of physics contained within an intermediate model of the El Nino-Southern Oscillation. These decadal changes in ENSO and tropical mean climate are more predictable than chance years in advance but even in these idealized experiments forecast skill is probably too small to be useful. Nonetheless, forecasts of the next two decades indicate that, according to this model, the 1998 El Nino marked the end of the post 1976 tropical Pacific warm period. Observations and atmosphere general circulation models are interpreted to suggest that decadal variations of the atmosphere circulation over the North Pacific between the 1960s and the 1980s are explained by a mix of tropical forcing and internal atmospheric variability. This places a limit on their predictability. The ocean response to extratropical atmosphere variability consists of a local response that is instantaneous and a delayed response of the subtropical and subpolar gyres that is predictable a few years in advance. It is shown that the wintertime internal variability of the Aleutian Low can weakly impact the ENSO system but its impact on decadal predictability is barely discernible.

Predictability of Tropical Pacific Decadal Variability in an Intermediate Model

Abstract The Zebiak–Cane (ZC) model for simulation of the El Nino–Southern Oscillation is shown to be capable of producing sequences of variability that exhibit shifts in the time-mean state of the eastern equatorial Pacific that resemble observations of tropical Pacific decadal variability. The models performance in predicting these shifts is compared to two naive forecasting strategies. It is found that the ZC model consistently outperforms the two naive forecasts that serve as a null hypothesis in assessing the significance of results. Forecasts initialized during anomalously warm and anomalously cold decades are shown to have the highest predictability. These modeling results suggest that, to a moderate extent, the state of the tropical Pacific in one decade can predetermine its time-mean state in the following decade. However, even in this idealized context decadal forecasting skill is modest. Results are discussed in the context of their implications for the ongoing debate over the origin of decada...